Washington University in St. Louis
Washington University Open Scholarship
All Theses and Dissertations (ETDs)
12-2-2013
The Relationship between Household Economic
Resources and Youth Academic Performance in
Ghana: A Multilevel Structural Equation Modeling
David Ansong
Washington University in St. Louis
Follow this and additional works at: http://openscholarship.wustl.edu/etd
Part of the Social Work Commons
Recommended Citation
Ansong, David, "The Relationship between Household Economic Resources and Youth Academic Performance in Ghana: A
Multilevel Structural Equation Modeling" (2013). All Theses and Dissertations (ETDs). 1194.
http://openscholarship.wustl.edu/etd/1194
This Dissertation is brought to you for free and open access by Washington University Open Scholarship. It has been accepted for inclusion in All
Theses and Dissertations (ETDs) by an authorized administrator of Washington University Open Scholarship. For more information, please contact
[email protected].
WASHINGTON UNIVERSITY IN ST. LOUIS
George Warren Brown School of Social Work
Dissertation Examination Committee:
Michael Sherraden, Chair
Gina Chowa
Carolyn Lesorogol
Amanda Moore McBride
Mungai Mutonya
Ramesh Raghavan
James Wertsch
The Relationship between Household Economic Resources and Youth Academic Performance in
Ghana: A Multilevel Structural Equation Modeling
by
David Ansong
A dissertation presented to the
Graduate School of Arts and Sciences
of Washington University in
partial fulfillment of the
requirements for the degree
of Doctor of Philosophy
December 2013
St. Louis, Missouri
Copyright by
David Ansong
2013
Table of Contents
List of Figures………………………………………………………….……………………...iv
List of Tables………………………………………………………….……………………….iv
List of Abbreviations………………………………………………………………………….v
Acknowledgements………………………………………………………….……………......vi
Abstract………………………………………………………………………………………vii
1. Chapter 1: Introduction…………………………………………………………………..1
1.1. Background of the Study……………………………………………………………....1
1.2. Statement of Problem………………………………………………………………….2
1.2.1. Low Academic Performance…………………………………………………...3
1.2.2. High Cost of Transition………………………………………………………...5
1.3. Significance of the Study……………………………………………………………...7
1.4. Purpose of the Study…………………………………………………………………..9
1.5. History and Structure of the Ghanaian Educational System…………………………..9
1.6. Definition of Concepts………………………………………………………………..12
1.6.1. Youth…………………………………………………………………………..12
1.6.2. Formal Education and Educational Outcomes………………………………...12
1.6.3. Economic Resources…………………………………………………………..13
1.7. Conclusion …………………………………………………………………………...15
2. Chapter 2: Theoretical Framework …………………………........................................17
2.1. Overview…………………………..............................................................................17
2.2. Human Capital Development …………………………..............................................17
2.3. Effects of Income on Educational Outcomes………………………….......................21
2.4. Effects of Assets on Educational Outcomes………………………….........................23
2.4.1. Direct Effects of Economic Resources on Educational Outcomes……………24
2.4.2. Indirect Effects of Economic Resources on Educational Outcomes………….25
2.4.2.1.
Academic self-efficacy pathway…………………………...................24
2.4.2.2.
Academic expectations pathway…………………………...................26
2.4.3. The Asset Experience …………..………………………….............................30
2.5. Summary ………………………….............................................................................32
3. Chapter 3: Empirical Evidence…………………………...............................................34
3.1. Overview…………………………..............................................................................34
3.2. Evidence from Developed Countries…………………………...................................34
3.2.1. Effects of Income on Educational Outcomes…………………………............34
3.2.2. Effects of Assets on Educational Outcomes………………………….............36
3.3. Evidence from Sub-Saharan Africa………………………….....................................39
3.3.1. Effects of Income on Educational Outcomes…………………………...........40
ii
3.3.2. Effects of Assets on Educational Outcomes…………………………............41
3.4. Additional Correlates and Predictors of Educational Outcomes……………………45
3.4.1. Individual-Level Correlates………………………….....................................46
3.4.2. Household- and Family-Level Correlates………………………....................47
3.4.3. School- and Community-Level Correlates……………………….................48
3.5. Conclusion ………………………….........................................................................50
4. Chapter 4: Methodology…………………………..........................................................51
4.1. Overview………………………….............................................................................51
4.2. Conceptual Framework………………………….......................................................51
4.2.1. Direct Effects of Household Economic Resources…………………………..51
4.2.2. Indirect Effects of Household Economic Resources through Self-Efficacy
and Expectations……………………………………………………………..52
4.3. Research Questions………………………….............................................................53
4.3.1. Direct Relationships………………………….................................................53
4.3.2. Indirect Relationships…………………………...............................................54
4.3.3. Moderated Relationships…………………………..........................................55
4.4. Research Design and Setting.…………….................................................................56
4.4.1. Sampling ………………………….................................................................57
4.4.2. Data Collection ………………………….......................................................58
4.5. Instruments…………………………..........................................................................59
4.6. Measures………………………….............................................................................60
4.6.1. Dependent Variables…………………………................................................60
4.6.2. Independent Variables…………………………..............................................61
4.6.2.1.
Parents’ income…………………………............................................61
4.6.2.2.
Household asset ownership………………..........................................61
4.6.3. Hypothesized Mediators…………………………...........................................62
4.6.3.1.
Academic self-efficacy………………………….................................62
4.6.3.2.
Academic expectation ………………………….................................62
4.6.4. Covariates………………………….................................................................63
4.6.4.1.
Individual-level characteristics……………….....................................63
4.6.4.2.
Household-level characteristics……………………............................64
4.6.4.3.
School-level characteristic……………………....................................64
4.7. Data Analysis Plan…………………………..............................................................65
4.7.1. Measurement Models………………….………..............................................67
4.7.2. Multilevel Structural Models…………….………….......................................68
4.7.2.1.
Multilevel Mediation Analyses….……………....................................68
4.7.2.2.
Moderation Test………………….………...........................................70
iii
5. Chapter 5: Results…………………………....................................................................71
5.1. Overview………………………….............................................................................71
5.2. Descriptive Results…………………………..............................................................71
5.2.1. Individual- and Household-Level Characteristics.…………….......................71
5.2.2. School-Level Characteristics…………………………………………………73
5.2.3. Dependent Variables …………………………...............................................74
5.2.4. Independent Variables…………………………..............................................74
5.2.5. Mediators (Academic Self-Efficacy and Expectations)...................................75
5.3. Results of Multilevel Measurement Models………………….………......................75
5.4. Results of Direct Relationships between Economic Resource and
Academic Performance …………………………………..........................................78
5.4.1. Direct Relationship between Economic Resources and English Scores.…….78
5.4.2. Direct Relationship between Economic Resources and Math Scores….…….82
5.5. Results of Indirect Relationships between Economic Resource and
Academic Performance ………………………………………...……………………90
5.5.1. Results of Self-Efficacy Pathway…………………………..............................90
5.5.1.1.
Mediating role of academic self-efficacy in English score outcomes...90
5.5.1.2.
Mediating role of academic self-efficacy in math score outcomes…...92
5.5.2. Results of Expectation Pathway…………………………................................94
5.5.2.1.
Mediating role of youths’ expectations in English score outcomes…..94
5.5.2.2.
Mediating role of youths’ expectations in math score outcomes …….95
5.6. Conclusion…………………………………………………………………………....96
6. Chapter 6: Discussion …………………………..............................................................97
6.1. Summary of Findings…………………………...........................................................97
6.2. Direct Relationship between Household Economic Resources and
Academic Performance………………………............................................................99
6.3. Mediated (Indirect) Relationships between Household Economic Resources and
Academic Performance…………………………………….......................................102
6.4. Gender as a Partial Moderator in the Relationship between Household Economic
Resources and Academic Performance…………………………...............................104
6.5. Relationship between School-Level Variables and Academic Performance………..106
6.6. Relationship between other Control Variables and Academic Performance..............107
6.7. Challenges and Limitations………………………….................................................109
7. Chapter 7: Implications and Conclusion…………………………................................112
7.1. Implications for Policy…………………………........................................................112
7.2. Implications for Future Research…………………………........................................115
7.3. Conclusion…………………………...........................................................................118
References……………………………………………………………………………………120
Appendices.……………………………………………………………………………..160-162
iv
List of Figures
Figure 1. Illustration of a multiple mediation design with two mediators…………………….52
Figure 2. Two-level path diagram for the direct and indirect relationships between
household economic resources and youth academic performance…………………69
Figure 3. Multilevel exploratory factor analysis for academic self-efficacy and
commitment to school scales at the within- and between-school levels..…………..77
Figure 4. Estimated unstandardized path coefficients for direct and indirect association
between parents’ income and English grade for the full sample at the
within- and between-school levels…………………….……………………………79
Figure 5. Estimated unstandardized path coefficients for direct and indirect association
between household assets and English grade for the full sample at the
within- and between-school levels…………………………………………………..80
Figure 6. Estimated unstandardized path coefficients for direct and indirect association
between parents’ income and math grade for the full sample at the
within- and between-school levels…………………………………………...….......83
Figure 7. Estimated unstandardized path coefficients for direct and indirect association
between household assets and math grade for the full sample at the
within- and between-school levels ………………….................................................84
List of Tables
Table 1. Descriptive characteristics of the full, female-only, and male-only samples………..72
Table 2. Summary of model fit statistics for measurement and structural models……………76
Table 3. Coefficient estimates of independent variables and mediators
in the English models………………………………………………………………..86
Table 4. Coefficient estimates of covariates in the English models…………………………...87
Table 5. Coefficient estimates of independent variables and mediators in the math model…..88
Table 6. Coefficient estimates of covariates in the English math model……………………...89
Table 7. Summary of hypotheses supported and not supported ………………………………98
v
List of Abbreviations
BCS
BECE
BHPS
CCT
CFA
CFI
CI
CGAP
CSD
FCUBE
GES
GHC
GNA
GoG
GPA
GSS
JHS
LEAP
MSEM
NAF
NLSY
NYPG
OECD
PIH
PSID
PTA
RMSEA
ROSCA
SACMEQ
SCT
SE
SEM
SHS
SSA
TRA
UNESCO
UNICEF
British Cohort Study
Basic Education Certificate Examination
British Household Panel Study
Conditional Cash Transfer
Confirmatory Factor Analysis
Comparative Fit Index
Confidence Interval
Consultative Group to Assist the Poor
Center for Social Development
Free Compulsory Universal Basic Education
Ghana Education Service
Ghana Cedi
Ghana News Agency
Government of Ghana
Grade Point Average
Ghana Statistical Service
Junior High School
Livelihood Empowerment against Poverty
Multi-group Structural Equation Modeling
New America Foundation
National Longitudinal Survey of Youth
National Youth Policy of Ghana
Organization for Economic Co-operation and Development
Permanent Income Hypothesis
Panel Study of Income Dynamics
Parent-Teacher Association
Root Mean Square Error of Approximation
Rotating Savings and Credit Association
Southern and East African Consortium for Monitoring Education Quality
Social Cognitive Theory
Standard Errors
Structural Equation Modeling
Senior High School
Sub-Saharan Africa
Theory of Reasoned Action
United Nations Educational, Scientific and Cultural Organization
United Nations Children's Fund
vi
Acknowledgements
This dissertation would not have been possible without the support of many people. I
wish to express my enormous gratitude for the advice, leadership, encouragement, and assistance
that my advisors Michael Sherraden PhD (dissertation chairperson) and Gina Chowa PhD
(dissertation committee member) so generously offered to me throughout my doctoral training
and the dissertation process. Their counsel and dedication to my research training inspired me to
pursue my research interest. I am thankful to have had the opportunity to work and study under
them as a Research Associate at the Center for Social Development.
To the other dissertation committee members, Carolyn Lesorogol PhD, Ramesh
Raghavan MD, PhD, Mungai Mutonya PhD, Amanda Moore McBride PhD, and James Wertsch
PhD, I am grateful for your support and steadfastness and for believing in me and in the
importance of this work. Your guidance and encouragement sustained me in this research.
Thank you to my family who stood by me from the beginning through to the end. My
wife Bernice, my son Kwesi, my brother Eric, and my friend Abena Ampomah have been my
strongest source of love, support, and strength throughout the dissertation process.
Many special thanks to my colleagues and friends at the Center for Social Development,
Li Zou, Lissa Johnson, Tiffany Trautwein, Tiffany Heineman, and Julia Steven for their research
and administrative assistance.
vii
ABSTRACT OF THE DISSERTATION
The Relationship between Household Economic Resources and Youth Academic Performance in
Ghana: A Multilevel Structural Equation Modeling
by
David Ansong
Doctor of Philosophy in Social Work
Washington University in St. Louis, 2013
Professor Michael Sherraden, Chair
The Government of Ghana recognizes the importance of education for improving the
socioeconomic well-being of young Ghanaians and the development of the nation. Education
currently accounts for the largest share (31%) of Ghana’s national budget (GNA, 2012).
Educational reforms and investments have had remarkable success in improving access to
education up to the Junior High School (JHS) level, but progression beyond JHS remains a
challenge. More than half of JHS graduates do not gain admission into Senior High School
(SHS) because of poor academic performance and the inability of those who qualify to afford the
drastic increase in educational costs.
Stakeholders have focused on different ways to address these educational challenges.
Researchers who study the developmental effects of owning assets (e.g., savings, home, land,
livestock, etc.) in particular have begun paying attention to ways in which personal (e.g.,
savings) and household (e.g., assets) economic resources can improve young people’s
educational outcomes. This emerging area of work is guided by the asset-effects framework,
which posits that economic resources may influence educational outcomes directly through being
able to pay for tuition and school supplies and indirectly through their influence on cognitive
viii
processes (e.g., academic self-efficacy and expectations). Empirical evidence from around the
world suggests potential connections, but there are significant research gaps on how specific
types of economic resources affect educational outcomes. For instance, while most studies that
link income and educational outcomes find strong relationships between income and children’s
academic performance, others find mixed or contradictory results. The mixed results suggest the
need for further conceptualization and empirical research to clarify the nature of the relationships
between different types of economic resources and academic performance.
To help address the research gaps, this study uses nested cross-sectional data from an
ongoing youth savings experiment to examine potential direct and indirect associations between
two types of economic resources—parents’ income and household assets—and math and English
scores of middle school-age youth in Ghana. In addition, the study assesses the possibility that
gender moderates the relationships between the aforementioned types of economic resources and
math and English scores.
Using multilevel structural equation modeling techniques, the study does not find strong
evidence to support the hypothesized direct relationship between economic resources and
educational outcomes. However, the study finds strong evidence that parents’ income and
household assets indirectly affect math and English scores through young people’s academic
self-efficacy and expectations. These findings suggest that psychological factors such as
academic self-efficacy and expectation are more predictive of youth academic performance than
economic resources. Subgroup analyses also show that gender moderates the indirect
relationships between economic resources and English scores but not math scores. This finding
concerning possible gender differences could inform policymakers about economic resources
that may promote parallel outcomes for boys and girls.
ix
Chapter 1: Introduction
Background of the Study
Formal education is the principal institutional route by which people are equipped with
the knowledge and skills they need to become economically productive and competitive
(Baffour-Awuah, 2011; Maliyamkono & Ogbu, 1999; Oduro, 2000). Like most young people
around the world, school-age youth in Ghana need formal education to acquire knowledge, skills,
and competencies to participate in an increasingly knowledge-driven society. Emerging
empirical evidence from sub-Saharan Africa (SSA) suggests that having economic resources is
associated with educational success (Chowa, Ansong, & Rainier, 2010). However, such research
does not explain how different types of economic resources—such as accumulated savings and
income—might have varying associations with specific types of educational outcomes (e.g.,
academic performance) or how these relationships may vary by gender. Thus, this dissertation
examines:
(1) how two types of economic resources—income and assets—are associated with
academic performance1 of Junior High School (JHS) students in Ghana,
(2) whether students’ psychological factors—academic self-efficacy and academic
expectations—mediate the relationships between economic resources and academic
performance, and
(3) whether the nature of the relationships between economic resources and academic
performance depends on the child’s gender.
Because formal education is critical to economic growth and human development
through human capital, especially in developing countries (Barro, 2006; Oduro, 2000), the
1
For the purposes of this study, academic performance refers to how well students are doing with respect to grades
and standardized tests.
1
Government of Ghana (GoG), public and private organizations, and other education stakeholders
have made and continue to make significant investments in the country’s educational system. For
the individual, formal education has both monetary and nonmonetary benefits, as it boosts:

earning potential, economic well-being, and social mobility (Baum & Payea, 2005;
Haveman & Wolfe, 1994; Keeley, 2007; Loke & Sacco, 2011);

social skills, including the ability to engage in peer relationships, self-control, compliance
with laws and social codes, and assertiveness (Gesthuizen, Van der Meer, & Scheepers,
2008); and

cultural capital through their acquisition and/or reinforcement of norms and values (e.g.,
tolerance, hard work, civility, and respect) (National Research Council, 2005).
Statement of Problem
Given that there are more returns on attaining higher levels of education, the youth in
Ghana must pursue educational advancement beyond the basic education level to benefit fully. In
Ghana and many developing countries, the youth with higher education have comparative
advantage over those who have only basic education. This is because those without higher
education often have to direct their job search toward the lower-tier and informal sector jobs that
are mostly casual and temporary, but the youth with higher education can be competitive in both
the formal and informal job market (El Badaoui & Rebiere, 2012). For instance, the 2007 job
tracking survey by the Ghana Statistical Service (GSS) revealed that more than 80% of job
openings in the formal employment sector were available only to individuals with no less than
Senior High School (SHS) education (Amankrah, 2007). Thus, there are better job prospects in
the formal sector for individuals who are able to enroll in formal education beyond the JHS level.
However, there are enormous challenges to the smooth transition from JHS to SHS in Ghana,
2
two of which are low academic performance and high cost of higher education (Ministry of
Education, Science, and Sports, 2007).
Low Academic Performance
Admission to SHS in Ghana is partly merit-based. That means successful transition from
JHS to SHS partially hinges on good grades. Yet, many JHS students have poor academic
performance and therefore they cannot gain admission to SHS (Ajayi, 2012). For example,
between 1992 and 1997, less than one fifth of JHS students reached mastery level2 in English
language and math subjects (Mereku, 2003). From 2006 to 2008, 38% of JHS students who sat
for the Basic Education Certificate Examination (BECE)3 did not qualify for SHS because of
poor performance (Daily Graphic, 2008). In 2008 and 2010, students from 34 schools had a 0%
pass rate on the BECE and therefore did not qualify for SHS (Okyerefo, Fiaveh, & Lamptey,
2011). Several inter-country studies also have corroborated the observation that academic
performance in Ghana is comparatively low. For instance, Ghana placed last among other
African countries, including South Africa, Morocco, Botswana, and Egypt in the Trends in
International Math and Science Study (TMISS) in 2003 (Abukari, 2010). A repetition of the
TMISS study in 2007 and 2011 revealed that academic performance in Ghana continues to trail
other countries with similar income levels (Etsey et. al., 2009; National Center for Education
Statistics, n.d.).
Poor academic performance in Ghana has been attributed to inadequate infrastructure,
lack of educational materials, and inadequate training and motivation of teachers (Ghana News
Agency, 2010; Norviewu-Mortty, 2012). Nevertheless, individual- and family-owned economic
2
As of 2003, the Ghana Education Service’s (GES) cutoff for mastery in a subject was 60% (i.e., Average) for
English language and 55% for math (Mereku, 2003). The GES uses the following 9-scale grading system: Excellent
(80%-100%), Very Good (75%-79%), Good (70%-74%), High Average (65%-69%), Average (60%-64%), Low
Average (55%-59%), Low (50%-54%), Lower (45%-49%), and Lowest (below 40%).
3
The BECE is the national standardized examination offered at the time of completion of JHS (i.e., Ninth grade).
3
resources might help improve academic performance through investment in educational
resources and needs. While empirical research has not shed adequate light on this possibility in
Ghana, evidence from Elliott’s (2011, 2013) research on the effects of economic instability on
children’s educational outcomes in the US suggests that economic resources provide families
with an advantage. Families with accumulated assets are able to access educational resources—
such as textbooks and writing materials—that are proven to affect academic performance
(Chowa, Masa, Wretman, & Ansong, 2012; Etsey, 2005; World Bank & UNICEF, 2009).
The GoG has made efforts to address many of the challenges facing the educational
system. However, most of these efforts have been geared toward achieving universal primary
education by 2015—Millennium Development Goal 2 —and less about improving academic
performance. The GoG’s efforts to increase enrollment at the JHS level have had little or no
impact on academic performance (Atta-Quayson, 2007), one of the major barriers to successful
transition from JHS to SHS. Enrollment rates at the JHS level have increased steadily since the
1980s, while academic performance has decreased over the same period (Akyeampong, 2009;
DANIDA, EFW, & The World Bank, 2011; Okyerefo, Fiaveh, & Lamptey, 2011; The World
Bank, 2010). See Appendix A for enrollment trends of public and private JHS from 2001 to
2010. Using econometric estimation models, Osei and others (2009) find that government tuition
subsidies do not have a significant impacts on BECE pass rates. Thus, ensuring access through
an increased focus on enrollment does not guarantee other important educational outcomes, such
as regular attendance, active engagement in schoolwork, and good academic grades. Without
efforts to improve academic performance, many JHS students will not obtain passing grades in
the BECE to qualify for admission into SHS.
4
High Cost of Transition
Apart from low academic performance, the significant increase in educational expenses is
another major barrier to successful transition to SHS. Basic education (primary and JHS) has
been made mostly free through GoG programs that subsidize educational costs by providing
School Capitation Grants (free tuition), a School Feeding Program (free school lunch), school
uniforms, and textbooks (Osei, et. al., 2009). The GoG provides these subsidies at the basic
education level partly because the Constitution of the Republic of Ghana4 stipulates that basic
education is a fundamental human right for all Ghanaian children, and must be free to all
regardless of social or economic background (Akyeampong, 2011). However, government
subsidies drop substantially at the SHS level as the Constitution recommends but does not
require the GoG to guarantee or fund second cycle institutions. For instance, in the 2011/2012
academic year, SHS students paid an equivalent of about USD 500 in annual tuition (Public
Agenda, 2011) and almost USD 100 in direct costs of school charges and materials (Innovation
for Poverty, n.d.). Clearly, as one transitions from JHS to SHS, the number of school-related
supplies5 increases and the cost of education rises to more than 50% of many families’ annual
incomes (Ghana Health and Education Initiative, n.d.). The drastic increase in educational
expense may account for the high dropout rate (70%), which is higher during the transition from
JHS to SHS than the dropout rate during JHS (22%). This may be an indication that many
families cannot bear the cost of SHS education. See Appendix B for illustration of the rates of
transition from primary school through SHS.
4
Chapter 5 of the 1992 Constitution of the Republic of Ghana stipulates that (a) basic education shall be free,
compulsory, and available to all; (b) secondary education in its different forms, including technical and vocational
education, shall be made generally available and accessible to all by every appropriate means, and in particular, by
the progressive introduction of free education.
5
Common school-related supplies at the SHS level include school uniforms, school bags, exercise books, textbooks,
and other stationeries.
5
School-related expenses create enormous challenges for individuals and households,
especially those who fall within low-income brackets. Data from the Ghana 2003 Core Welfare
Indicators Questionnaire (CWIQ) survey reveal that persons from households in the highest
wealth quintile in Ghana are nearly three times likely to have completed middle school than
persons in the lowest wealth quintile (Oduro, 2008). This gap could be due to poor households’
inability to afford school expenses. Like many developing countries, Ghana’s economy is
growing, and there are competing demands on meager public funds available for social programs
like education. Conventional mitigating measures—such as scholarships and fellowship
opportunities—at the SHS level are scarce, thus individuals and families have to find ways to
pay for educational expenses.
As Ghana moves toward a model of more cost sharing in education at the SHS and higher
levels, individual- and family-owned economic resources—income and assets—may become an
even more important factor in determining whether or not one can afford schooling. Knowing
that they are able to pay for current and future educational expenses enables students to
concentrate on their academic performance. It is becoming increasingly clear that provision of
free access to school, particularly beyond the primary school level, is unlikely because of the
continued rise in population, high levels of inflation, low economic growth rates, and competing
demands on limited government funds (Maliyamkono & Ogbu, 1999). Lewin (2008) predicts
that the already low SHS enrollment level will continue to plunge unless the GoG allocates at
least 5% of Ghana’s gross domestic product (GDP) to education6. In terms of percentage of total
education expenditure, public expenditure dropped from 19% to 18% for JHS and from 19% to
12% for SHS from 2004 to 2008 (DANIDA, EFW, & The World Bank, 2011). Taking the
6
In 2011, 8.1% of GDP was spent on education, which was 24.4% of the total government (UNESCO Institute for
Statistics, 2013).
6
current trends in education funding into account, cost sharing will continue to be a key part of
government policies at higher levels of education.
In principle, the Free Compulsory Universal Basic Education (FCUBE) program initiated
in 1996 to increase access and ensure that parents pay only limited non-tuition fees has not
completely removed the financial burden on parents (UN Girl’s Education Initiative [UNGEI],
n.d.). There are over 70 different types of fees ranging from UDS 0.33 to UDS 13.13 that parents
must pay in addition to stationery, transportation, and 10% of the cost of required textbooks (The
World Bank & UNICEF, 2009). These costs have resulted in 11% of children working while in
school and 3% dropping out entirely (The World Bank & UNICEF, 2009). Despite the current
share of education expenses already borne by students and parents, some government officials
are calling for increment in the amount of fees and levies they are required to pay (Caillaud et al.,
2009; Lewin, 2008). Thus, government subsidies for education are not likely to increase enough
to absorb costs, a situation that raises the questions: Does the lack or limited availability of
economic resources affect students’ educational expectations? How does this, in turn, affect
effort and performance in school?
This dissertation will examine whether access to economic resources can lead to a strong
sense of self-efficacy and positive educational expectations, which may motivate students to
focus on their academic work in order to perform well in school.
Significance of the Study
Much of what we know about the impacts of economic resources on educational
outcomes is derived from studies in developed countries and other African countries. Findings
suggest that economic resources—owned mostly by parents—are positively associated with
children’s school-related behaviors (e.g., regular school attendance) (Zhan & Sherraden, 2011;
7
Elliott & Beverly, 2010) and educational outcomes (e.g., graduation, better grades, and
progression) (Elliott & Beverly, 2010; Kim, 2010; Loke & Sacco, 2011). While these results may
suggest that there is a relationship between economic resources and education in Ghana, there
are knowledge gaps. For instance, the mediating mechanisms through which economic resources
affect academic performance have not been fully understood (Elliott, Choi, Destin, & Kim,
2011). Knowing such mediation mechanisms will clarify the processes that underlie the nature of
the relationship between economic resources and academic performance.
Also, the educational effects of economic resources—particularly of those owned by
youth in SSA—are not well documented (Deshpande & Zimmerman, 2010), and the few
empirical studies in developing countries have been limited to small sample sizes or specific
countries (Deshpande & Zimmerman, 2010). There are nuances such as different measures and
important covariates that empirical research does not yet account for (Chowa, Ansong, & Masa,
2010).
In addition, economic, social, and political differences exist between Ghana and
developed or even fellow SSA countries (Green, 2013; Pasquali & Aridas, 2013). Therefore,
results from different settings may not be generalizable to the Ghanaian context (Fentiman, Hall,
& Bundy, 1999), and the mechanisms by which economic resources are associated with
academic performance are less clear.
By helping address these gaps, this study seeks to inform practitioners and policymakers
on how economic empowerment opportunities may create educational advantage for Ghanaian
youth and their households. Findings from this study and others can guide efforts to reverse the
lagging academic performance at the JHS level, one of the major barriers to educational
progression.
8
Purpose of the Study
Further research is needed to shed light on the relationships between economic
resources—income and assets—and academic performance in Ghana. Thus far, empirical
evidence from around the world suggests potential connections, but there are significant research
gaps on how specific types of economic resources (e.g. household income versus assets) affect
academic performance. In the SSA context, it is not clear whether other factors fully or partially
mediate the relationships between different types of economic resources and academic
performance, and whether these possible direct and indirect relationships differ for girls and
boys. This dissertation uses data from a random sample of Ghanaian youth and parents
participating in the YouthSave Ghana Experiment to provide clarity on the relationship between
economic resources (income and assets) and academic performance (math and English language
subject scores). Specifically, this dissertation aims to examine:
1. the kinds of household economic resources associated with youth’s academic
performance,
2. the mediating role of academic self-efficacy and expectations in the relationship
between economic resources and academic performance,
3. whether the unmediated (i.e., direct) and mediated (i.e., indirect) relationships
between economic resources and academic performance vary by gender.
History and Structure of the Ghana Education System
A brief review of its history and current structure is necessary to put the Ghanaian
educational system into perspective. I divide the educational history of Ghana into three
epochs—precolonial, colonial, and postcolonial. Prior to the introduction of Western literacy to
Ghana in the 1530s by the Dutch and Danish (Education Encyclopedia, n.d.), young people
9
developed knowledge and skills through indigenous education in the form of social or
apprenticeship training. This training covered practical, social, spiritual, and intellectual needs
appropriate for the environment and community (Egbo, 2000). Artisans and professionals taught
crafts, trades, and other professions through apprenticeships (Obide, 1995). Males and females
received instructions in different activities.
Indigenous education was overshadowed and dwarfed by western educational policies in
the colonial era. Beginning in the 1830s, missionaries introduced Western literacy as a means of
spreading Christianity (Education Encyclopedia, n.d.; Ghana Education Service [GES], 2004;
Graham, 1971, 1976). Later, colonialists trained some older youth in Western literacy so they
could assist in running the colonies. The colonial administration gradually introduced formal
education and later built infrastructure for schools (Obide, 1995). By the end of the colonial era
in the late 1950s, Ghana had moved toward mass formal education and a Western educational
system. Later in 1961, the country passed the Education Act (Act 87), which pushed for the
construction of more schools and adoption of the concept of universal primary education (GES,
2004; Oduro, 2000).
The postcolonial era has seen a gradual move from an agricultural-based economy—with
24.65% of the country’s GDP in this sector—to an industrial- and service-based economy
(Kolavalli, et. al., 2012), which has resulted in an increasing demand for workers who have
mastered numeracy and literacy skills. These structural shifts refocused the approach to
education and employment. The formal educational system has formalized the acquisition of
knowledge and skills. Youth who wish to develop their capabilities for future employment and
boost their earnings prospects may have to go through some form of formal education. This is
recognized generally as the most suitable means of acquiring literacy and numeracy skills and
10
acquiring broader knowledge, reasoning, and problem-solving skills essential for being
successful in the changing economy (National Research Council, 2005).
The educational structure in Ghana has undergone many reforms since the country gained
independence. The most recent and significant changes were the result of the 1987 educational
reforms, which shortened pre-university education from 17 years to 12 years (Tonah, 2009).
Currently, Ghana has the 6-3-3-4 educational structure. Students begin formal education at age
six and spend six years in primary school, three years in JHS, three years in SHS, and four years
in the university. Basic education is universal and covers the first nine years of schooling
(primary school and JHS). The goal of universal education is to expose children to a wide variety
of ideas and skills to (a) build their capabilities to fit into society, (b) stimulate their interests in
specific areas for future training, and career development. Students are taught eight to nine
subjects: mathematics, English language, integrated science, social studies, Ghanaian language
and culture, basic design and technology, information and communication technology, religious
and moral education, and the French language (optional). A school year at the primary and JHS
level lasts 40 weeks. At the end of the third year in JHS, students are tested using scores from an
internal continuous assessment and Basic Education Certificate Examination (BECE) which is a
national standardized examination.
By law, basic education in Ghana is compulsory, universal, and tuition-free. According to
the last population and housing census conducted in 2010, most Ghanaians (53.7%) have had at
least basic education (Ghana Statistical Service, 2012). The current gross enrollment ratio at JHS
is 79.5%, but the completion rate is only 66% (Global Partnership for Education & Ghana
Development Partner Group, 2012). Students are not obliged to continue their education beyond
JHS, but doing so is a goal of many students because in SHS they begin to acquire advanced
11
knowledge, specialized skills, and competencies that increase their higher education
opportunities, enhance their employment prospects and access to most formal jobs. The SHS
curriculum covers core subjects—English language, mathematics, integrated science, and social
studies—and elective subjects in one of five program areas—agriculture, general arts/sciences,
business, vocational, and technical studies.
The current gross enrollment ratio at SHS is 40% (UNESCO Institute for Statistics,
2013). Results of the 2010 census show that among the entire population of Ghana, 23.4% have
no formal education, 33.4% have a primary level education, 26.7% have a JHS education or its
equivalent, 10.9% have an SHS education or its equivalent, and 4.1% have a post-SHS education
(Ghana Statistical Service, 2012). This means more than four in five Ghanaians have never
progressed from JHS to SHS even though an SHS education equips young people with the
knowledge and skills that broaden their chances of future educational prospects, economic wellbeing, and social mobility. Contrasted with previous generations, more Ghanaian youth are
transitioning from JHS to SHS today (Fatusi & Hindin, 2010), but a large pool of youth still do
not progress because of poor performance in JHS and the inability to pay for SHS education.
Definition of Concepts
Youth
In this study, youth are defined as 12- to 18-year-olds enrolled in JHS or SHS. While
there is no universally accepted definition of youth in Ghana, for practical and policy purposes,
the National Youth Policy of Ghana (NYPG) defines it as persons from ages 15 to 35 years
(Ministry of Youth and Sport, 2010). The United Nations’ definition of youth as those aged 15 to
24 years serves as the basis for the NYPG age range. This study limits the definition of youth to
12
12- to 18-year-olds because it is consistent with the age range of typical JHS and SHS students,
the target population of this study.
Other youth characteristics considered in this study are based on socially defined
activities (e.g., school and work) that youth typically are involved in (Jacobo, n.d.). Youth in
Ghana can be categorized in four ways: (1) in school but not working, (2) in school and working,
(3) working only, and (4) neither working nor in school. Given that one of the primary
phenomena of interest in this study is academic performance, the current study focuses on youth
enrolled in the formal educational system, including those who also work. This study uses the
terms youth, young people, and JHS students interchangeably.
Formal Education and Educational Outcomes
Education includes the knowledge, skills, and attitudes that people cultivate to prepare for
future occupations, responsible citizenship, and leisure activities. Formal and informal education
represents acquired human capital and the process of acquiring it, but the focus of this study is on
formal education. Unlike informal education, formal education is more systematic, has a
predefined curriculum and minimum attendance requirements (Dib, 1988).
Typical measures of formal educational outcomes fall under two categories: academic
performance and educational attainment (Kim, 2010). Some of the measures of academic
performance are test scores and subject grades (Elliott, Kim, Jungj, & Zhan, 2010; Huang, Guo,
Kim, & Sherraden, 2010; Jutte et al., 2010; Reynolds & Temple, 1998; Watkins, 1997).
Educational attainment, on the other hand, encompasses achievement of milestones such as
school completion, graduation, and transition to higher levels of education (Elliott & Beverly,
2010; Reynolds, Temple, & Ou, 2010). Other studies have focused on school behaviors,
including school attendance (Nam & Huang, 2009; Elliott & Beverly, 2010), school engagement,
13
and establishing networks and bonding in school (Buhs, 2005; Iyer, Kochenderfer-Ladd,
Eisenberg, & Thompson, 2010) or psychological measures, including academic self-efficacy and
expectations (Kim, 2010; Zhan & Sherraden, 2003).
Different studies model these education-related measures differently as outcomes,
predictors, mediators, and correlates. In this study, educational outcomes are limited to academic
performance as demonstrated by math and English language subject scores. Given that the
literature on educational outcomes has used academic performance and achievement
interchangeably, the current study will use both academic performance and achievement to refer
to educational outcomes, especially when discussing extant literature.
Economic Resources
In this study, the term economic resources has a broad meaning, encompassing both
income and assets. Although these two concepts are connected, they are distinct and therefore
should be treated as separate concepts (Oliver & Shapiro, 1997; Sherraden, 1991; Wolff, 1995;
Zhan, 2006). Income denotes the inflow of financial resources (Sherraden, 1991) and becomes an
asset only if saved for future use (Oliver & Shapiro, 2006). Assets refer to accumulated
resources, including deferred income. Assets can be abstract (i.e., intangible) or concrete (i.e.,
tangible) (Sherraden, 1991; Sparr & Moser, 2007), but a broad definition of assets, though useful
in some cases, “runs the risk of blurring a great many meanings of assets and may not serve very
well in thinking about social issues related to government policies” (Nam, Huang, & Sherraden,
2008, p. 2).
This study focuses on tangible assets (e.g., wealth and physical property), which are the
most relevant for educational outcomes in Ghana. Measuring tangible assets is a more focused
way to gauge/measure economic resources, and studying their impact on educational outcomes
14
ensures that this study draws on the theoretical perspective that assets are a tool for
socioeconomic development of young people (see Nam, Huang, & Sherraden, 2008). Specific
examples of tangible assets include but are not limited to (1) financial savings; (2) stocks, bonds,
and other financial securities; (3) real property; (4) hard assets other than real estate (e.g.,
automobiles); (5) machines, equipment, tools, and other tangible components of production; (6)
durable household goods; and (7) livestock (Nam, Huang, & Sherraden, 2008).
In this study, the term economic resources refers to income and assets, unless specified
otherwise. As necessary, ownership of an economic resource—by student, parent, or
household—will be clarified.
Conclusion
For many in Ghana, progressing beyond JHS is critical to future socioeconomic wellbeing. There are several infrastructural and personnel inadequacies that may contribute to the
problem, but two of the major barriers to progression are low academic performance and
inability to afford fees and school-related expenses. However, individual and household
economic resources might help fund students’ education, which could affect their academic
performance at the JHS level. Emerging theoretical and empirical work on asset effects—
especially in developed countries and some SSA countries (e.g., Uganda)—are the basis for this
reasoning (Chowa, Ansong, & Masa, 2010). However, empirical research has not yet shown the
effects of economic resources on academic performance in Ghana. It is not clear what type of
economic resources or whose economic resources matter most in improving academic
performance. We know little about how the direction and magnitude of the relationship may
change when individual, classroom, and school factors are considered. We do not know whether
economic resources directly influence academic performance or whether behavioral and personal
15
psychological factors mediate these relationships. This study contributes to the knowledge base
by examining these questions and shedding light on how findings may confirm or deviate from
what we know empirically about the relationship between economic resources and academic
performance in other contexts.
16
Chapter 2: Theoretical Framework
Overview
Using the human capital development framework, I address the sources of and
motivations for individual and household investment in formal education. I also explore the
direct and indirect effects of income and assets on educational outcomes.
Human Capital Development
Understanding the relationships between economic resources and educational outcomes
requires a focus on the broader question of what motivates educational investment. Why and
how do others, especially parents and relatives, get involved in education of youth? Why does it
matter to invest in youth education beyond basic school? The human capital development
perspective provides a framework for understanding important aspects of investment in youth
education.
In their seminal work in the early 1960s, Theodore Schultz, Gary Becker, and Jacob
Mincer set the stage for debate on what constitutes human capital and how human capital relates
to future well-being, productive activities, and earnings (Beaulieu & Mulkey, 1995; Hayami,
2009; Nureev, 2010). Human capital is an individual’s stock of skills, specialized knowledge,
experience, and acquired abilities that result from formal and informal education. In this study, I
will focus on the benefits of formal education without discounting apprenticeships, on-the-job
training, and other avenues for human capital accumulation.
A key argument in favor of human capital investment is that knowledge and skills bring
returns to individuals and communities, organizations, and civil society (Hayami, 2009; Kwon,
2009). The human capital development framework is grounded in the assumption that people are
rational and weigh the benefits and costs of their investment in human capital (Becker, 1996;
17
Schultz, 2003). Economists view education as capital good because individuals’ knowledge and
skills affect productivity levels and lead to economic growth through spending of earnings
(Beaulieu & Mulkey, 1995; Ng & Feldman, 2010; Olaniyan & Okemakinde, 2008). In addition
to these economic returns of human capital investment, there are social benefits of self-respect
and honor (Acemoglu & Angrist, 1999; Moretti, 2004). As people acquire more education, they
acquire more competencies that boost their self-worth, and they may become more useful to
others who may need their help.
One underlying assumption of the human capital development perspective is that capital
acquisition is multileveled with individual, family, community, and national levels influencing
knowledge and skills acquisition (Hayami, 2009; Kwon, 2009). At the individual-level,
personality traits that influence human capital accumulation include genetic makeup, physical
and mental qualities, strength, and intelligence (Nureev, 2010). Family- and household-level
characteristics that constrain or facilitate investment in children’s human capital include parents’
education, interests, decision-making power dynamics, and economic status (Akresh, Bagby, de
Walque, & Kazianga, 2012a). Communities and countries contribute to the development of
human capital through education by devoting resources to educational infrastructure.
Human capital development is not always a personal decision (Olaniyan & Okemakinde,
2008), especially for children and youth who depend on parents’, relatives’, and guardians’
investments in their knowledge and skills acquisition. Also, if parents’ resources are limited,
their ability to spend money and time on children’s education will be affected. The type of
investment also depends on “…financial capabilities in the families [and] the prospective
utilization of the capital which is being accumulated” (Mincer & Polachek, 1974, p. S77).
Parents’ financial resources and their knowledge of options, returns, and costs influence
18
investment in their children’s education. Many parents and families may make deliberate
financial sacrifices hoping for future benefits (Hayami, 2009). For example, a Ghanaian parent
might sell or use property as collateral to raise money to support a child’s education in the hope
of better returns in the future.
Similarly, governments’ and other entities’ investments through policies—such as
compulsory school attendance—hinge on the premise of financial capacity and prospects of
future benefits to individuals and society. When the government makes an educational
investment, youth seldom have a say in how it happens. Even when children have resources and
are aware of available human capital investment options, they may not have enough information
about those options or may lack maturity to make informed and rational choices. As Fischhoff
(1992) explains, youth who lack information often do not make the best decisions. Even when
they have adequate information, “…they often do a poor job of understanding the probabilities of
consequences… [and] they are more likely to rely on personal experiences” (Fischhoff, Crowell,
& Kipke, 1999, p. 3). Some youth may not have had the personal experiences that are pivotal in
making informed decisions about educational investments.
Human capital is developed through accumulation over time (Cisternas, 2011; Tamura,
2004). Within the context of formal education, the human capital accumulation process starts
from preschool and builds progressively through higher education and beyond. This means it is
important that individuals go beyond the basic school level if they want to acquire advanced
knowledge and skills to be competitive in the future. More human capital investments (e.g., more
years of education and training) often yield greater returns. In Ghana, for instance, individuals
who acquire formal education up to the basic education level may not be as marketable as those
who attend school beyond JHS (Sackey, 2008). The higher the levels of education attained, the
19
greater the returns. Two individuals with the same job titles but different levels of education
might earn different salaries, even if they are similar in productivity (College Board, 2005; Day
& Newburger, 2002). Research on youth earnings in Harare, Zimbabwe, finds that individuals
with higher education earn about 21% more than those with a GCE O-level (SHS) certificate
(Zhou, 2002). Individuals who have O-level certificates earn 25% more than those without it.
Although more years of education and higher credentials may not always translate into
better performance, having them positively influences the chances of employment and earnings.
These differential returns on human capital investment can motivate parents and youth to strive
for more education or training beyond the basic level to reap more benefits.
Gender also affects the allocation of investments in human capital (Mincer & Polachek,
1974), partly because in many societies, returns on education differ by gender. In developed and
developing countries, occupational differences still exist between men and women. Even when
the labor force has a high representation of women, they frequently are employed in low-earning
activities (Heintz, 2005). In the US for instance, the average salary of female researchers in the
medical field is lower than that of men in the same field, regardless of specialty, rank, leadership
positions, and number of publications (Jagsi et. al, 2012).
If the theoretical argument holds that human capital investment depends on expected
returns, investment in girls’ education is lower than investment in boys’ in many societies
(Jensen, 2010). For some families, the idea that most girls will join their husbands’ families after
marriage means that the return on any human capital investment may not always have direct
benefits to their parents (Foster & Rosenzweig, 2009), which could be a disincentive (Jensen,
2010). Thus, it is important that studies such as this one take into account gender dynamics that
20
influence how people make human capital investment decisions, which may lead to different
outcomes.
The foregoing review of the human capital development perspective focused on
motivations for, sources of, and gender dynamics in human capital investment. Next, I discuss
the theoretical frameworks which explain how income and assets affect academic success.
Effects of Income on Educational Outcomes
Income is an important economic resource that enables people to afford educational
resources and services. The relevance of income to education lies mostly in its ability to fund
important educational expenses, such as daily school lunch, stationery, school supplies, and
tuition. Because of the importance of income for education, governments often step in to provide
income support to families (Mayer, 2010). Indeed, the increase in income-maintenance
strategies—including conditional cash transfer (CCT) programs in Ghana and other SSA
countries—is due to the impact they are having on school enrollment and school attendance
(Garcia & More, 2012).
While the presence of income boosts educational investment, which results in better
educational outcomes, the absence of income has negative effects on educational investments.
When the steady flow of income is interrupted by job loss or other events, investment in
education also may diminish. Some families in developing countries such as Ghana respond to
these income shocks by increasing children’s labor to supplement household income (Chowa,
Ansong, & Masa, 2011; Beegle, Dehejia, & Gatti, 2003; World Bank & UNICEF, 2009).
There are also situations where changes in flow of income compels some parents and
guardians to frequently change jobs and housing locations, which leads to school changes (Ross
& Roberts, 1999). Continuous residential and school mobility is associated with lower math
21
scores, more grade failures, and school behavior problems (McDonnell, Crook, & Chellman,
2012; Ross & Roberts, 1999). Frequent changes of school make it difficult for students to feel
connected to a school and build important alliances and friendships, which are important for
staying on course in school.
Friedman’s permanent income hypothesis (PIH) is a cornerstone of the theorized
connection between income and educational expenditures. His hypothesis is that an individual’s
ability to forecast future income determines how they spend or invest current income (Wang
2006). Someone convinced that human capital will result in more future income is more inclined
to spend current income on its development. Proponents of the PIH expect in-school youths’
future incomes to be higher than their current incomes because they will obtain higher paying
jobs that require superior knowledge and skills. In other words, those who expect education to
result in higher future incomes may be motivated to spend on education now. Studies in Ghana
suggest that the economic motivation to invest in education is boosted further when the
household size is small and the head has formal education (Alhassan, 2010; Donkoh &
Amikuzuno, 2011).
Income is important for human capital development, but relying solely on it may not
result in long-term educational well-being (Sherraden, 1991; Boshara, 2009). When it comes to
education, long-term and short-term investments are both important. Uncertainties about
continuous income may lead to doubts that resources will be available to pay for future
educational costs, especially for students who plan to progress to higher education levels.
Another challenge with the income perspective is that young people in particular may not
have income. As the neoclassical lifecycle theory asserts, income does not flow throughout all
stages of the lifecycle. School-age youth are likely to have limited or no income, which means
22
most depend on employed parents and adult relatives to assist with their educational investments.
However, parents’ and relatives’ incomes often are susceptible to inconsistent flow because of
job loss, death, or illness (Ansong, Chowa, Grinstein-Weiss, 2013; Sherraden, 1991). When
these unexpected events happen, a young person may have no choice but to drop out of school,
unless there are other economic resources to draw on.
In essence, income is important for short-term educational expenditure but may not be the
most reliable means of funding long-term human capital development. For income to have a
significant effect on future educational well-being, it has to create a storehouse of future
consumption through asset accumulation. Thus, in addition to needing income for short-term
educational expenses, the youth and their families also need assets to serve as insurance against
future reductions or losses of income. According to Shapiro and Wolff (2005, p. 12), “wealth
(assets) signifies a command over financial resources that, when combined with income, can
produce the opportunity to secure ‘good life’ in whatever form is needed or desired – education,
training…” With this distinction between income and assets, how distinct are the impacts of
income and assets on educational outcomes? One would expect variations in the magnitude and
nature of the associations because of the inherent differences in these types of resources. A
discussion of the relationship between assets and educational outcomes follows.
Effects of Assets on Educational Outcomes
If income is important but sometimes insufficient for supporting long-term educational
outcomes, can assets fill the gap? The asset-effect perspective asserts that assets can have shortterm, long-term, direct, and indirect impacts on educational outcomes. Michael Sherraden (1991)
hypothesized that beyond consumption effects, assets have important effects on several domains
of individuals’ and families’ economic, psychological, social, and educational outcomes. If
23
assets influence young people’s educational outcomes, by what mechanisms does this happen in
the Ghanaian context? Next, I discuss how the asset-effect framework theorizes the direct and
indirect mechanisms by which assets affect educational outcomes.
Direct Effects of Economic Resources on Educational Outcomes
One may hypothesize a direct link from assets to educational outcomes (Elliott, Kim,
Jung, & Zhan, 2010), because people often rely on wealth and fungible resources to pay for
school-related expenses. As in other SSA countries, Ghanaian children often are denied access to
education because their families cannot afford fees (Agbewode, 2009). Even though basic school
is supposed to be free because of government policies and programs, some schools charge
additional fees and levies that lead some families to tap into their economic resources. Some
pupils cannot go to school because they have not paid their fees (Siziya, Muula, & Rudatsikira,
2007).
Education is high-priority expenditure for many families in Ghana and many African
countries, and some individuals and families accumulate savings from their jobs and other
productive assets for the specific purpose of paying school tuition and expenses for their children
(Bouman, 1995). Those with limited or no financial resources may sell off personal or family
properties, and some use assets as collateral to secure loans (Lerman & McKernan, 2008) to
finance educational investments. Student loan programs in Ghana have formidable problems
(Woodhall, 1992), and obtaining educational loans is a challenge for most individuals, especially
at the SHS level. Basic-level education is subsidized, and student loan programs are readily
available for tertiary-level (i.e., college) students. That leaves SHS-level students with limited
options for funding their education. For those who cannot easily access student loans or the
24
limited scholarships, one option is to use non-moveable personal or family assets (e.g., houses
and land) as collateral for commercial loans.
The direct effects of assets on educational access are more evident in countries such as
Ghana where most individuals cannot depend on the limited number of scholarships available,
which often do not cover living expenses or school-related needs (OECD and The World Bank,
2007). Even students who receive scholarships must bear school-related expenses in the course
of their education, which makes having personal and family assets critical. Kimuyu (1999)
explains that some households in rural East Africa participate in savings schemes such as
RoSCAs to accumulate enough savings to pay school fees. Students who do not have sufficient
financial assets or other forms of economic resources may have to forego school until resources
become available.
Indirect Effects of Economic Resources on Educational Outcomes
Some asset researchers have built on the original conceptualization of direct asset effects
by suggesting that assets also have indirect effects on educational outcomes (Ansong, Chowa, &
Grinstein, 2013; Shobe & Page-Adams, 2001). The indirect asset-effect framework suggests that
assets may change people’s thinking and behavior, which in turn, may affect how students
perform in school. Two pathways—academic self-efficacy and academic expectations—explain
how this happens for JHS students.
Academic self-efficacy pathway: Defined as a belief in one’s ability to accomplish a
particular task successfully, self-efficacy often depends on whether one has the opportunity to
participate in knowledge and skills acquisition through formal education. The sense of control
obtained by accumulating, managing, and spending economic resources also improves selfefficacy (Bynner & Paxton, 2001). This is consistent with the vicarious learning perspective,
25
which suggests that observational learning (i.e., modeling) is a key determinant of self-efficacy
(Bandura, 1997; Gist & Mitchell, 1992; Law & Hall, 2009).
External social and economic factors, including the presence or absence of income and
assets, influence the development of self-efficacy. When these economic resources are available,
people react to them in varying ways that may affect their beliefs and how they feel about their
capabilities (Pajares & Schunk, 2001; Shobe & Page-Adams, 2001). Bynner and Paxton (2001,
p.5) suggest that “having an asset allows a household to be more confident looking to the future
and will enable people to take more active control of their lives.”
As economic resources influence self-efficacy, self-efficacy may in turn affect
educational outcomes (Bean & Sidora-Arcoleo, 2012; Chowa, Masa, Wretman, & Ansong, 2013;
Page-Adams & Sherraden, 1996; Sherraden, 1991). Since the 1980s, motivation researchers and
educational psychologists have focused considerable attention on the relationship between selfefficacy and positive educational outcomes. So far, studies have shown that academic selfefficacy predicts college grades (Bong, 2001; Zajacova, Lynch, & Espenshade, 2005), hours
spent on schoolwork (Torres & Solberg, 2001), and the propensity to stay in school (Zhang &
RiCharde, 1998). Young people’s beliefs about their capabilities influence their task choices and
the effort they put into their schoolwork (Bandura, 1997; Schunk, 1995; Schunk & Pajares,
2002). Students who feel capable tend to work harder and seldom give up when faced with
obstacles (Schunk & Pajares, 2002). Bandura’ social cognitive theory posits that when students
believe in their self-efficacy, they are motivated to use acquired knowledge and skills to tackle
academic challenges, which eventually result in better academic outcomes (Zajacova, Lynch, &
Espenshade, 2005). Confident young people tend to view difficult school assignments as
challenges and work hard to address them (Pajares & Schunk, 2001).
26
In Bandura’s view, students’ feelings of self-efficacy determine how they set personal
academic goals and behave in school. Research on academic self-efficacy of students suggests
that perceived self-efficacy influences the setting of academic goals and final academic
performance (Bandura, Barbaranelli, Gian, & Pastoreili, 1996; Grabowski, Call, & Mortimer,
2001; Zimmerman, Bandura, & Martinez-Pons, 1992). The more students believe in their
abilities, the more they would exercise control over how they think, feel, and behave in school.
Concerns about limitations or a lack of economic resources to pay for educational
expenses can arouse academic stress or doubts, which may disrupt learning abilities (NIH News,
2012; Yousefi, et al., 2010). In the absence of such anxieties and self-doubts, students are
motivated to concentrate on schoolwork, increase their study hours, and seek help from teachers
and/or colleagues, all of which are important for performing well in school (see Oyserman,
Bybee, & Terry, 2006). Even students with superior academic skills may need motivation to use
their academic prowess. In other words, competent and brilliant students need some level of
belief in their own capabilities to excel in school. The availability of sufficient economic
resources to pay for school expenses allows students to be confident and proactively use their
skills. Bandura (1986) explains that, “judgments of efficacy …determine how much effort people
will expend and how long they will persist in the face of obstacles or aversive experiences. The
stronger their perceived self-efficacy, the more vigorous and persistent are their efforts” (p. 394).
In short, the presence of economic resources may influence students’ beliefs about what they can
accomplish. Confidence in one’s capabilities may in turn affect amount of effort and academic
performance.
Academic expectations pathway: Economic resources also affect educational outcomes
indirectly through academic expectations. Most people, including young students, have hopes,
27
goals, and plans for the future. Those who have positive expectations anticipate better
opportunities in the future, which form the basis for goal setting, planning, and behaving in ways
to optimize goal attainment (Christy-McMullin, Shobe, & Wills, 2009; Hideg & Novaky, 2002;
Seginer, 2009).
Motivation for positive expectations often is stimulated by individuals’ interactions with
their physical and social environments (Seginer, 2009). When parents invest substantial
economic resources into their children’s education, they expect their children to perform well
because they have provided for their school needs. Some parents may not only have high
academic expectations of their children but will also get involved in their children’s schooling to
make sure their educational investments are not wasted. Children who are aware of their parents’
educational expectations may be motivated to work harder in school in order not to disappoint
their parents. On the other hand, those who feel overly pressured to perform may be inclined to
rebel against what they perceive as unrealistic expectations and incessant pressure; such rebellion
could negatively affect their academic performance (Akhtar & Aziz, 2011; Peixoto, 2011).
Some theorists suggest that an active process of interconnection and coexistence occurs
between people and their physical environments (i.e., the environment influences people’s
thoughts and behaviors just as people manage and shape their environments) (Werner & Altman,
2000; Werner, Brown, & Altman, 2002). Stimulation to think and behave in certain ways also
comes from children’s physical environments. The ongoing interconnections and interactions
with the physical environment include the processes of accumulating resources and managing
those already accumulated. Managing personal economic resources might prompt people to think
about future possibilities. Young people who assist their parents manage a productive asset (e.g.,
rental properties such as a commercial vehicle or house) may think about what they can do to
28
accumulate similar assets or achieve even greater success in the future. Conversely, youth from
families with fewer or no assets may not even consider attaining higher education level because
they lack the financial resources to sustain the process. It is possible that some people with few
resources may also be even further motivated because of their desire to experience better life
situations that will be different from the difficult experiences of their parents. Notwithstanding,
their perception of what might be feasible, in terms of being able to pay for higher education,
may become an obstacle. In order for students from poor households to intensify and sustain
their hard work toward higher education (i.e. future identity), they must first be convinced that
resources are available to help pay for future tuition and other school related expenses (Elliott,
Chowa, & Loke, 2011).
Stimuli that influence thoughts and behaviors vary in intensity, frequency, duration,
number of sources, and type (Hutchison, 2008; Sundstrom, Bell, Busby, & Asmus, 1996). This
means that different levels, sources, and types of economic resources or expectations could have
varying effects on behavior. Hutchison (2008) explains that an overload or deprivation of
environmental stimuli (i.e., physical and social) can have negative influences on human
behavior. That means overly high parental expectation or lack of any form of parental
expectation could hinder academic achievement.
Young people have sophisticated cognitive abilities that allow them to respond to stimuli
from available economic resources. They can reconstruct past and current events and imagine
likely future outcomes of their actions (Taylor & Pham, 1996). This cognitive stimulation is
reflected in their expectations for and worries about the future. Depending on how young
people’s experiences and social contexts influence their thoughts, they may develop fears or
hopes that affect how they behave and plan for the future. For instance, the fear of not
29
transitioning from JHS to SHS because of associated costs might lead a young person to give up
on schooling altogether. Conversely, a young person who has financial security because of their
personal or family income and assets may be more optimistic about the future and would think
about how they can take advantage of available resources.
Once people have pondered and decided how they would wish their future to look, they
engage in behaviors that are consistent with their thoughts and expectations. The theory of
reasoned action (TRA), developed by Fishbein and Ajzen in the mid-1970s, explains the
relationship between cognitive processes and actions. The TRA holds that beliefs and intention
(i.e., cognitive representation) are immediate antecedent of behavior (Fishbein & Ajzen, 1975;
Rutter & Bunce, 1989). In other words, individuals’ assessment of future consequences plays a
critical role in deciding how to behave. After evaluating likely outcomes, people may engage in
exploratory behaviors, such as seeking advice and gathering information about available options.
Behaviors also could be direct commitments to a specific option (Seginer, 2009), such as
engaging more seriously in academic work and studies (Horstmanshof & Zimitat, 2007).
The Asset Experience
Related to the idea that assets have indirect effects on educational outcomes is the idea of
asset experience. Bynner and Paxton (2001) expanded the original concept of asset effects by
suggesting that one should understand asset effects in terms of not only psychological, social,
and economic outcomes but also benefits acquired during asset accumulation, management, and
spending. This process—termed the asset experience—could serve as a learning opportunity for
many young people in Ghana. Some parents and guardians involve their children in asset
accumulation efforts to impart entrepreneurial and money management skills. Many
communities in Ghana historically have had community norms and traditional institutional
30
structures that foster occupational socialization of younger members of the household (Apt,
2005; Apt, Blavo, & Opoku, 1992; Centre for Social Policy Studies [CSPS], 2001). As Mbebeb
(2009) explains:
In the African traditional setting, occupations and vocational training of the youth were
tied to family or kinsman. The primary obligation of the family was to niche the child for
a comprehensive task in the family and society. Socialization of family members and
kinsmen into economic roles became a cultural value in respect of the old age utilitarian
education practice. Learning most occurred through imitation, observation and
apprenticeship where learners were socialized into life skills in a life space. Learning in a
real life situation had direct impact on vocational awareness, entrepreneurial drive and
employability. (p. 26)
Young people who participate in accumulation and management of economic resources gain
knowledge and develop their human capital, both of which are valuable to their future careers
and endeavors. In many communities and cultures, youth have responsibilities in a family
business—a type of asset—which teaches them how, where, and when to accumulate economic
resources. Young people still in school may bring their resource accumulation and management
experience to the classroom, which may influence academic performance. Like internships, the
asset experience might offer helpful practical experiences and opportunities to apply theories
taught in the classroom. This is consistent with the participatory action research approach, an
effective framework for helping youth take meaning from what they learn in school (Engelman
& Hazel, 2010). In fact, young people who participate in internships, job shadowing, and SHSbased businesses gain early occupational experience and vocational exploration (Shanahan,
Mortimer, & Kruger, 2002).
31
Benefits of resource accumulation and management might be more evident at the SHS
level where students start specializing in specific areas and courses of study. For example, a
student specializing in business studies or economics might be able to relate theoretical
knowledge to their work during weekends and long vacations during which students typically
help their families manage businesses.
However, involvement of school-age youth in asset-generating activities can be
problematic because it could get in the way of school work. In a systematic review of studies on
assets and child wellbeing, Chowa, Ansong, and Masa (2010) found that certain types of asset
accumulation might interfere with young people’s schooling and other developmental outcomes
in developing countries. Nevertheless, there is a learning component of asset accumulation.
Building and managing assets may teach youth what to repeat or avoid, and these valuable life
lessons can be applied to coursework and future choices.
Summary
This chapter reviewed the human capital development perspective that attempts to
explain motivations for investing in education. Also discussed are the effects that income and
assets may have on educational outcomes. The nature and amount of assets that people own play
a significant role in their ability to access educational resources and services for personal
development. Similarly, the chapter draws on the asset-effects framework combined with the
social cognitive theory (SCT) and theory of reasoned action (TRA) to explain how the
management and use of assets and interactions with them can change people’s behaviors. Direct
and indirect effects of assets are critical if they are to have a significant impact on young
people’s education. The forgoing discussions on theoretical perspectives of income and asset
ownership offer logical ways through which economic resources can influence educational
32
outcomes. The key question is whether there is enough evidence to support the theoretical
explanations. The next chapter reviews empirical studies on the nature of the relationship
between economic resources and educational outcomes.
33
Chapter 3: Empirical Evidence
Overview
This chapter presents empirical evidence of the connection between economic resources
and educational outcomes in developed and developing countries. Although existing studies
examine a wider range of ages than the current work, results suggest that there are positive
relationships between economic resources and educational outcomes for youth of all ages.
Studies included here have found positive, negative, and mixed effects, and some test moderated
and mediated effects of economic resources. Because other factors affect educational outcomes, I
conclude with a review of literature on individual, classroom, and school covariates that might
mediate, suppress, or confound the relationship between economic resources and educational
outcomes.
Evidence from Developed Countries
Empirical studies in developed countries—mostly the US and Europe—suggest that
income and assets have direct and indirect positive impacts on several educational outcomes,
including school grades, enrollment and graduation (Corwyn & Bradley, 2002; Destin &
Oyserman, 2009; Nam & Huang, 2009; Williams Shanks, Kim, Loke, & Destin, 2010).
Effects of Income on Educational Outcomes
Most studies that investigate the relationship between income and educational outcomes
find significant positive associations (Nam & Huang, 2009). A study by Acemoglu and Pischke
(2001) that uses data from three longitudinal surveys of high school graduates in the US finds
that a 10% increase in family income is associated with 1.4% increase in the likelihood of
enrolling in college. Brooks-Gunn, Klebanov and Duncan (1996) follow children from birth to
34
age 5 and find that parents’ income during a child’s first three years has a strong effect on the
child’s intelligence test scores, regardless of family structure and mother’s education and age.
Using National Longitudinal Survey data, Dahl and Lochner (2012) find that a $1,000 increase in
income improves children’s academic test scores by 6%.
Other studies investigate the relationship between income and other types of educational
outcomes. One of the few studies that have investigated the relationship between income and
school behavior finds that higher family income is associated with lower risk of school dropout
among ninth graders in US urban schools (Randolph, Fraser, & Orthner, 2006). The study also
finds differences in dropout by race but not gender. Another study by Hilger (2012) finds causal
effects of parental income on college enrollment in the US and that females have slightly
higher—but not statistically significant—enrollment rates than males. Belley and Lochner (2007)
also find that higher family income is strongly positively associated with educational attainment,
which is consistent with the hypothesized association between income and educational outcomes
in Chapter 2.
The evidence of positive relationships between income and educational outcomes in the
US is consistent with findings from European studies. In Norway, Loken (2010) uses
longitudinal data from 1967 to 1969 to examine the causal effects of permanent family income
on children’s educational attainment and finds a positive relationship. Using data from the
British Cohort Study (BCS) and the British Household Panel Study (BHPS), Blanden and Gregg
(2004) find that family income has a causal relationship with educational attainment.
While most studies that link income and educational outcomes find strong positive
relationships between income and children’s academic performance, others find mixed or
contradictory results. In a national survey of families in the US, Corwyn and Bradley (2002) find
35
no statistically significant effect of household income on European American children but a
small effect on children in minority groups, including African Americans and Hispanics. Using
data from the Panel Study of Income Dynamics (PSID), Shea (2000) finds a negligible impact of
parents’ permanent income on children’s years of schooling, especially when fathers have more
years of education. Shea also finds that results are more robust for boys, possibly because fathers
tend to bequeath more resources to sons than daughters. Another study by Chevalier, Harmon,
O’Sullivan, and Walker (2005), finds that controlling for parental education, permanent income
does not significantly explain whether or not a student will drop out of school by age 16.
Though most income-related studies examine the direct relationship between income and
educational outcomes, few have specified and tested the mediating role of parental expectations.
For example, Davis-Kean (2005) uses a structural equation modeling approach to show that
parents’ income indirectly affects academic performance through their expectations and behavior
and that girls perform slightly better than boys on reading tests. Using data from Baltimore
fourth-grade students and their parents, Alexander, Entwisle, and Bedinger (1994) find that
parental expectations directly affect test scores and school behavior and also mediate the
influence of income and home environment on educational outcomes. The current study
examines similar relationships by assessing whether income is directly associated with academic
performance in Ghana or indirectly through youth academic expectations, and whether this
varies by gender.
Effects of Assets on Educational Outcomes
In developed countries, different types of assets are connected to different educational
outcomes. Using data from the 1979 National Longitudinal Survey of Youth (NLSY79) Orr
(2003) finds that U.S. households’ income-producing assets (i.e., estates, farm, stocks, and
36
bonds) were signficantly associated with standardized math scores, regardless of children’s
gender. Destin and Oyserman (2009) also used datasets from Chicago and Detroit to investigate
the impact of financial assets on adolescents’ planned academic effort and expectations for
higher grades. The study reveals that when students believe financial assets are available to pay
for school, they tend to have higher expectations of their grades and plan to put more effort into
their school work. Destin and Oyserman (2009) also find no significant variation by gender.
Some studies investigate the effect of family homeownership and consistently find
positive impacts on children’s educational outcomes. In one such study, Zhan and Sherraden
(2003) find that mothers’ homeownership is associated with children’s higher grade point
averages (GPAs), but results are estimated to be signficantly higher for girls than for boys. Other
studies find that homeownership is associated with improved math and reading test scores
(Essen, Fogelman, & Head, 1978), and school attendance (Aaronson, 2000; Green & White,
1997). Lovenheim (2011) finds that college enrollment increases by 5.7% when housing value
increases by $10,000 in the US. What is not clear is the relationship between homeownership
and educational outcomes is whether the positive impacts of homeownership are due to the
stability of residence, neighbor characteristics or the homeownership itself.
Youth-owned assets, such as savings in a bank or money kept at home, may also have
implications for several educational outcomes. Using data from the US PSID and its Child
Development Supplement, Elliott (2008) documents a positive relationship between youthowned economic resources and academic outcomes. Elliott finds that young people aged 18 and
younger who have savings designated for school score higher in math than those who do not
have such savings. The study also controls for gender and finds that while boys score 3.71 points
higher on math tests than girls, girls are 31% more likely to expect to attend college.
37
Another dimension of the conceptual connection between assets and educational
outcomes is the mediational role of other factors, including parents’ and youth’s expecations and
future orientation. Sherraden (1991) explains that assets boost self-efficacy and create a sense of
orientation toward the future. Evidence—mostly from the US—supports this hypothesis that
assets may lead to psychological outcomes (e.g., expecations, self-efficacy, and future
orientation). A study by Elliott, Kim, Jung, and Zhan (2010) finds that parents’ asset ownership
affects their expectations, which indirectly affects their children’s math scores. This is congruent
with Zhan and Sherraden’s (2003) finding that mothers’ expectations of children’s educational
success act as a mediator in the relationship between assets (e.g., home ownership and savings)
and educational achievement. Using data from the NLSY79, Zhan (2006) finds that parents’
future expectations mediate the association between parental assets and educational outcomes.
Kim (2010) similarly finds that children’s educational expectations partially explain the positive
impact of net worth and financial assets on school dropout rates.
Findings from the studies discussed above are consistent with the theorized
relationships—direct and indirect—between economic resources and educational outcomes, but
there are nuances in the directions and mechanisms of the relationships. Economic resources
(e.g., income and assets) and educational outcomes are not always positively associated, and the
relationships between them are not always statistically significant. For example, an individual
could become overly confident about educational outcomes, which could result in complacency
of effort. Children from advantaged families might not work hard in school because their
parents’ economic resources assure a secure future.
Some studies have found no statistically significant association between assets and
educational outcomes. Elliott, Kim, Jung, and Zhan (2010) find a positive relationship between
38
wealth (net worth, homeownership, and children’s savings for school) and children’s math scores
among Caucasian parents but not among African American parents, and Kim (2010) finds that
income is associated with other educational outcomes (high school completion, college
attendance, and college degree attainment) but not high school dropout rates. Similarly, a study
by Orr (2003) in the United States finds that income-generating assets are statistically
significantly associated with math scores but non-income-generating assets are not associated
with math scores. Variation in type and measurement of educational outcomes, ownership
structure of economic resources (i.e., whether the resources are owned by youth, parent, or
houshold), and other socio-demographic factors (e.g., gender) could impact results. Some
variation could be attributed to methodological issues and dataset limitations. For instance,
studies use varying approaches to account for differences in outcomes by gender, including
splitting samples by gender, modeling interaction term between gender and other variables, or
entering gender as a control variable in regression models. This study will help fill some of the
research gaps by examining how gender and asset ownership may affect academic performance
as measured by math and English subject grades.
Evidence from Sub-Saharan Africa
Given the evidence of mixed but mostly positive direct and indirect associations between
economic resources and educational outcomes in developed countries, can we expect similar
results from Ghana and other SSA countries? Research is lacking, but rigorous empirical
research is emerging in Ghana (Chowa, Masa, Wretman, & Ansong, 2013). Until recently,
studies on economic resources and education in Ghana have focused on resources owned by
parents and other family members and have had inconsistent results. Empirical evidence
discussed in this section suggests that the relationship could be positive, negative, or both.
39
Effects of Income on Educational Outcomes
Empirical research on the independent effects of income on educational outcomes in SubSaharan Africa (SSA) is limited, and most studies examine income effects in the larger context
of household socioeconomic status (SES). Tansel (1997) conducts one of the few empirical
studies on the independent effect of income on educational outcomes and finds that household
income in Ghana and Côte d’Ivoire is the most significant predictor that families will invest in
children’s education and that children will advance beyond basic education. Acheampong (2007)
also finds that Ghanaian children in households with limited income are less likely to attend JHS
or transition from JHS to SHS than children from more advantaged families. Conversely,
Cockburn and Dostie (2007) report that household income in Ethiopia negatively affects school
attendance because students’ work time competes with school hours. This finding is consistent
with other studies in Ethiopia by Rose and Al-Samarrai (1997), and Weir (1997) that find that
income opportunities such as child work are associated with low school enrollment. This means
some children miss school because they are engaged in income generating activities.
Some studies in SSA investigate how cultural beliefs and practices related to gender roles
influence the relationship between income and educational outcomes. Using data from
Mozambique’s Household Survey of 2002/2003, Bilale (2007) finds that household income is
significantly associated with enrollment and academic performance and that, boys in primary
school perform better than girls. Handa, Simler, and Harrower (2004) use data from
Mozambique’s Household Survey of 1996/1997 and find that household income is strongly
associated with primary school enrollment, with boys in rural areas recording significantly
higher enrollment rate than girls. In a study of 1,725 households in Guinea, Glick and Sahn
40
(2000) find that increases in household income have a significantly greater influence on girls’
enrollment than boys’.
The few studies in SSA that have disentangled the effects of income from other economic
resources and socio-economic status indicators have mostly ignored the potential mediational
role of psychological factors (e.g., self-efficacy and expectations). Empirical research—such as
what is presented in the current study—is needed to test the hypothesized indirect pathway from
economic resources (e.g., income) to educational outcomes.
Effects of Assets on Educational Outcomes
Other studies on economic resources and educational outcomes in SSA have focused on
household assets and generally reveal a positive association between asset ownership and school
enrollment, attendance, and completion (Filmer and Pritchett, 2001; Montgomery, Grant,
Mensch & Roushdy, 2005; Montgomery & Hewett, 2005). Filmer and Pritchett (1999) use
survey data from 35 developing countries, including Ghana, to examine how household wealth
affects educational attainment of 15- to 19-year old youth. Using an asset index as a measure for
household wealth, the authors find that wealth is positively associated with basic school
enrollment. According to the study, in regions like Central Africa, many children from lowincome households never enroll in school, possibly because such low-income households lack
resources to pay for school-related expenses. Cockburn and Dostie (2007) also find that a
family’s ownership of assets—in the form of ploughs and sickles—increases school attendance.
Perhaps the amount of time youth are required to work on the farm is reduced—and
subsequently the amount of time youth are able to spend in school is increased—if the family
owns assets such as farm equipment. Clearly, more rigorous research is needed to investigate
these asset ownership mechanisms.
41
Just as there is limited mediational studies on the relationship between income and
education in SSA, few empirical studies investigate assets’ influence on educational outcomes in
SSA. One of such few studies is the 2005 study by Ssewamala and Curley who find that
Ugandan students with access to a savings account have positive plans and become more
confident about them. Although the authors did not specifically test mediational roles of these
psychological factors, they suggest that they are important antecedents to changes in school
behavior and improvement in academic performance.
A number of studies in SSA examine the differences in how assets’ influence educational
outcomes for rural and urban youth as well as for girls and boys. Studies by Jensen and Nielson
(1997) and Buchmann (2000) find that although land ownership has a negative effect on school
attendance, the negative relationship tends to be stronger in rural areas. Because of their
proximity to owned land, families in rural areas may expect children to work on farmlands and
they may perceive it as the best future career for the children as opposed to formal education,
whereas children in urban areas likely live too far from the land for farm work to be feasible.
Using data from the 2000 Multiple Indicator Cluster Survey in Senegal, Montgomery and Hewett
(2005) find that children from households in the lowest wealth quintile are less likely than those
in higher wealth quintiles to ever attend school or complete at least primary education. The
authors also reveal that the absence of assets exerts a stronger negative influence in rural areas.
Montgomery and Hewett believe this rural–urban difference can be explained by urban families’
relatively easier access to educational infrastructure. This suggests that rural–urban differences in
infrastructure might moderate the relationship between parent and household economic resources
and educational outcomes.
42
One might argue that those living in rural areas—regardless of their economic
resources—would have comparatively low educational outcomes because school infrastructure is
lacking. This may be true for some but not all rural dwellers. In rural areas with poor academic
performance and low attendance, family resource constraints might worsen the already poor
educational experience of students. For example, the Resource-Dilution perspective explains that
when resources cannot support the education of all children in the household, parents might ask
some children, particularly younger ones, to stay home. If resources become available, out-ofschool children in the household may then go to school.
Unlike the consistent results in the rural–urban variation, studies show mixed results for
the moderational effect of gender in the relationship between assets and educational outcomes.
For instance, Lincove (2009) examines the association between household wealth and school
attendance in Nigeria and finds significant relationship with girls’ schooling but not boys’.
Similarly, Ssewamala, Karimli, Han and Ismayilova (2010) find that children’s savings accounts
in Uganda have a significantly larger positive influence on adolescent girls’ educational plans
than boys’. In contrast, Curley, Ssewamala, and Han (2010) do not find gender variations in the
relationship between children’s savings accounts and better educational outcomes in Uganda.
While studies in Ghana reveal that educational attainment does vary by gender (Abukari, 2010;
Dunne & Leach, 2007; Mensch & Lloyd, 1998), research has not yet sufficiently explained the
role of gender dynamics. Thus, this study will test the moderation effect of young people’s
gender in the relationship between economic resources and their academic performance.
Generally, assets positively influence education outcomes directly (e.g., by financing
educational needs) or through psychological factors (e.g., self-efficacy and expectations).
However, is it possible for some kinds of assets to influence educational outcomes negatively?
43
Assets that require a significant amount of time to manage or maintain may negatively influence
school attendance. Cockburn and Dostie (2007) show that the ownership of permanent crops
decreases school attendance for boys and that maintenance of household farms requires a great
deal of time and effort by all family members, including children, which may take students’ time
away from schoolwork. Empirical evidence from Ghana, Kenya, and Ethiopia also shows that
some types of assets negatively affect school attendance. Using data from Ethiopia, Admassie
(2002) finds an inverse relationship between household ownership of cattle and school
attendance. A family’s ownership of cattle or other large livestock may increase the probability
that youth must herd animals during school hours and attend school when work permits. Youth
who work may not perform well in school because of the stress associated with combining the
two activities (Chowa, Ansong, & Masa, 2010). Also, youth who work before or after school
may not have time to do homework or may have difficulty concentrating in class because of
fatigue. In a study of Kenyan youth aged 17 years and younger, Moyi (2007) finds that family
ownership of arable land is associated with a child’s propensity to combine farm-work with
schooling. Similarly, a study by Bhalotra and Heady (2003) shows that, an increase in family
farmland ownership negatively affects girls’ school attendance and participation. These negative
effects are consistent with Lerman and McKernan’s (2008; p.198) admission in their in-depth
discussion of asset-effects that, “…problems can arise in the process of building and managing
assets.”
Some studies show that economic resources have both positive and negative impacts on
educational outcomes. For example, Jensen and Nielson (1997) use data from Zambia’s Priority
Survey II and find that savings and assets, such as bicycle, refrigerator, and plough, have positive
effects on school attendance, while land ownership has negative effects. Using data from 596
44
households in Kenya to examine individual, family, and contextual factors that affect school
enrollment, Buchmann (2000) finds that children in households with more financial resources are
more likely to enroll. However, the study also finds that children whose families own land are
more likely to work rather than attend primary and secondary schools. According to these two
studies, land ownership has a negative effect on schooling. Because of the appreciation most
Ghanaians have for land ownership—partly because of its association with wealth and power
(Gilman, 1984)—it is possible that land ownership in Ghana also has profound effects on
children’s educational well-being.
As mentioned earlier, studies on occupational socialization of children suggest that many
youth assist their parents to manage assets such as shops and businesses (Kisselburgh, Berkelaar,
& Buzzanell, 2009; Lucas, 2006). Some young people must herd livestock daily as part of their
household responsibilities. Combining this type of work with school or doing such work during
school hours impedes educational attainment (Chowa, Ansong, & Masa, 2010).
The mixed results suggests the need for further conceptual and empirical research of the
contextual effects (e.g., student age, income level, and number of parents in the household) in the
relationships between assets and educational outcomes in developing countries. Younger
children who are unable to manage assets may not benefit from the asset experience (Bynner &
Paxton, 2001), but they may then have more time to focus on schoolwork and fewer worries
about economic resources.
Additional Correlates and Predictors of Educational Outcomes
Correlates and predictors of educational outcomes, which may have implications for the
effects of economic resources, fall into three major categories: individual, household/family, and
school/community.
45
Individual-Level Correlates
Individual-level factors that affect educational outcomes include biological and
psychological well-being. A study by Johnson, Wolke, Hennessy, and Marlow (2011) finds that
children born prematurely have relatively lower levels of academic attainment and a higher
prevalence of learning difficulties. The study also finds that neonatal factors (e.g., breech
delivery and abnormal cerebral ultrasound) independently predict educational outcomes by age
11. Fentiman et al., (1999) also find that the age at which a child is first enrolled in school is
strongly associated with dropout rates.
How students think about schooling is a predictor of performance. A student’s selfconcept, self-efficacy, attitudes, and motivations positively affect academic performance (Ajayi,
2012; Bandura et al., 1996; Jegede et al., 1997; Kiamanesh, 2005). In Nigeria, self-concept and
academic motivation are significant predictors of students’ attitudes towards math, personality
traits such as agreeableness and extraversion are associated with student educational aspirations,
and academic performance are associated with educational aspirations of students (Ajayi, 2012;
Jegede et al., 1997).
Around the globe, behaviors that reflect students’ perceptions and experiences have an
influence on academic performance. In Italy, one study shows that students who believe in their
academic self-efficacy and have educational aspirations not only have high achievement but also
exhibit positive social behaviors, have low behavioral and emotional problems, and are accepted
by their peers (Bandura et al., 1996). In addition, school attachment and commitment positively
influence academic performance (Stewart, 2007).
46
Household- and Family-Level Correlates
Household and family factors (e.g., parents’ educational level, involvement, and
expectations) affect youth educational outcomes. Research suggests that parents’ educational
level is positively associated with children’s educational outcomes. Perhaps it is easier for
educated parents to help their children with homework because they have literacy and numeracy
skills. In addition, educated parents may show more interest in their child’s school performance
and may follow up regularly on how they perform in school, which could influence children’s
perceptions and subsequent behaviors. Although parents’ educational level generally is
associated with children’s educational outcomes, the magnitude of the relationship may vary by
geographical and cultural contexts. For instance, empirical evidence shows that fathers’
educational level has a stronger positive effect on children’s educational performance in Ghana
and Cote d’Ivoire (Tansel, 1997) and in Uganda, Kenya, Tanzania, Malawi, Mozambique, and
Zambia (Smith & Barrett, 2010). Mothers’ educational level, on the other hand, has a stronger
influence on children’s educational outcomes in Guinea (Glick & Sahn, 2000), Botswana,
Namibia, Lesotho, and Swaziland (Smith & Barrett, 2010).
Similarly, parental expectations seem to affect children’s educational performance (Zhan,
2006). Most empirical studies show that children perform better in school when parents are
involved (Fantuzzo, McWayne, Perry, & Childs, 2004; Muller & Roberts, 2000; Nyarko &
Vorgelegt, 2007; Topor, Keane, Shelton, & Calkins, 2010), but a few show that parental
involvement may be associated negatively with children’s educational performance (Izzo &
colleagues, 1999). Could this be that parents become involved only after problems have arisen?
Some parents actively become involved in their children’s education outside school hours, and
others help in the management and running of the school. For instance, many parents participate
47
in school governance through parent-teacher associations (PTA). The nature of parental
expectations and the way such expectation are communicated to children could encourage them
to work hard and perform well in school.
School- and Community-Level Correlates
School and community factors—in the form of human and physical resources—also
affect children’s educational outcomes. The qualification of teachers and the student–teacher
ratio may make a difference in the quality of education that students receive. Huebler (2008)
cites data from UNESCO to show that SSA’s average student–teacher ratio (40.7:1) is the
highest contrasted with other regions, including Southern Asia (37.8:1), Southeast Asia (26.5:1),
and the world (24.6:1). Indeed, more than 90% of countries in SSA have a student–teacher ratio
higher than 40:1, with Mozambique (67.4:1), Rwanda (65.9:1), and Chad (63.2:1) being the
highest (Huebler, 2008). Ghana has a comparatively lower student-teacher ratio of 30:1
(DANIDA, EFW, The World Bank, 2011).
When large numbers of students are enrolled in a school or the school day is divided into
shifts (i.e., with teachers being responsible for teaching a day’s worth of lessons in back-to-back
groups of students to serve more children in one facility), student performance and teacher
satisfaction suffers (Lee et al., 2005; Michaelowa & Wittmann, 2007). Using data from South
Africa’s Third International Mathematics and Science Study, Howie (2005) finds that heavier
teacher workload is associated with poor performance on math tests. Lee, Zuze, and Ross (2005)
use data from 42,000 students in 23,000 schools in 14 SSA countries to show that students tend
to have low reading test scores in schools in which teachers have large numbers of students.
However, not all empirical studies find negative or significant impacts of high pupil–teacher
ratios, heavy teacher workloads, and large class sizes on educational outcomes. A study in
48
Ethiopia by Abraha, et al., (1991) finds no statistically significant relationship between class size
and girls’ performance. These non-significant and mixed results suggest that variation in other
factors such as asset ownership may have effects.
Teachers’ commitment, qualifications, and experience also affect educational outcomes.
Using 2003 World Bank data on Indonesia, Suryadarma et al., (2006) finds that a lack of teacher
commitment—as measured by teacher absenteeism—negatively affects students’ performance on
math tests. Another study of fifth graders from Burkina Faso, Cameroon, Cote d’Ivoire,
Madagascar, and Senegal shows that teacher absenteeism has a statistically significant negative
effect on students’ performance on standardized tests (Michaelowa, 2001).
Teachers with better training and experience may be able to use their expertise to enhance
students’ learning. Nyagura and Riddell (1993) present evidence that seventh graders in
Zimbabwe are more likely to do well on math and English tests when taught by trained teachers.
A study by Michaelowa (2001) also finds that the qualification of teachers—through initial
education—and regular training are associated with higher standardized test scores of fifth
graders. Similarly, Lee, Zuze, and Ross (2005) find a strong direct relationship between the
quality of teachers and sixth graders’ reading comprehension abilities in Botswana,
Mozambique, Namibia, and Seychelles.
Other physical resources such as classrooms, furniture, and textbooks also influence
educational outcomes. For instance, the availability of textbooks is positively associated with
math performance in Zimbabwe (Nyagura & Riddel, 1993) and higher standardized test scores in
Francophone SSA countries (Michaelowa, 2001). Many schools in SSA do not have instructional
materials (e.g., textbooks, blackboards, libraries, and laboratory equipment) or other basic needs
(e.g., running water and electricity), which affects the quality of teaching and learning (Bilale,
49
2007; Howie, 2005). When some of these missing resources are available in students’ homes,
students do better in school. Using second-wave data from the Southern and East African
Consortium for Monitoring Education Quality (SACMEQ), Smith and Barrett (2010) find that
when books, artificial lighting, and reading tables (i.e., desks) are available to engage and
support learning in the home, students perform better in school.
Conclusion
In conclusion, the extant literature shows that economic resources—whether income or
assets—are associated with different educational outcomes in varied ways. What is unclear in the
case of Ghana is the relative importance and independent effects of (a) income and (b) assets on
academic performance. What we know thus far are inferences from studies conducted in other
SSA countries and—most often—developed countries that are economically, socially, and
politically different from Ghana. The key question is whether these mixed findings are applicable
to the Ghanaian context. This dissertation will help clarify how the two broad types of economic
resources—income and assets—are associated with academic performance in Ghana.
Also clear from the empirical evidence reviewed in this chapter is that income and assets
are not the only possible determinants of educational outcomes. Several individual, household,
classroom, school, and community characteristics are associated with educational outcomes,
regardless of the presence or absence of economic resources (Chowa, Masa, Wretman, &
Ansong, 2013). This means that a rigorous test of the relationship between economic resources
and educational outcomes cannot ignore these contextual factors. Another important question
that needs attention is whether the direction and magnitude of the relationships between
economic resources and academic performance is significantly different for boys and girls in
Ghana, which is addressed by a multi-group modeling approach in this study.
50
Chapter 4: Methodology
Overview
This chapter describes the proposed conceptual framework and study methodology for
addressing the main research questions of this study. Then the chapter presents research
hypotheses, description of variables in the models, and multilevel structural equation modeling
(MSEM) strategies used.
Conceptual Framework
Theoretical perspectives and empirical evidence discussed in Chapters 2 and 3 suggest a
positive relationship between economic resources and academic performance, but the process by
which this occurs is unclear. The current study offers three potential pathways—direct effects of
economic resources, indirect effects through academic expectations, and indirect effects through
academic self-efficacy—to explain the link between economic resources and academic
performance (Figure 1). The relatively few studies on this subject in SSA conceptualize and
examine this connection as a direct relationship (Curley, Ssewamala, & Han, 2010; Ssewamala et
al., 2008).
Direct Effects of Household Economic Resources
Figure 1 shows a direct pathway (Path c) from economic resources (i.e., parents’ income,
and household assets) to academic performance, as measured by English and math scores. This
direct pathway captures the idea that asset accumulation leads to educational advantage by
increasing individuals’ control over educational resources (Bouman, 1995; Elliott, 2011). For
instance, students can use household savings to pay for tuition, extra classes or lessons,
examination fees, and transportation to school, all of which have direct influence on students’
school performance (Kimuyu, 1999).
51
Figure 1. Illustration of a multiple mediation design with two mediators. (A) Economic resources
affect academic performance. (B) Economic resources exert indirect effects on academic
performance through academic self-efficacy and expectations.
Independent Variables/
Covariates
Mediators
Individual level
1. Age
2. Gender
3. School attendance
4. Planned effort
5. Commitment to school
Academic
self-efficacy
Path a1
Household level
1. Parents’ income
2. Household assets
3. Household size
Dependent Variables
Path b1
Academic
performance
1. English scores
2. Math scores
Path c
Path a2
Youth’s academic
expectations
Path b2
School level
1. Average class size
2. Teachers’ qualification
Indirect Effects of Economic Resources through Self-Efficacy and Expectations
Elliott et al. (2011) observe that the indirect “linking mechanisms” connecting economic
resources to educational outcomes are missing from most studies. Based on prior theorization
and empirical studies, this study suggests that two indirect pathways exist. First is the path from
economic resources to academic performance through academic self-efficacy (Path a1 and Path
b1). Second is the path from economic resources to academic performance through academic
expectations (Path a2 and Path b2). This study suggests that parents’ incomes and household
assets will result in young people’s higher academic expectations and feelings of self-efficacy,
52
which in turn will motivate them to be serious about their education and behave in ways that
improve their performance in school.
It is important to note that these hypothesized relationships exist in a multilevel
framework because individual students grow up and interact within dual contexts: their
households and schools. This means that, besides the influence of individual preferences and
dispositions on their educational outcomes, other factors at the household and school levels exert
influence on their expectations, self-efficacy, and ultimately academic performance. Therefore,
the direct and indirect associations between economic resources and academic performance
should be conceptualized from a multilevel perspective.
Before these hypothesized multilevel relationships can inform policy, practice, and future
research in Ghana and SSA, it is important to test them because of several unknowns. First, if the
theorized indirect pathways hold in the SSA context, it is not clear whether the hypothesized
mediators fully or partially mediate the relationships between all types of economic resources
(i.e., parents’ income and household assets) and academic performance in English and math
subjects. Second, it is unclear whether the type of economic resource (i.e., income or asset) has
different effects on outcomes. Third, it is unclear if these possible direct and indirect
relationships affect girls and boys differently.
Research Questions
The study proposes six research questions and corresponding hypotheses based on the
above conceptual model and the gaps in existing research. They are categorized into three
relationships: direct, indirect, and moderated.
Direct Relationships
Research question 1: Do economic resources directly affect academic performance?
53
Hypothesis 1a: Parents’ income is positively associated with young people’s English
scores.
Hypothesis 1b: Household assets are positively associated with young people’s
English scores.
Hypothesis 1c: Parents’ incomes are positively associated with young people’s math
scores.
Hypothesis 1d: Household assets are positively associated with young people’s math
scores.
Indirect Relationships
Research question 2: Do economic resources indirectly affect academic performance by
increasing young people’s academic self-efficacy?
Hypothesis 2a: Academic self-efficacy mediates the association between parents’
income and young people’s English scores.
Hypothesis 2b: Academic self-efficacy mediates the association between household
assets and young people’s English scores.
Hypothesis 2c: Academic self-efficacy mediates the association between parents’
incomes and young people’s math scores.
Hypothesis 2d: Academic self-efficacy mediates the association between household
assets and young people’s math scores.
Research question 3: Do economic resources indirectly affect academic performance by
increasing young people’s academic expectations?
Hypothesis 3a: Academic expectations mediate the association between parents’
income and young people’s English scores.
54
Hypothesis 3b: Academic expectations mediate the association between household
assets and young people’s English scores.
Hypothesis 3c: Academic expectations mediate the association between parents’
income and young people’s math scores.
Hypothesis 3d: Academic expectations mediate the association between household
assets and young people’s math scores.
Moderated Relationships
Research question 4: Are the direct effects of economic resources on academic performance
different for boys and girls?
Hypothesis 4a: Gender moderates the direct association between parents’ income and
young people’s English scores.
Hypothesis 4b: Gender moderates the direct association between household assets and
young people’s English scores.
Hypothesis 4c: Gender moderates the direct association between parents’ incomes and
young people’s math scores.
Hypothesis 4d: Gender moderates the direct association between household assets and
young people’s math scores.
Research question 5: Are the indirect effects of economic resources on academic performance
through academic self-efficacy different for boys and girls?
Hypothesis 5a: Gender moderates the indirect association between parents’ incomes
and young people’s English scores through academic self-efficacy.
Hypothesis 5b: Gender moderates the indirect association between household assets
and young people’s English scores through academic self-efficacy.
55
Hypothesis 5c: Gender moderates the indirect association between parents’ incomes
and young people’s math scores through academic self-efficacy.
Hypothesis 5d: Gender moderates the indirect association between household assets
and young people’s math scores through academic self-efficacy.
Research question 6: Are the indirect effects of economic resources on academic performance
through academic expectation different for boys and girls?
Hypothesis 6a: Gender moderates the indirect association between parents’ income
and young people’s English scores through academic expectations.
Hypothesis 6b: Gender moderates the indirect association between household assets
and young people’s English scores through academic expectations.
Hypothesis 6c: Gender moderates the indirect association between parents’ income
and young people’s math scores through academic expectations.
Hypothesis 6d: Gender moderates the indirect association between household assets
and young people’s math scores through academic expectations.
Research Design and Setting
This study uses a cross-sectional, multilevel research design to establish the existence of
correlational relationships between economic resources and academic performance (Kumar,
2011). The cross-sectional data are from a baseline survey taken in the ongoing YouthSave
Ghana Experiment, one of the largest social experiments in the developing world to test the longterm impacts of youth savings accounts on educational, psychosocial, health, and financial
capability outcomes and household financial well-being. YouthSave is a consortium project and
is funded by the MasterCard Foundation. Save the Children implements the project in
partnership with the Center for Social Development (CSD) at Washington University in St.
56
Louis, the New America Foundation (NAF), and the Consultative Group to Assist the Poor
(CGAP). CSD is leading the research agenda for the project.
As part of the YouthSave project, HFC Bank in Ghana is offering savings accounts to
JHS students in randomly assigned experimental schools. HFC Bank is a leading universal
banking institution in Ghana that offers diversified financial services, including commercial,
investment and mortgage banking and micro financing. The YouthSave Consortium selected
HFC Bank from a pool of interested financial institutions because it was the best fit for the
project.
The intervention includes intensive marketing of a tailored youth savings product called
Enidaso Account (which literally means Hope Account in the Akan language) to 50 experimental
schools and offers financial education and in-school banking to half of the experimental schools.
Youth who open accounts receive a photo ATM card that they can use only to check their
balance. There are no charges for checking account balance. Withdrawals are not allowed for the
first three months after account opening. Parents must accompany youth to withdraw from the
account, but the youth can make deposits by themselves. Deposits are branch-based for half of
the experimental group and both branch- and school-based for the other half.
As mentioned earlier, the current study focuses on only the baseline data of the
YouthSave experiment and therefore this study does not compare the experimental group with
the control group. See Appendix C for an illustration of how the current study fits into the larger
YouthSave project.
Sampling
Because YouthSave is an experiment, YouthSave research team used probability
sampling to select junior high schools as the primary sampling unit and students as the secondary
57
unit in the project. Specifically, the sampling frame included all public JHSs in eight of the ten
administrative regions of Ghana. Two regions—Upper East and Upper West—were excluded
because HFC Bank, the financial institution partner of the YouthSave project in Ghana, does not
operate in those regions. After creating the sampling frame, the YouthSave research team used a
simple random sampling approach to choose 100 JHSs to be included in the study. Again,
because the larger YouthSave project is an experiment, the 100 schools were assigned randomly
into 50 experimental schools and 50 control schools. However, the present study does not
attempt to establish causality and so it uses the baseline sample as a single observational sample
without comparing the experimental to the control schools (Hofler, 2005).
In obtaining the individual-level sample, the research team selected 60 students from
each school using the “random sampling with replacement”7 approach to ensure that each student
had an equal and independent probability of being selected (Kumar, 2011). To allow for attrition,
participants were oversampled (Rootman, 2001) with a final sample size of 6,252 JHS students.
Researchers interviewed parents of youth participants, successfully surveying 73% (4,576) in
person, which resulted in a youth sample of 6,252 and a parent sample of 4,576. Because
household and parents’ characteristics (e.g., parents’ income and household assets) are central to
the current study, youth whose parents were not interviewed at baseline are excluded from the
current study.
Data Collection
This randomly sampled observational study merges three types of data sets for statistical
modeling. The first are baseline survey data obtained by 40 interviewers from 4,576 JHS
7
With the sampling with replacement approach, elements randomly selected from the sampling frame are put back
into the sampling frame before another element is selected from the pool. This approach ensures that each subject
has an equal and independent chance of being selected.
58
students in schools and their parents at homes and workplaces. YouthSave researchers from
CSD, the University of North Carolina, and the Institute of Statistical, Social and Economic
Research (ISSER) at the University of Ghana trained interviewers to collect survey data. To
ensure quality and rigorous data collection, interviewers were required to have prior data
collection experience with ISSER and a bachelor’s degree. The second type of data comes from
copies of youths’ report cards kept by the schools. Interviewers worked with headmasters of the
selected schools to transfer the youths’ data—including math and English language scores and
school attendance—onto a data abstraction form (Appendix D). The third type of data come from
one-page self-administered surveys of headmasters and include school characteristics of
infrastructure, personnel/staff, classroom conditions, and programs organized by the school (e.g.,
excursions and free school feeding programs).
The institutional review boards of the University of Ghana, Washington University in St.
Louis, and the University of North Carolina approved all data collection methods.
Instruments
Researchers carefully developed and tested all four data collection instruments—the
youth questionnaire, parent questionnaire, school record data abstraction form, and headmaster
questionnaire—and adapted questions from established scales to ensure relevance to youth in
Ghana. Based on suggested best practices, researchers used a multi-method pretesting protocol
(Miller, 2003; Schneider et al., 2009; Willis, 2005), and a panel with expertise in measurement
theory and educational, health, and financial capability outcomes reviewed the instruments to
identify potential problems. Next, the researcher team pilot tested the instruments in one private
and three public JHSs to validate the applicability and appropriateness of the questions and
scales. Pilot testing involved 55 interviewer-administered face-to-face surveys, 20 cognitive
59
interviews, and 6 in-depth qualitative interviews. The final data collection instruments are
proprietary and will be available upon completion of the YouthSave project.
Measures
Dependent variables
The two dependent variables are youth’s overall math scores and English scores during
an academic term. Students’ performances in math and English language or reading subjects
have been used consistently as proxies for overall academic performance in other studies in
Ghana (Chowa et al., 2013; Etsey, 2005; Osafo-Acquah & Asamoah-Gyimah, 2009) and in other
countries (Loke & Sacco, 2011; Williams Shanks, 2007; Yeung & Conley, 2008; Zhan, 2006).
Using math and English subject scores as proxies for academic performance is based on the
reasoning that developing literacy and numeracy skills is fundamental for success in a range of
developmental areas, including social, economic, technical, and scientific training (Department
of Education and Skills, 2011).
Each outcome variable includes continuous assessment and final examination scores.
Continuous assessment scores reflect student performance on learning activities—including
paper-and-pencil tests, in-class and take-home assignments, and oral presentations in class—
completed throughout the academic term and are supposed to be a formative measure of
students’ learning. The Ghana Education Service (GES) adopted the continuous assessment
program during the 1987 education reforms with the goal to improve teachers’ assessments of
students’ ability (Amuah, 1996).
The study uses a ratio of 70:30 as weights for the final examination and continuous
assessment scores, respectively. This weighting criterion is consistent with the GES weighting
60
system at the time of baseline data collection (Dery & Addy-Lamptey, 2010; Keteku, 1999).
These outcome variables are continuous variables ranging from zero to 100.
Independent Variables
The two independent variables—parents’ income and household assets— are types of
household economic resources.
Parents’ income. Parents’ income is a continuous variable that measures all sources of
parents’ income in a typical month. Because incomes in developing countries like Ghana
fluctuate from month to month and season to season, interviewers emphasized during data
collection that parents should report the typical monthly income during the last 12 months. This
best practice was adapted from the household questionnaire of the Ghana Living Standards
Survey Round Four (GLSS4, n.d.), and researchers expected respondents to report average
income. Parents self-reported income from full-time jobs, part-time jobs, income from rental
properties, pensions, remittances, and other sources, all in Ghana Cedis (GHC).
Household asset ownership. This is a continuous variable measured by an asset index
constructed using an approach recommended by Filmer and Pritchett (2001) and Filmer and
Scott (2008). The survey asked youth about family ownership of 19 asset items in three broad
categories: nine household possessions (e.g., radio, electric or gas stove, kerosene stove, electric
iron, box iron, refrigerator, television, cellular phone, and landline phone), six types of livestock
(e.g., cattle, goats, sheep, donkeys, pigs, and chickens), and four transportation-related property
(e.g., bicycle, motorcycle, canoe or boat, and car or truck). Moser and Felton’s (2007)
suggestions for constructing asset index informed the types of assets included in this study. The
study used principal component analysis to determine the weight of each asset item. Using the
equation, Ai = (b1a1i) + (b2a2i) + … (bkaki), where Ai is the asset index for the “i” household, a1i,
61
a2i, …, aki are the k indicators of the 19 asset items, and b1, b2, …, bk are weights used to
aggregate the indicators into an index. A positive index value indicates a high level of asset
ownership and a negative value indicates low level of assets. The use of a composite asset
variable is a standard practice in asset-related studies (Ansong, Chowa, & Grinstein-Weiss,
2013; Chowa, 2008; Elliott, Kim, et al., 2010; Huang, Guo, Kim, & Sherraden, 2010).
Hypothesized Mediators
This study models two variables as potential mediators in the relationship between
economic resources and educational performance: academic self-efficacy and youth’s academic
expectations.
Academic self-efficacy. Academic self-efficacy is measured by an 8-item subscale that
determines “perceived capability to manage one’s own learning behavior, to master academic
subjects, and to fulfill academic expectations” (Muris, 2001, p.146). Muris’ original 5-point scale
was expanded to an 11-point scale for YouthSave, which is consistent with Bandura’s guidelines
for constructing self-efficacy scales (Bandura, 2006). Respondents answered questions about
how easily they can perform certain tasks (e.g., study for a test, pass a test, pass all subjects,
finish homework, pay attention in class, satisfy parents with schoolwork, focus on their studies,
and get teachers to help with schoolwork). This study uses confirmatory factor analysis (CFA) to
reduce the academic self-efficacy items into one self-efficacy index. A positive index value
indicates a strong sense of self-efficacy and a negative value indicates a weak sense of selfefficacy. The data analysis section of this chapter explains this data reduction process.
Academic expectation. Academic expectation is a continuous variable that measures
scores students expect to receive at the end of the academic term in math and English language
subjects. This is a self-reported score ranging from zero to 100 percentage points. Math
62
expectations are included in all math models, and English expectations are in all English models.
The academic expectation questions were adapted from studies conducted in Chicago and Detroit
by Destin and Oyserman (2009) and the “I Can Save” project survey initiated by CSD
(Sherraden et al., 2007).
Covariates
Evidence presented in Chapters 2 and 3 indicate that personal, household, school, and
community characteristics might affect educational outcomes by suppressing or confounding the
effects of economic resources. To understand the independent effects of economic resources on
educational outcomes, these additional factors—described below except community
characteristics—are controlled for in all statistical models. Community characteristics are not
controlled for because community level data were not collected as part of the YouthSave project.
Individual-level characteristics. The study controls for youth’s age and school
attendance. Age is a continuous variable measured in years. Many education-related studies
typically account for age (Chowa, et. al., 2013; Elliott, Chowa, & Loke, 2011). School
attendance is a continuous variable measured as the number of days that a study participant
attended school in the last academic term. The attendance records were obtained from the
attendance registers of schools in the study.
The study also controls for students’ commitment to school and planned effort.
Commitment to school is a continuous variable measured using a nine-item, 11-point Likert-type
scale ranging from 0 (strongly disagree) to 10 (strongly agree). The commitment to school scale
is an adapted version of the Rochester Youth Development Study (Thornberry, Lizotte, Krohn,
Farnworth, & Jang, 1991). Youth were asked to describe their attitudes toward school, doing
homework, and getting good grades. A high value indicates high level of commitment to school.
63
This study uses confirmatory factor analysis (CFA) to reduce the commitment to school scale
into one index. The data analysis section of this chapter explains the data reduction process.
Planned effort is a continuous variable measured in hours per week. The planned effort question
was adapted from Destin and Oyserman (2009). Youth were asked, on average, how many hours
a week they spend on school work outside of school. A high value indicates higher level of
planned effort.
Household-level characteristic. Household size is included in all multilevel structural
models as a household-level control variable. It is a continuous variable measured by the number
of people who are economically dependent (i.e., rely for food, shelter, clothing, and other basic
needs) on the head of household. Studies control for household size because of its influence on
determinants of academic performance (Bachman, 2002; Chowa, et al., 2013; Marks, 2006).
Parent’s expectation was excluded from the final models because it was highly collinear with
household size.
School-level characteristics. Two school-level characteristics known to influence
academic performance (i.e., student-to-teacher ratio and teacher qualification) are controlled in
all structural models. The student-to-teacher ratio for each school is a continuous variable.
Research in SSA shows that the number of students relative to number of teachers in a class
affects math performance (Nyagura & Riddel, 1993), standardized test scores (Michaelowa,
2001), and quality of teaching and learning (Bilale, 2007; Howie, 2005; Smith & Barrett, 2010).
Teacher qualification is a continuous variable that measures the number of teachers
without postsecondary certification in each school in this study. This variable was included
because other studies from Ghana show that the quality of teaching and learning in the classroom
64
partly depends on academic and professional qualifications of the teacher (Agyemang, 1993;
Etsey, 2005).
Data Analysis Plan
This study employs multilevel structural equation modeling (MSEM) approach to address
the research questions. The multilevel approach ensures that hierarchical clustering of youth in
schools is accounted for. Multilevel analysis is recommended when the intraclass correlation
(ICC), a measure of the proportion of variance in academic performance that is accounted for by
between schools variation, is greater than 10% (Lee, 2000). In this study, ICC was found to be
.4578 for math and .4023 for English (See Equations 1). These high ICCs are an indication that
variation between schools account for 45.78% of the variability in math performance and
40.23% in English performance (See Equation 1 below).
Equation 1
ICC =
ơ 2 u0
____________________________
(ơ 2u0 + ơ 2r)
where ơ 2u0 is the estimate of school-level variance, and ơ2r is the estimate of individuallevel variance.
Math performance ICC =
128.07
= 0.4578 (i.e., 45.78%)
(128.07 + 151.67)
English performance ICC =
114.53
= 0.4023 (i.e., 40.23%)
(114.53 + 170.14)
Structural equation modeling (SEM) approach is combined with the multilevel approach
because it allows researchers to test mediation effects efficiently in one model rather than
separate models in multiple steps as in conventional regression analysis (Bowen & Guo, 2011;
Miranda & Russell, 2011: Nasser & Hagtvet, 2006; Zwieg, Yahner, & Rossman, 2012). In all
multilevel structural models, the independent variables and covariates are assessed at Levels 1
65
and 2, and the mediators and dependent variables at Level 1 (i.e., 2-1-1 design). A two-step
MSEM process was used (Kline, 2005; Schumacker & Lomax, 1996). The study assesses the fit
between all theorized (latent) constructs and their corresponding indicators before assessing the
hypothesized direct and indirect relationships between economic resources and educational
outcomes. The two-step method removes the possibility that the structure of individual latent
factors could be the source of misfit when testing the structural model (Regambal & Alden,
2009).
The study also conducts subsample analyses to examine measurement and structural
differences across gender, which tests whether the set of observed variables measure the same
relationships in the male group as in the female group. Differences in results would indicate that
gender moderates the relationship between economic resources and academic performance,
regardless of other control variables. I used the student version of LISREL 9.1 statistical
software program to conduct all analyses.
Recognizing the current debate on misuse of SEM significance testing (Glaser, 1999; Hu
& Bentler, 1999; Kline, 2005), this study based its assessment of model fit on a number of
generally recommended indices: a nonsignificant chi-square statistic (x2) and a root mean square
error of approximation (RMSEA) ≤ .05 with an upper bound ≤ 0.08 for its 90% confidence
interval. Chi-square values and their degrees of freedom were reported as customary, but chisquare values were not used to judge model fit because of their sensitivity to this study’s large
sample size (Thompson, 2004; Warne, 2011). However, chi-squares are used in situations in
which chi-square difference tests are needed to help identify the best fit among a series of
models.
66
Missing data—assumed to be missing at random—in all indicators accounted no more
than 10% of the total sample. Thus, the study uses the model-based regression imputation to
replace the missing data and analyzed all measurement and structural models with the maximum
likelihood estimation method. Response scales of all observed variables in the models were wide
enough to qualify as continuous variables. Maximum likelihood works well under the
assumption that all data are multivariate normal. The study assessed the skewness and kurtosis
statistics of all indicators for possible violations of multivariate assumptions. Skewed variables
were log-transformed. All variables included in the measurement and structural models had
skewness statistics ranging from -1.38 to 1.05 and kurtosis statistics from -1.39 to 2.40. These
ranges fall within the acceptable departures from normality (i.e., skewness values of ±1.5 and
kurtosis values of ±3.5), which maximum likelihood is robust enough to handle (Hau & Marsh,
2004). All MSEM analyses were based on covariance matrices because they produce more stable
test statistics and parameter estimates (Chowa, 2008; Cudeck, 1989).
Measurement Models
The study used data reduction techniques to create indices for the multidimensional
measures of academic self-efficacy and commitment to school. First, I used confirmatory factor
analysis (CFA) to specify the factor structure of these constructs. In conducting the CFA, all
eight self-efficacy indicators and ten commitments to school indicators were specified priori to
loading onto the latent constructs. To improve the model fit, I dropped indicators with factor
loadings below 35. I also allowed errors between some indicators to correlate using theory and
meaningful modification indices as a guide (Hong, 1998; Whittaker, 2012). I standardized all
latent variables in the measurement models by fixing the factor loading of one indicator for each
latent variable to equal 1.0 (Kline, 2005, Ledermann & Macho, 2009).
67
After confirming the factor structures of both constructs, I transformed the self-efficacy
and commitment to school items into separate indices (DiStefano, Zhu, & Mindrila, 2009). I used
the factor scores of the two latent variables as weights to produce composite factors for selfefficacy and commitment to school. Treating the nine indicators as “causal indicators” was
necessary for two reasons. First, this approach is more parsimonious than the more complex
structural model that includes all nine self-efficacy and commitment indicators (Zajacova,
Lynch, & Espenshade, 2005). Moreover, because of limitations on the number of variables that
can be included in a structural model in the student version of LISREL 9.1, including nine
indicators in the measurement model would limit the number of covariates allowed in the main
structural models.
Multilevel Structural Models
Multilevel Mediation Analyses
The study tested three types of associations (i.e., direct effect, partial mediation, and
complete mediation) simultaneously to assess all hypothesized pathways between economic
resources and educational outcomes.
To find support for the direct relationship hypothesis, the model should be a good fit, and
the direct path from economic resources to educational outcomes at the individual-level (Pathw c
in Figure 2) must be significant. The complete mediation hypothesis will hold if the direct
relationship between economic resources and educational outcomes is not statistically significant
when controlling for other covariates (Kenny, 2012). Thus, there will be complete mediation if
Pathw c in Figure 2 is zero or not statistically significantly different from zero. To retain the
partial mediation hypothesis, the overall model should fit the data well, and the parameter
estimate of the direct relationship between economic resources and educational outcomes at the
68
individual-level must be significant (i.e., Pathw c must be significantly different from zero) when
controlling for expectations, self-efficacy, and other variables in the model (Kenny, 2012).
Lastly, the paths from economic resources through the mediator to educational outcomes at the
individual-level (Pathw a and Pathw b) should be statistically significant. If the conditions for
partial mediation are met but the direct relationship (Pathw c) and indirect relationship (the
product of Pathw a and Pathw b) are in opposite directions (i.e., + and –), it would be considered
suppression (MacKinnon, Krull, & Lockwood, 2000; Shrout & Bolger, 2002).
Figure 2. Two-level path diagram for the direct and indirect relationships between economic
resources and academic performance. Pathw = relationships at the individual level (i.e., within
schools); PathB = relationships at the school level.
- Academic self-Efficacy
- Academic expectations
Pathw a
- Parents’ income
- Household assets
PathB a
- Parents’ income
- Household assets
Pathw c
Pathw b
- English scores
- Math scores
Within
schools
Between
schools
- Academic self-Efficacy
- Academic expectations
PathB b
- English scores
- Math scores
PathB c
All multilevel structural models account for the mediation effect of academic selfefficacy and expectations—the two main mediators—simultaneously, which has the advantage
of establishing whether the effect of each mediator is independent of the effects of the other
mediator (Kenny, 2012). Kenny (2012) recommends this approach as long as the two mediators
are not too highly correlated: (r(efficacy and math expectation) = .32; r (efficacy and English expectation)=.35; p<.05).
69
I modeled youths’ math expectations as a mediator in all math models and English expectations
as a mediator in all English models. The study included the following covariates in all multilevel
structural models: four individual-level covariates (i.e., youth’s age, youth’s school attendance
record, commitment to school, planned effort), one household-level covariate (i.e., household
size), and two school-level covariates (i.e., student-to teacher ratio and teacher qualification).
Moderation Test
In addition to accounting for seven covariates in the structural models, the study used the
multi-group approach to assess gender moderation of the relationships between economic
resources and educational outcomes. First, the study analyzed structural models with the full
sample to establish and confirm the best multilevel structural relationships between the main
variables. Next, the models were replicated with separate female and male samples. If the models
showed satisfactory fit, I compared the male-only and female-only models with the full-sample.
To reject the hypotheses that gender moderates the relationship between economic resources and
academic performance, both models being compared have to be statistically significant and
should have similar effect size.
In total, I tested three English models: Model 1a is the English model for the full sample,
Model 1b is the English model for the female-only sample, and Model 1c is the English model
for the male-only sample. Similarly, I tested three math models: Model 2a is the math model for
the full sample, Model 2b is the math model for the female-only sample, and Model 2c is the
math model for the male-only sample.
70
Chapter 5: Results
Overview
This chapter presents descriptive characteristics of the study sample and results of the
structural equation modeling. In addition, the study compares the characteristics of males and
females because one of the main questions addressed in this study was whether gender moderates
the potential relationships between economic resources and educational outcomes. Exceptions
include school-level control variables (e.g., student-to-teacher ratio and teacher qualification),
which were not measured at the individual level. There is less emphasis on interpretation of
parameter estimates because the overarching goal of this study was to test the overall trends in
the relationship between economic resources and academic performance, and also to assess selfefficacy and expectations as possible mediating pathways. The magnitude of parameter estimates
are only interpreted whenever possible.
Descriptive Results
Table 1 presents the descriptive statistics of all dependent and independent variables, the
two mediators, and control variables for the full, male-only and female-only youth samples.
Individual- and Household-Level Characteristics
Fifty-two percent of the sample are female. The average age of the entire youth sample is
15.30 years, which is more than a year older than the expected JHS-level upper age of 14 years.
Males (M = 15.49, SD = 2.07) are slightly but statistically significantly older than females (M =
15.12, SD = 1.88) (t = 6.55, p < .01). The youngest female in the youth sample is 10 years old,
the youngest male is aged 9, and the oldest male and female are aged 26.
71
Table 1. Descriptive characteristics of the full, female-only, and male-only samples
Full Sample (N=4452)
Female (n=2314)
Male (n=2138)
p-value‡
M(SD)
Range
M(SD)
Range
M(SD)
Range
English score
50.88(17.25)
8.20↔100
50.02(17.08)
10↔98.60
51.82(17.38)
8.20↔100
<.001
Math score
50.20(16.92)
4.51↔100
48.60(16.08)
9.20-100
51.92(17.08)
4.51↔99.40
<.001
Parents’ income (log)
13.60(0.09)
-0.10↔0.55
1.05(0.01)
-0.10↔0.55
1.04(0.01)
-0.10↔0.55
.74
Household assets index
2.37(1.79)
-6.33↔13.03
2.29(1.77)
-6.33↔13.03
2.46(1.81)
-5.33↔10.76
<.01
0.02(0.89)
-5.56↔1.65
-0.01(0.87)
-4.49↔1.65
0.02(0.92)
-5.56↔1.65
<.05
English expectations
66.54(16.11)
5↔100
67.18(16.34)
0↔100
67.56(16.60)
4-100
.43
Math expectations
67.36(15.99)
0↔100
65.16(16.49)
8↔100
68.05(15.54)
5↔100
<.01
Age (years)
15.30(1.98)
9↔26
15.12(1.88)
10↔26
15.49(2.07)
9↔26
<.01
Attendance (days)
54.32(7.79)
4↔68
54.90(7.49)
4↔68
53.70(8.07)
7↔67
<.01
Planned effort (hours)
7.53(4.81)
0↔40
7.70(4.90)
0↔40
7.36(4.72)
0↔40
<.05
Commitment to school
77.39(9.32)
36↔90
77.01(9.35)
36↔90
77.77(9.28)
39↔90
<.05
Dependent variables
Independent variables
Mediators
Self-efficacy index
Individual-level covariates
Household-level covariates
Household size
10.14(7.49)
0↔28
10.18(7.41)
0↔28
10.09(7.57)
0↔28
School-level covariates
Student-teacher ratio
25.01(12.36)
6↔69
Teachers without
1.47(2.54)
0↔14
postsecondary certificate
‡ Probability values indicating whether females and males are statistically significantly different on each of the variables in the
models; M = Mean; SD = Standard deviation
72
.70
Young people in the study come from relatively large households. The median and most
frequent household size (i.e., mode) is 4, but the typical household includes almost 5 (M = 4.67,
SD = 2.74) dependents. Statistically, male and female youth in the study live in households that
are not significantly different in size (t = 1.44, p = .15). Female-headed households (M = 4.61,
SD = 2.85) are a bit larger than male-headed households (M = 4.73, SD = 2.62).
Generally, young people are committed to school. The mean commitment to school index
is 77.39 with a possible range 0 to 90. The least committed student scored 36 and the most
committed scored 90 on the commitment to school scale. Statistically, male students (M = 77.77,
SD = 9.28) are slightly more committed to school than female children (M = 77.01, SD = 9.32) (t
= 2.89, p < .05).
With regard to the number of hours students spend on school work after normal school
hours, students spend nearly eight hours on average (M=7.53, SD=4.81) on their school work
every week. Female students spend more hours per week on their school work (M = 7.70, SD =
4.90) than male students (M = 7.36, SD = 4.72) (t = 2.48, p < .05). This difference of less than
one hour of school work is statistically significant.
Overall, school attendance rate is high with the typical student attending school about 54
days (M = 54.32, SD = 7.79) in an academic term8. Female students (M = 54.90, SD = 7.49)
attend school about one day more than male students (M = 53.70, SD = 8.07), which is
statistically significant (t = 7.14, p < .001).
School-Level Characteristics
The average student-to-teacher ratio is about 25 (M=25.01, SD=12.36). About 20 of the
100 schools in the sample have student-to-teacher ratio as low as 15, and others have as high as
8
The total number of school days varied from school to school, ranging from 50 to 68 days.
73
69. In nearly half of the schools (49%), all teachers have post-secondary certification, but other
schools have as many as 14 teachers without post-secondary certificate.9
Dependent Variables
Overall, students’ performance was slightly better in English than math. The average
English score for the full sample was 50.88 (SD = 17.25). Females scored a little lower (M =
50.02; SD = 17.08) than males (M = 51.82; SD = 17.38), which is statistically significant (t =
3.49, p < .001). Females’ English scores range from 10 to 98.60, while males’ scores range from
8.20 to 100. The best and lowest scoring students in English is are males.
In math, the average score for the full sample was 50.20% (SD = 16.92). As in English,
males did better in math subject. Males scored statistically significantly higher (M = 51.92, SD =
17.08) than their female counterparts (M = 48.60, SD = 16.08) in math (t = 6.55, p < .001).
However, the worst male student scored 4.51 in math, while the worst female student scored
9.20.
Independent Variables
The mean of household asset index for the full sample is 2.37 (SD = 1.59) with a range
from -6.33 to 13.03. It is important to note that for this scale, negative values do not mean
negative assets or lack of assets. A negative value means fewer household assets and a positive
value indicates more household assets. Likewise, a value of zero does not mean the household
does not own assets, but rather the median value. Statistically, male youths’ households have
significantly more household assets (M = 2.46, SD = 1.81) than female youths’ households (M =
2.29, SD = 1.77) (t = 3.32, p < .01). The average monthly income for parents (log-transformed)
9
A 2009 report by the Ghana Education Campaign Coalition (GNECC) shows that 24% of JHS teachers nationwide
are untrained and as high as 38% of JHS teachers from rural areas.
74
is 13.60 (SD = 0.09). There is no statistically significant difference in the income earned by
parents of males and the parents of females (t = -0.32, p = .74).
Mediators (Academic Self-Efficacy and Expectations)
The mean of academic self-efficacy index for the full sample is 0.02 (SD = 0.89) with a
range from -5.56 to 1.65. A negative value indicates a weaker sense of self-efficacy and a
positive value indicate a stronger sense of self-efficacy. Statistically, males participants (M =
0.02, SD = 0.92) rate their self-efficacy higher than females (M = -0.01, SD = 0.87) (t = 3.32, p
< .05). The typical youth expects to score above average (i.e., 50%) in both English and math,
but they have slightly higher expectations for math scores (67.36%) than English scores
(66.54%). Statistically, there is no considerable gender difference in expected English scores (t =
0.79, p = .43), but there is a statistically significant gender difference in expected math scores (t
= 6.07, p < .01). Male students expect to score 0.38 higher than female students in math.
Results of Multilevel Measurement Models
The hypothesized multilevel two-factor (e.g., self-efficacy and commitment to school)
structure did not fit the data well: x2=2191.41(df = 155), RMSEA = .08 (90% CI: .08, .09). Respecification of the model with correlated errors resulted in a good model fit: x2 = 189.96(df =
48), RMSEA = .04 (90% CI: .03, .05). See Table 2 for summaries of model fit indices. In the
final measurement model, within-schools (i.e., individual-level) factor loadings of both latent
constructs are above the recommended cutoff of .35, which further indicates a good two-factor
structure (Warne, 2011). The factor loadings for academic self-efficacy range from .50 to .67,
and the loadings for commitment to school range from .36 to .60. Figure 3 shows the within- and
between-schools factor structure and loadings of the academic self-efficacy and commitment to
school.
75
For the purposes of multilevel structural equation modeling, the factor scores of the two
latent variables were used as weights to produce two separate composite factors for self-efficacy
and commitment to school. The Cronbach’s alphas for academic self-efficacy (full sample: α =
.74; males: α = .74; females: α=.75) and commitment to school (full sample: α = .70; males: α =
.70; females: α = .71) had adequate internal consistency reliabilities based on the widely
accepted .70 cutoff (Lance, Butts, & Michels, 2009; Spector et al., 2002).
Table 2. Summary of model fit statistics for measurement and structural models
RMSEA (90% CI)
x2(df)
.04(.03; .05)
189.96(48)
Full sample (N = 4452)
.03(.03; .04)
112.82(33)
Male sample (n = 2138)
.05(.03; .07)
40.57(10)
Female sample (n = 2314)
.04(.03; .05)
66.63(21)
Full sample (N = 4452)
.03(.02; .04)
91.58(33)
Male sample (n = 2138)
.06(.04; .07)
49.38(10)
Female sample (n = 2314)
.04(.02; .05)
55.53(21)
Model
Measurement Model
English Models
Math Models
RMSEA = Root Mean Square Error of Approximation;
90% CI = Lower and upper bounds of a 90% confidence interval
76
Figure 3. Multilevel exploratory factor analysis for academic self-efficacy and commitment to
school scales at the within- and between-school levels
Have a clear image of
myself being
successful in life
.61**
Know how I don’t
want my life to turn
out
.50**
Have a good sense of
what it takes to be
successful as an adult
On the “right track”
for future success
Selfefficacy
.67**
.65**
.55**
Try to make good
choices to increase my
chances for a good
future
.59
I like school
.53**
I belong at school
.36**
Commitment
.44**
Within
schools
I finish homework
.60**
Good grades is
important
Have a clear
image of myself
being successful
in life
I know how I
don’t want my
life to turn out
Have a good sense
of what it takes to
be successful
On the “right
track” for future
success
Between
schools
.61**
.15**
Self-efficacy
.38**
.51**
*
.52**
Try to make good
choices to increase
my chances for a
good future
I like school
I belong at
school
I finish
homework
.71
.53**
.32**
Commitment
.46**
.69**
Good grades is
important
*p < .05, **p < .01, ***p < .001
77
Results of Direct Relationships between Economic Resources and Academic Performance
Direct Relationship between Economic Resources and English Scores
Direct effects of two types of economic resources—household assets and parents’
income—on each dependent variable were tested simultaneously. The overall English model
showed satisfactory fit with the data (x2 = 112.82, df =33, RMSEA = .04). Table 3 presents path
coefficients (b) and standard errors (SE) of the independent variables and mediators for the
within- and between-school models for English scores. Results at the individual level show that
at a 95% confidence interval, household assets (b = -0.01, p > .05) and parents’ income (b = 0.08, p > .05) were not directly associated with English scores, controlling for other covariates in
the full sample English model. Likewise, these direct relationships were not statistically
significant at the school level (i.e., household assets: b = -0.87, p > .05; parents’ income: b=1.65, p > .05). Results for the hypothesized direct relationships between economic resources and
English performance for the full sample at the within- and between-school levels are illustrated
in Figures 4 and 5.
Results for males and females are reported separately from the full sample to highlight
possible gender differences. As the moderation test results show, there is no evidence of
significant gender difference in the English model with respect to direct relationship between
economic resources and English performance. In the female English model, household assets (b
= -0.39, p > .05), and parents’ income (b= -0.05 p> .05) were not directly or significantly related
to English scores at a .05 significance level. Similarly, in the male English model, household
assets (b = 0.17, p > .05), and parents’ income (b= -0.18, p> .05) were not directly or
significantly related to English scores. Evidence from all three English models—full, female, and
male—show that the data do not support the hypotheses (1a–1d) that parents’ income and
78
household assets directly affect English scores. In other words, the amount of parents’ income
and household’s ownership of assets does not directly predict how well young females and males
will perform in English.
Figure 4. Estimated unstandardized path coefficients for direct and indirect association between
parents’ income and English grade for the full sample at the within- and between-school levels.
-1.98*
Academic
expectation
0.21**
-0.08
Parents’
income
-5.18
-69.39
Parents’
income
Academic
self-efficacy
Academic
expectations
English
grade
0.12*
Between
Schools
0.43
-1.65
789.68*
Academic
self-efficacy
*p < .05, **p < .01, ***p < .001;
Within
Schools
English
grade
-0.35
: means path coefficient is not statistically significant
79
Figure 5. Estimated unstandardized path coefficients for direct and indirect association between
household assets and English grade for the full sample at the within- and between-school levels.
Academic
expectations
0.21**
-0.58
-0.01
Household
assets
49.81*
Academic
self-efficacy
English
grade
0.12*
Within
Schools
56.33
Academic
expectations
Between
Schools
0.43
-0.87
Household
assets
1177.29*
*p < .05, **p < .01, ***p < .001;
Academic
self-efficacy
English
grade
-0.35
: means path coefficient is not statistically significant
Although the independent variables of interest—household assets and parents’ income—
are not directly associated with English scores, several covariates are statistically significantly
associated with them. Table 4 presents path coefficients (b) and standard errors (SE) of the
control variables for the within- and between-school English models. In the full sample, two
80
youth characteristics—commitment to school and school attendance—are directly and
significantly related to English performance. Student’s commitment to school is directly
associated with English scores (b = 0.13, p < .05), which means that for every one unit increase
in students’ commitment to school, English scores increase by 0.13 percentage points. There is
gender difference in the magnitude but not the direction of the relationship. In the female-only
sample, youths’ performance in English decrease by .19 percentage points for every one unit
increase in commitment to school (b = 0.19, p < .05). For males, English scores increase by .14
percentage points for every one unit increase in commitment to school (b = 0.14, p < .05).
Like the positive relationship between commitment to school and academic performance,
young people’s school attendance rates are positively and significantly associated with English
scores (b = 0.22, p < .05). Each additional day of attendance results in a 0.22 percentage-point
increase in English scores. The direction of the relationship is the same in the female- and maleonly samples, except that the extent of the positive association is stronger in the male group.
Males’ English scores increase by 0.25% when they increase their school attendance by one day
(b = 0.25, p < .05). For females, English scores increase by only 0.21% for every one additional
day that they attend school (b = 0.21, p < .05).
Student-to-teacher ratio, one of the two school-level covariates, is not statistically
significantly associated with English scores in the full (b = -0.002, p > .05), female (b = 0.01, p
>.05), and male models (b = -0.001, p >.05). Results also show that teacher qualification is not
significantly associated with English scores in all three samples (full sample: b = -0.17, p > .10;
female-only sample: b = -0.08, p > .01; and male-only sample: b = 0.0.34).
81
Direct Relationship between Economic Resources and Math Scores
A separate SEM test was conducted to test the hypotheses that household assets and
parents’ income are directly related to youths’ performance in math (1c-1d). Table 5 presents
path coefficients (b) and standard errors (SE) of the two independent variables of interest and
mediators for the within- and between-school models for math scores. Statistically, household
assets (b = -0.30, p > .05) and parents’ income (b = 0.15, p > .05) are not directly associated
with math scores. Similarly, results at the school level show a statistically nonsignificant direct
relationship between economic resources and math performance. Therefore, the data do not
support the hypotheses that household assets (1c) and parents’ income (1d) are directly,
positively associated with math scores. The within- and between-school results for the
hypothesized direct relationships between economic resources and math performance for the full
sample are illustrated in Figures 6 and 7.
This study also finds that these results of nonsignificant direct relationships are similar
for males and females. Therefore, the data failed to support the moderation hypotheses (4c-4d)
that results of the direct relationships between household assets and math scores (females: b =0.37, p > .05; males: b = 0.13, p > .05) and between parents’ income and math scores (females: b
= 0.08, p > .05; males: b = 0.29, p > .05) vary by gender.
82
Figure 6. Estimated unstandardized path coefficients for direct and indirect association between
parents’ income and math grade for the full sample at within- and between-school levels.
0.15
Academic
expectations
0.14**
0.15
Parents’
income
0.33*
30.24*
Academic
self-efficacy
Academic
expectations
Math grade
0.09*
Between
Schools
0.18
-2.86
Parents’
income
-7.86
*p < .05, **p < .01, ***p < .001
*p < .05, **p < .01, ***p < .001;
Academic
self-efficacy
Within
Schools
Math
grade
-0.33
: means path coefficient is not statistically significant
83
Figure 7. Estimated unstandardized path coefficients for direct and indirect association between
household assets and math grade for the full sample at within- and between-school levels.
2.39*
Academic
expectations
0.14**
-0.30
Household
assets
-0.36
-13.12
Academic
self-efficacy
Academic
expectations
Math grade
0.09*
Between
Schools
0.18
0.54
Household
assets
-8.68
*p < .05, **p < .01, ***p < .001;
Academic
self-efficacy
Within
Schools
Math
grade
-0.33
: means path coefficient is not statistically significant
84
Table 6 presents path coefficients (b) and standard errors (SE) of the control variables for
the within- and between-school models for math scores. Unlike household assets and parents’
income, some control variables are significantly associated with total math scores, including
school attendance (b = 0.27, p < .01) and commitment to school (b = 0.08, p < .05). For each
additional day of attendance, math scores increase slightly by 0.27%. Contrary to results from
the English model, there is a weak but statistically significantly relationship between teacher
qualifications and math performance (b = 0.39, p < .10). Although this relationship is significant,
the results are not compelling because the relationship is only significant at the .10 significance
level.
Overall, the nature, magnitude, and number of significant variables in both the English
and math models are very similar. However, while neither parents’ income nor household assets
are directly, significantly related to English scores, commitment to school and attendance are
directly, significantly associated with English and math scores at the .05 significance level. Also,
teacher qualification is positively associated with math scores but only at the .10 significance
level.
85
Table 3. Path coefficients of independent variables and mediators in the English models
Within Schools
Between schools
Model 1w
Full sample
b(SE)
Model 2w
Females
b(SE)
Model 3w
Males
b(SE)
Model 1B
Full sample
b(SE)
Model 2B
Females
b(SE)
Model 3B
Males
b(SE)
Household assets  English score
-0.01(0.41)
-0.39(0.57)
0.17(0.61)
-0.87
2.42
0.33
Parents’ income English score
-0.08(0.19)
-0.05(0.27)
-0.18(0.32)
-1.65(3.45)
-1.66
4.43(2.56)+
Efficacy  English score
0.12(0.03)**
0.11(0.05)*
0.13(0.05)*
-0.35(0.42)
-0.54(0.61)
-0.48(0.54)
Household assets  Efficacy
49.81(23.68)*
-4.00(1.29)**
32.52(9.77)*
Parents’ income  Efficacy
-5.18(5.61)
0.47(0.56)
0.45(5.09)
789.68(183.41)*
6.59(11.69)
326.45
0.21(0.02)***
0.22(0.03)***
0.21(0.03)***
0.43(0.28)
0.37(0.29)
-1.34(0.59)*
Household assets  Expectations
-0.58(1.55)
16.61
-0.86(1.96)
56.33
-598.02
54.46
Parents’ income  Expectations
-1.98(0.75)*
10.84(8.94)
-1.93(1.06)+
Relationships
Direct effects of economic resources
Indirect effects of academic self-efficacy
1177.29(394.48)* -95.53(43.11)*
0.79
Indirect effects of academic expectations
Expectations  English score
-69.36(25.26)* 692.822(245.22)* -43.81(23.81)+
+p < .10, *p < .05, **p < .01, ***p < .001; b= Path Coefficients (equivalent to regression weights); SE=Standard Errors
86
Table 4. Path coefficients of covariates in the English models
Within schools
Between schools
ExpectationEfficacy
Model 1w
Full sample
b(SE)
-23.61(9.57)*
Model 2w
Females
b(SE)
1.79(0.16)***
Model 3w
Males
b(SE)
-14.25(0.29)***
Model 1B
Full sample
b(SE)
-131.49(25.67)*
Model 2B
Females
b(SE)
6.57(1.47)*
Efficacy Expectation
6.29(0.41)***
-48.61(13.01)*
5.37(0.41)***
19.15(2.67)**
-195.09(28.54)**
Age  Efficacy
-23.69(9.92)*
1.79(0.41)*
-11.26(2.92)*
413.04(146.25)*
-2.41
-1.44
Age  Expectations
-0.36(-0.48)
-7.72(5.97)
-0.61(0.61)
34.91
-57.16
49.45(30.08)+
Commitment Efficacy
10.22(4.02)*
-0.12(0.12)
6.34(0.89)***
33.33(17.19)+
-3.28(1.08)*
0.77(0.40)+
Commitment  English score
0.13(0.04)*
0.19(0.06)*
0.14(0.06)*
0.19(0.24)
0.26(0.47)
0.77(0.40)+
Commitment  Expectation
-2.79(0.25)**
24.35(6.93)*
-2.22(0.27)***
-7.41(1.36)*
85.64
66.17(38.75)+
Household size  assets
0.03(0.004)**
0.03(0.01)**
0.02(0.007)*
0.03(0.32)
-0.02(0.03)
0.11(0.05)*
-0.36(0.36)
1.29(4.23)
0.38(0.47)
-32.12(8.93)*
234.75
-8.23(2.14)**
Attendance  English score
0.22(0.02)**
0.21(0.04)**
0.25(0.03)**
0.24(0.21)
0.26(0.22)
0.02(0.26)
Parental income  assets
0.05(0.01)*
0.03(0.01)*
0.06(0.02)*
-0.04(0.07)
-0.10(0.07)
0.06(0.06)
Planned effort  Efficacy
8.69(3.96)*
-0.67(0.16)*
3.62(1.30)*
282.28
0.49
317.88(150.54)*
Planned effort  English score
-0.05(0.05)
-0.02(0.07)
-0.09(0.08)
-0.65(1.39)
-0.28(1.41)
3.21
Planned effort  Expectation
-0.93(0.21)*
9.60(3.27)*
-1.05(0.27)**
-24.18(7.23)*
335.76(111.55)*
-18.09(7.29)*
Class ratio  English scores
--
--
--
-0.002(0.5)
0.01(0.06)
-0.001(0.06)
Relationships
Household size  Expectations
Model 3B
Males
b(SE)
-161.71(74.16)*
Teachers without postsecondary
---0.17(0.22)
-0.08(0.26)
certificate  English scores
+
p < .10, *p < .05, **p < .01, ***p < .001; b= Path Coefficients (equivalent to regression weights); SE=Standard Errors
87
0.34
Table 5. Path coefficients of independent variables and mediators in the math models
Within Schools
Between schools
Model 4w
Full sample
b(SE)
Model 5w
Females
b(SE)
Model 6w
Males
b(SE)
Model 4B
Full sample
b(SE)
Model 5B
Females
b(SE)
Model 6B
Males
b(SE)
Household assets  Math score
-0.30(0.39)
-0.37(0.56)
0.13(0.65)
0.54
12.99(30.41)
0.05(6.99)
Parents’ income Math score
0.15(0.19)
0.08(0.25)
0.29(0.32)
-2.89(3.01)
-3.93
-1.98(2.88)
Efficacy  Math score
0.09(0.03)*
-1.38(0.07)***
-5.18(0.13)***
-0.33(0.37)
-1.15(0.89)
23.25(6.51)*
Household assets  Efficacy
-0.36(0.28)
-0.24(0.41)
-1.60(0.64)*
-8.68(5.83)
-46.19(30.41)
12.68
Parents’ income  Efficacy
0.33(0.13)*
0.21(0.19)
0.29(0.33)
-7.86(4.77)
0.45
0.57(5.52)
Relationships
Direct effects of economic resources
Indirect effects of academic efficacy
Indirect effects of youth’s academic expectations
Expectations  Math score
0.14(0.02)**
0.12(0.02)**
0.13(0.03)*
0.18(0.27)
-0.88(1.07)
0.27(0.56)
Household assets  Expectations
2.39(0.65)*
2.08(0.97)*
5.46(2.32)*
-13.12
-15.57
65.80
Parents’ income  Expectations
0.15(0.31)
0.31(0.45)
1.83(1.23)
30.24(9.74)*
26.43
-75.21(40.87)+
+
p < .10, *p < .05, **p < .01, ***p < .001; b= Path Coefficients (equivalent to regression weights); SE=Standard Errors
88
Table 6. Path coefficients of covariates in the math models
Within schools
Between schools
ExpectationEfficacy
Model 4w
Full sample
b(SE)
0.38(0.02)***
Model 5w
Females
b(SE)
0.41(0.2)**
Model 6w
Males
b(SE)
0.92(0.05)***
Model 4B
Full sample
b(SE)
2.33(0.39)**
Model 5B
Females
b(SE)
3.96(1.07)
Model 6B
Males
b(SE)
3.99(0.91)*
Efficacy Expectation
-1.14(0.05)***
-1.38(0.07)**
-5.18(0.13)***
-7.03(0.86)**
-11.83(2.29)**
23.25(6.51)*
0.28(0.09)*
0.34(0.14)*
0.87(0.19)*
-7.65(3.48)*
-5.26(5.46)
-5.09(4.64)
Age  Expectations
-1.14(0.19)**
-1.38(0.32)*
-1.15(0.71)
-8.57
-4.94(9.42)
76.92(48.22)
Commitment Efficacy
0.38(0.02)***
0.35(0.04)**
0.29(0.33)
0.03(0.23)
-1.53(1.18)
-0.73(0.55)
Commitment  Math score
0.08(0.04)*
0.14(0.05)*
0.29(0.32)
0.36(0.20)
1.56(0.74)*
0.12(0.45)
Commitment  Expectation
0.94(0.06)***
0.98(0.09)**
1.86(1.23)
3.05(0.50)**
6.85(1.59)*
-9.61(2.76)*
Household size  assets
0.03(0.004)**
0.03(0.01)**
0.02(0.01)*
0.03(0.03)
-0.03(0.03)
0.11(0.05)*
-0.02(0.15)
0.29(0.22)
-0.81(0.54)
11.54(3.46)*
8.82
-69.99(47.79)
Attendance  Math score
0.27(0.02)**
0.31(0.04)**
0.29(0.03)**
0.55(0.20)*
0.58(0.21)*
0.31(0.28)
Parental income  assets
0.05(0.01)**
0.03(0.01)*
0.08(0.06)
-0.03(0.06)
-0.09(0.07)
0.08(0.07)
Planned effort  Efficacy
0.05(0.04)
0.001(0.05)
-0.07(0.09)
-2.55(1.25)+
1.89(3.17)
-6.06(2.64)*
Planned effort  Math score
0.07(0.05)
0.10(0.07)
0.04(0.08)
-0.83(1.14)
-2.64(1.79)
-0.70(1.94)
Planned effort  Expectation
0.66(0.08)**
0.71(0.12)**
1.54(0.31)**
10.57(2.75)*
12.28(4.08)*
-23.16(10.61)*
--
--
--
-0.06(0.05)
0.00(0.06)
-0.09(0.06)
Relationships
Age  Efficacy
Household size  Expectations
Class ratio  Math scores
Teachers without postsecondary
---0.39(0.22)+
0.21(0.82)
0.62(0.32)+
certificate  Math scores
+
p < .10, *p < .05, **p < .01, ***p < .001; b= Path Coefficients (equivalent to regression weights); SE=Standard Errors
89
Results of Indirect Relationships between Economic Resource and Academic Performance
To test the potential mediational role of academic self-efficacy and youth expectations, I
assessed four major components of the mediation model: overall model fit, direct pathway (Pathw
c in Figure 2), and indirect pathways (Pathw a and Pathw b in Figure 2). If results of the overall
model suggested the presence of statistically significant mediation (i.e., indirect pathways), the
results were compared to results of the direct pathway to establish complete or partial mediation.
Statistically significant direct and indirect relationships suggest partial mediation. Statistical
significance of the indirect relationship only suggested complete (i.e., full) mediation. When an
indirect relationship was confirmed, the magnitude of the indirect effect was calculated using the
product-of-coefficients strategy (Kenny, 2012; Preacher & Hayes, 2008) by multiplying the path
coefficient from the independent variable to the mediator (Pathw a) by the path from the mediator
to the dependent variable (Pathw b).
Results of Self-Efficacy Pathway
Mediating role of academic self-efficacy in English score outcomes. To assess academic
self-efficacy as a potential mediator, I examined the overall model for acceptable fit. The fit
indices of the overall mediation model showed good model fit (x2 = 112.82, df = 33, RMSEA =
.04). Results show that the direct path from academic self-efficacy to English scores is
statistically significant for the full sample (b = 49.81, p < .05), the female-only sample (b = 4.00, p < .01), and the male-only sample (b = 32.52, p < .05). Because of the statistically
significant relationship between self-efficacy and English scores, any significant association
between economic resources and self-efficacy would indicate the presence of a mediational
effect.
90
Results show that the direct path from household assets to academic self-efficacy is
statistically significant (i.e., household assets  academic self-efficacy: b = 49.81, p < .05),
which means that there is a statistically significant indirect pathway from household assets to
English scores via youths’ academic self-efficacy. The extent of the indirect relationship is 5.98
(i.e., household assets  academic self-efficacy  English scores: 49.81 × 0.25 = 5.98). The
results support the hypothesis that a young person’s academic self-efficacy mediates the
association between household assets and English scores (2b). The mediational effect is
complete (i.e., full mediation) because direct association between the independent variable
(household assets) and the outcome (English scores) is not statistically significant. Thus,
household assets are related to English scores but only through students’ self-efficacy. For every
one additional unit increase in household assets, English scores improve by 5.96 points through
the intervening effect of academic self-efficacy.
Next, I examined whether the mediational effect is consistent across the two gender
groups. First, there are statistically significant relationships between academic self-efficacy—the
mediator—and English scores for females (i.e., academic self-efficacy  English scores: b =
11.00, p < .05) and males (i.e., academic self-efficacy  English scores: b = 0.13, p < .05). With
these significant results, any significant relationship between economic resources and academic
self-efficacy across the gender groups would indicate an indirect relationship between economic
resources and English scores through the intervening effect of academic self-efficacy.
Results show that academic self-efficacy mediates the relationship between household
assets and English scores for females (i.e., household assets  academic self-efficacy: b = -4.00,
p < .01) and for males (i.e., household assets  academic self-efficacy: b = 32.52, p < .05), but
at different significance levels, magnitudes, and directions. Therefore, the data support the
91
hypothesis that gender moderates the indirect association between household assets and young
people’s English scores through academic self-efficacy (5b). The magnitude of the indirect
relationship for females is small and negative (i.e., -4.00 × 0.11 = -0.44) because there is an
inverse relationship between household assets and academic self-efficacy of females (b = -4.00,
p < .01). Conversely, the magnitude of the indirect relationship for males is positive and bigger
(i.e., household assets  academic self-efficacy  English scores: 32.52 × 0.13 = 4.23) because
the statistically significant path from household assets and academic self-efficacy is positive and
bigger for the male sample (b = 32.52, p < .05). The extent of the indirect relationship in the
male sample is closer to that of the full sample.
The study also assessed whether academic self-efficacy mediates the association between
parents’ income and English scores. Upon examining the paths for the full sample and the two
gender samples, none of the samples has a statistically significant path from parents’ income to
self-efficacy (full sample: b = -5.18, p > .10; female sample: b = 0.47, p > .10; male: b = 0.45, p
> .10). Evidence from the data does not support the hypothesis (2a) that self-efficacy mediates
the association between parents’ income and English performance. Likewise, because the path
from parents’ income to self-efficacy is not statistically significant for females and males, the
data do not support Hypotheses 5a, which posits that gender moderates the indirect association
between parents’ incomes and young people’s English scores through academic self-efficacy.
Mediating role of academic self-efficacy in math score outcomes. To assess the
potential mediational role of academic self-efficacy in the math models, I first reviewed the
overall model for adequate model fit, which was good (x2 = 91.58, df = 33, RMSEA = .03). To
establish a mediational effect, the direct path from academic self-efficacy—the potential
mediator—to math scores must be statistically significant for the full sample model. The results
92
show that academic self-efficacy statistically explains math scores for the full (b = 0.09, p < .05),
the female-only (b = -1.38, p < .001), and the male-only (b = -5.18, p < .001) samples. Because
the academic self-efficacy pathway is significantly associated with math scores, any significant
association between economic resources and self-efficacy would indicate the presence of a
mediational effect.
Results show that the direct path from household assets to self-efficacy is not statistically
significant (b = -0.36, p > .10), which means that there is no statistically significant indirect
pathway from household assets via youths’ academic self-efficacy to math scores for the full
sample. The results do not support the hypothesis that a young person’s academic self-efficacy
mediates the association between their household assets and math scores (2d). However, gender
moderates the mediational role of self-efficacy in the relationship between household assets and
math scores. That is, the indirect relationship between household assets and math scores via selfefficacy is significant for males (b = -1.60, p < .05) but not females (b = -0.24, p > .10). Thus,
household assets are related to math scores through students’ self-efficacy but only for males.
The extent of the indirect relationship is 8.29 (i.e., -1.60 × -5.18 = 8.29). The data supports the
hypothesis (5d) that gender moderates the mediational role of academic self-efficacy in the
relationship between household assets and math scores.
Results show that the direct paths from parents’ income to self-efficacy is significant for
the full sample (b = 0.33 p < .05), but not the female-only (b = 0.21, p > .10), and the male-only
samples (b = 0.29, p < .10). This means, for the full sample, there is a statistically significant
indirect pathway from parents’ income to math scores via youths’ academic self-efficacy. The
data supports the hypothesis (2c) that there is an indirect relationship between parents’ income
and math scores through academic self-efficacy. However, the data do not support the
93
hypotheses (5c) that gender moderates the indirect association between parents’ incomes and
young people’s math scores through academic self-efficacy.
Results of Academic Expectation Pathway
Mediating role of youths’ academic expectations in English score outcomes. Before
assessing whether an indirect relationship exists between economic resources and English scores
via youths’ expectations, the overall model for the full sample was assessed and was deemed a
good fit (x2 = 112.82, df = 33, RMSEA = .04). The female and male samples have sufficient
overall model fit (female: x2 = 66.63, df = 21, RMSEA = .04; male: x2 = 40.57, df = 10, RMSEA
= .05), suggesting there is support for the moderation hypotheses for the English model.
The path from academic expectations to English scores is statistically significant in the
full (b = 0.21, p < .001), female-only (b = 0.22, p < .001), and male-only (b = 0.21, p < .001)
samples. In all three samples, the paths from household assets to youth academic expectations
are not statistically significant (full sample: b = -0.58, p > .10; female-only sample: b = 16.61, p
> .10; male-only sample: b = -0.86, p > .10). Thus, there is no support for the hypothesis (3b)
that academic expectations mediate the association between household assets and young people’s
English scores. Likewise, because the potential mediational effect is not statistically significant
in the female and male samples, there is no evidence to support the hypothesis (6b) that gender
moderates the indirect association between household assets and young people’s English scores
through academic expectations.
The path from parents’ income to youths’ academic expectations for the English model is
negatively and statistically significant in the full sample (b = -1.98, p < .05), which means that
there is an indirect path from parents’ income to English scores through academic expectations.
Thus, the data supports Hypothesis 3a, which posits that academic expectations mediate the
94
association between parents’ income and young people’s English scores. The extent of the
indirect association is 3.91 (i.e., -1.98 × -1.93 = 3.91). Similarly, there is a statistically significant
direct path from parents’ income to academic expectations in the male-only sample (b = -1.93, p
< .10) but not in the female-only sample (b = 10.84, p > .10). The gender difference means there
is support for Hypothesis 6a that gender moderates the indirect association between parents’
income and young people’s English scores through academic expectations.
Mediating role of youths’ academic expectations in math score outcomes. There is
evidence of mediation by academic expectations in the full, female-only, and male-only samples.
First, I established that the pathway from young people’s academic expectations to math scores
is statistically significant for full (b = 0.14, p < .01), female-only (b = 0.12, p < .01), and maleonly (b = 0.13, p < .05) samples, which means there will be evidence of mediation if a
statistically significant path exists between any of the economic resources and the mediator (i.e.,
Path a). Results show that the path from parents’ income to academic expectations is not
statistically significant (b = 0.15, p > .10), which discounts the hypothesis (3c) that academic
expectations mediate the association between parents’ income and young people’s math scores.
Likewise, the path from parents’ income to academic expectations is not statistically significant
for the female-only (b = 0.31, p > .10) and male-only (b = 1.81, p > .10) samples, which means
there is no evidence to support the hypothesis (6c) that gender moderates the indirect association
between parents’ income and young people’s math scores through academic expectations.
However, there is a significant indirect relationship between household assets and math
scores. First, the path from household assets to academic expectations in the full sample is
statistically significant (b=2.39, p < .05), which supports the hypothesis (3d) that academic
expectations mediate the association between household assets and young people’s math scores.
95
The strength of the indirect relationship is 0.33 (i.e., 2.39 × 0.14 = 0.33). In addition, the path
from household assets to academic expectations in the female-only (b = 2.08, p < .05) and maleonly (b = 5.46, p < .05) samples are statistically significant. The magnitude of the indirect
relationship for males (i.e., 5.46 × 0.13 = 0.71) is nearly thrice that for females (i.e., 2.08 × 0.12
= 0.25). This finding supports that hypothesis (6d) gender moderates the indirect association
between household assets and young people’s math scores through academic expectations.
Conclusion
Overall, results show that there is no direct relationship between economic resources and
academic performance, at the 95% confidence level. However, at the same confidence level,
results show that both types of economic resources—parents’ income and household assets—are
associated with educational outcomes through different mediating mechanisms. While results
indicate that economic resources are associated with academic performance, it is through the
intervening variables of academic self-efficacy and academic expectations. The study also finds
evidence of gender differences. Nine (75%) of the 12 individual and household characteristics
vary by gender. Gender moderates all but of the statistically significant indirect relationships
between economic resources and academic performance. Where evidence of gender differences
exists, results consistently favor males over females.
96
CHAPTER 6: DISCUSSION
Summary of Findings
The purpose of this study is to examine the pathways through which economic resources
are associated with educational outcomes for Junior High School students in Ghana. Assetseffects is the main theoretical framework guiding the study. This framework posits that asset
ownership and the process of accumulating assets change young people’s mental representations
in ways that improve academic self-efficacy and future expectations, which in turn influence a
range of developmental outcomes (Scanlon & Adams, 2009; Sherraden, 1991). The theory of
assets effects also suggests that assets create educational advantages by facilitating access to
educational resources and services that improve educational outcomes (Elliott, 2013). Using this
theoretical framework, prior empirical evidence, and data from the cross-sectional baseline
survey from the YouthSave Ghana Experiment, I examined 24 hypotheses clustered into the
following three categories:
1. Direct relationships: Different types of economic resources have varying direct
effects on academic performance.
2. Mediated (indirect) relationships: Different types of economic resources affect
academic performance through young people’s academic self-efficacy and
expectations.
3. Moderated relationships: Direct and mediated relationships between economic
resources and academic performance differ by gender.
In all, the study finds support for eight hypotheses. None of the hypothesized direct relationships
are supported: Half (4 out of 8) of the hypothesized indirect relationships are supported. A third
97
(4 out of 12) of the hypothesized moderated relationships are supported. Table 7 summarizes the
list of hypotheses supported.
Table 7. Summary of hypotheses supported (8 out of 24 hypotheses supported)
Category of
hypotheses
Direct
relationships
Count of
supported
hypotheses
0 out of 4
Indirect
relationships
4 out of 8
Moderated
relationships
4 out of 12
List of supported hypotheses
None
1. Hypothesis 2b: Academic self-efficacy mediates the
association between household assets and young people’s
English scores.
2. Hypothesis 2c: Academic self-efficacy mediates the
association between parents’ incomes and young people’s
math scores.
3. Hypothesis 3a: Academic expectations mediate the
association between parents’ income and young people’s
English scores.
4. Hypothesis 3d: Academic expectations mediate the
association between household assets and young people’s
math scores.
1. Hypothesis 5b: Gender moderates the indirect association
between household assets and young people’s English
scores through academic self-efficacy.
2. Hypothesis 5d: Gender moderates the indirect association
between household assets and young people’s math scores
through academic self-efficacy.
3. Hypothesis 6a: Gender moderates the indirect association
between parents’ income and young people’s English
scores through academic expectations.
4. Hypothesis 6d: Gender moderates the indirect association
between household assets and young people’s math scores
through academic expectations.
Overall, findings reveal that the relationship between household economic resources—
parents’ income and household assets—and youth academic performance is more evident and
stronger when intervening variables—young people’s academic self-efficacy and expectations—
are taken into account. For the most part, these indirect relationships are positive, meaning the
98
more access young people have to economic resources, the higher their academic expectations
and sense of self-efficacy, and these in turn lead to better academic performance. The many
instances of no significant direct relationships between economic resources and academic
performance suggest that knowing the amount of economic resources owned by a youth’s
household is not enough to explain success in academic performance. Findings also suggest that
gender moderates all the significant indirect relationships between economic resources and
academic performance. Compared to school characteristics, more individual assets appear to
explain academic performance, which is discussed in this chapter.
Direct Relationship between Household Economic Resources and Academic Performance
Findings from this study suggest that there is no statistically significant direct
relationship between either type of the economic resources—parents’ income and household
assets—and academic performance. This finding is consistent with studies by Easton-Brooks and
Davis (2007), Williams Shanks (2007), and Phillips et al. (1998), as they also found no
significant relationship between assets and English scores. The difference between this nonsignificant finding and the significant findings from previous empirical studies in the US could
be a reflection of contextual and cultural differences. As discussed in Chapter 1, JHS education
in Ghana is constitutionally guaranteed and required for all school-age youth. Besides, the
government provides basic school supplies and free lunch in some deprived schools. Perhaps,
these policies and provisions have negated the potential effects that lack of household economic
resources could have on young people’s educational experience at the JHS level. In the absence
of these programs, household economic resources could make a difference in JHS students’
educational experience because only those who have economic resources would be able to afford
educational expenses.
99
Another reason for the contradictory results could be that the assets of most households in
the study are below a critical threshold to trigger an effect on academic performance. The worth
of household assets may not have been high enough to help pay for necessary school supplies.
Perhaps having household assets makes a significant difference in young people’s academic
performance once it reaches a certain threshold (Sherraden & McKernan, 2008). Empirical
evidence has yet to determine an asset threshold that would make a difference in educational
outcomes.
Besides, youth in the study did not personally own the household assets measured in the
study. Therefore they may not benefit from the assets-experience that comes with direct
ownership or management of assets. As discussed in Chapter 2, there is a learning component of
asset accumulation and ownership. When young people participate in accumulation and
management of assets, they gain practical experiences some of which may reinforce knowledge
acquired in the classroom and affect their academic performance. Perhaps, if these assets were
owned by youth in this study, they may have had positive asset-experiences that would directly
impact their academic work. On the other hand, we know certain types of assets lead to negative
asset-experience which may have negative influence on academic performance. Some youth in
the study may have assisted their households to manage assets, such as farmlands or business
ventures that are known to take students’ time away from academic work. Since this study did
not assess the nature of the asset-experience of youth, it is not possible to draw definitive
conclusions about the direction of the asset-experience. Nevertheless, there is a possibility that
some youth in the study spent more time on households assets compared to schoolwork, and that
may have neutralized the positive effects that household assets could have had on their academic
performance.
100
The finding that parents’ income is not associated with educational outcomes is not
atypical. Some studies find significant associations, but others fail to find support for such
relationships. Studies in SSA generally find significant relationships between parents’ income
and educational outcomes. For instance, studies from SSA countries such as Ghana, Côte
d’Ivoire (Acheampong, 2007; Tansel, 1997), Mozambique (Bilale, 2007; Handa, Simler, &
Harrower, 2004), Guinea (Glick & Sahn, 2000), and Ethiopia (Cockburn & Dostie, 2007)
suggest that higher parents’ or household income correlates with improved educational
outcomes. On the other hand, studies by Corwyn and Bradley (2002), Shea (2000), and Chevalier
et al. (2005) find non-significant relationships between parents’ or household income and
educational outcomes in the US. One possible reason for the difference in findings is that the
studies in SSA do not test income separately from other measures of socio-economic status. To
assess the true association between parents’ or household income and educational outcomes, this
study separates income from other measures of socio-economic status.
Methodological difficulties in measuring income in developing countries also could
account for the variations in findings. Many individuals and households in SSA earn seasonal
income (Prakongsai, 2006), and the data in some studies may be unreliable depending on when
they were collected (Moser & Felton, 2007). Moreover, a large proportion of households in the
developing world earn income from the informal sector, through self-employment, and by
performing a variety of economic activities. Thus, reliance on study participants’ memory recall
for income data is insufficient (Sahn & Stifel, 2003). Studies use different methods of data
collection that may contribute to different findings.
Aside from the possible methodological explanations for differences in findings, perhaps
parents’ income is not as important until students transition from JHS to SHS. As discussed in
101
Chapter 1, there is no tuition fee for JHS, but there is little government support at the SHS level.
Thus, it is possible that parents’ income would be more relevant at the SHS level because that is
when more economic resources are needed to pay for tuition and school supplies. Some students
are even excluded from class or taking SHS standardized exams for non-payment of tuition fees
(Amoah, April 2013; Daily Guide, May 2012; GNA, April 2013; Owusu, 2012). Such situations
are unlikely at the JHS level because the cost of supplies is comparatively lower, and the
government covers tuition and subsidizes examinations fees (See GNA, March 2009).
Given that this study finds no significant direct relationship, it is possible that the
slightest contextual difference could explain the varying results. Because of the regional
inequality in school infrastructure in Ghana (Higgins, 2009; Jebuni, McKay, & Shepherd, 2007),
academic performance may be more sensitive to small variations in household assets. For
instance, many pupils in some districts in Ghana carry chairs and tables to school daily (GNA,
2012). Lack of household durables (i.e., assets) such as furniture could negatively affect youths’
schooling. Studies in Ghana show that not having a school desk and chair is a major reason for
school dropout (Akyeampong et al., 2012; Ampiah & Kwaah, 2010; Rolleston, et al, 2010).
Mediated (Indirect) Relationships between Household Economic Resources and Academic
Performance
This study aimed to test the potential mediational roles of academic self-efficacy and
expectations in the relationship between economic resources and academic performance. Overall,
results indicate that economic resources are indirectly associated with educational outcomes
through academic self-efficacy or expectations. This finding provides empirical support for the
theory that having economic resources affects the way people think about their capabilities,
which may in turn change their behaviors and lead to positive developmental outcomes (Scanlon
102
& Adams, 2009; Sherraden 1991). The findings also corroborate empirical evidence that
suggests that economic resources influence the way people perceive the world, which affects
their behavior (Yadama & Sherraden, 1996; Zhan, 2006).
One surprising result from this study is that the most significant way by which economic
resources are associated with academic performance appears to be through the mediating
variables of academic self-efficacy and expectations. Having economic resources does not
directly explain much of the variability in academic performance, which is evident in the no
direct relationship between household assets and math scores. The question then is why
academic self-efficacy and expectations are so important to the connection between economic
resources and educational outcomes?
One explanation from the asset-effect framework is that economic resources help create
the necessary conditions for motivating young people to perform well in school. Researchers find
that having savings has various cognitive and behavioral effects on young people (Chowa, 2008;
Sherraden et al., 2005). Having economic resources may facilitate acquisition of the self-efficacy
needed to perform well in school. For instance, Ghanaian parents who have more economic
resources may be able to make significant investments in their children’s capabilities. A study by
Akaguri and Akyeampong (2010) shows that it costs households in the lowest income quintile
16% of their income to enroll a child in a public school, but it costs households in the highest
income quintile only 1.7% of their income. Households in the lowest income quintile must invest
a considerable proportion of their resources to develop their children’s capabilities.
Similarly, when parents invest considerable resources into their children’s education and
school supplies, students are likely to have higher expectations of their own academic work. This
study suggests that household resources are associated with academic performance through their
103
influence on academic expectations. This finding is consistent with evidence from other studies
that shows that asset ownership even by young people instills confidence and raises expectations
(Elliott, 2008; Elliott, Chowa, & Loke, 2011; Loke, 2009).
More research on assets and education are using psychological factors, such as selfefficacy and expectations, to help explain the relationship between assets and educational
outcomes (Elliott, 2011). In general, these studies have focused on parents’ expectations
(Williams Shanks & Destin, 2009), and relatively few have investigated the effects of young
people’s self-efficacy and expectations as linking mechanisms between economic resources and
educational outcomes (Hayes et al., 2012; James, 2010; Schunk, 2011; Sun, Buys, & Wang,
2012). Because the present study suggests that these mediating psychological factors provide the
most compelling explanations of the link between economic resources and educational outcomes,
it is important that they are factored into how we understand the importance of assets for
educational success (Zhan & Sherraden, 2009).
Gender as a Partial Moderator in the Relationship between Household Economic
Resources and Academic Performance
In addition to demonstrating indirect associations between economic resources and
educational outcomes, results from this study also suggest that the nature of these relationships
sometimes depends on the young person’s gender. First, results show that gender does not affect
either the direct or indirect relationships between parents’ income and math scores, which
perhaps shows that the gender gap in math performance is narrowing (Kane & Mertz, 2012;
Hyde & Mertz, 2009). Traditionally, boys outperform girls in math, but providing equal access to
opportunities through economic resources may help reduce this gender gap. A cross-country
104
study of 19 African countries shows that even when boys outperform girls on math tests, the
difference is small (Dickerson, McIntos, & Valente, 2013).
A weak but statistically significant gender difference does exist when it comes to English
scores, which is consistent with other empirical evidence from SSA (Abukari, 2010; Johnson,
Crosnoe, & Elder, 2001; Sahn, 2000). However, this study’s moderation results are different
from others’ because the inclusion of the mediating factors of academic expectations sometimes
changes the direction of the relationship in favor of males. For instance, there is no significant
gender difference in the direct relationship between parents’ income and English scores until
academic expectations act as a mediator. In this case, the relationship is significant and negative
for males but positive for females.
Historically, girls have not had as many educational opportunities as boys because of
Ghana’s dominant patriarchal culture (Abukari & Laser, 2013). A family with few resources
often would invest in boys’ schooling, so girls may become more sensitive to and take greater
advantage of educational opportunities. The finding that higher parental income favors girls is
consistent with a study conducted by Appleton (1995b) in Cote d’Ivoire, which shows that girls
outperform boys when households have more economic resources to begin with.
On the contrary, when academic self-efficacy mediates the association between
household assets and English scores, results show a significant negative relationship for females
but not males. The negative effect of household assets on females’ self-efficacy reduces the
positive effect of self-efficacy on educational outcomes, which results in an indirect negative
effect of household assets on educational outcomes. Clearly, this is an unexpected impact that
possibly supports the idea that the asset experience could have negative effects on outcomes,
especially if the asset accumulation process takes young people’s time away from schoolwork.
105
Overall, the mixed findings for the moderating role of gender suggest different economic
resources have varying impacts on girls’ and boys’ education. Specifically, household assets may
not always enhance girls’ education. As this study reveals, when the indirect association between
household assets and educational outcomes is statistically significant, the relationship is positive
for males but negative for females. Such evidence should raise concerns about the effectiveness
of asset building strategies (Scanlon & Adams, 2009). Because different assets are combined to
form an assets index in this study, it is not clear whether specific types of assets contribute to the
negative relationships for female youth. Clearly, this is a question that needs further
investigation to explain the dynamics.
Relationship between School-Level Variables and Academic Performance
Results show that the relationships between the number of teachers without
postsecondary qualifications and math and English scores are weak and not statistically
significant at the 95% confidence level. The weak and nonsignificant relationships are consistent
with studies in Ghana that suggest that teacher qualifications influence student learning the most
when other related factors such as conditions of service are favorable (Agyemang, 1993; Etsey,
2005). In other words, while academic and professional qualifications of teachers are necessary,
they are not sufficient to transform student learning and academic performance. Making sure all
schoolteachers, particularly those in rural areas of Ghana, are professionally trained is laudable
but inadequate to transform academic performance of students if teachers are not well-motivated
or do not have instructional resources to do teach effectively. For instance, private schools in
Ghana are mostly staffed with “untrained” teachers and yet they outperform public schools
staffed with predominantly trained teachers because other factors such as school management
106
systems, community involvement, and the lack of instructional materials are key factors in
explaining the performance differences (Akyeampong, 2003).
This study also finds that student-to-teacher ratio is not significantly associated with
English and math scores, which is counterintuitive and contradicts previous research in which
students in smaller classes outperform those in larger classes (Etsey, 2005). The importance of
student-to-teacher ratio is based on the idea that teachers are able to give direct and individual
attention when there are fewer students in the class. However, it seems there is a threshold effect,
and student-to-teacher ratio begins to have a negative effect after it exceeds a certain number. A
study by Kraft (1994) finds that class size in Ghana does not have a negative effect on academic
performance until there are more than 40 students, and studies by Hanushek (1996, 1999)
suggest that class size has no effect on academic performance. Findings from the present study
and others suggest that class size may not be detrimental to academic performance in every case.
Smaller teacher-to-student ratio may be beneficial at earlier grades when students need more
individualized support and attention from teachers (Lee et al., 2005; Greenwald & Hedges, 1996;
Nye & Hedges, 2000; Wößmann & West, 2006; Michaelowa & Wittmann, 2007).
Relationship between other Control Variables and Academic Performance
Numerous control variables are significantly associated with educational outcomes. First,
young people’s rate of school attendance has a statistically significant and positive relationship
with math and English scores, which means that youth who attend school regularly tend to do
better academically. This is consistent with a study by Norviewu-Mortty (2012) who finds that
regular school attendance in Ghana boosts subject grades in continuous assessment and final
examination scores at the JHS level. Other studies find strong relationships between school
attendance and other educational outcomes (Allensworth & Easton, 2007; Railsback, 2004).
107
When students attend school, they acquire more knowledge and skills by learning directly from
their teachers and peers. The Government of Ghana has adopted free tuition, free school
uniforms, and school feeding programs at the basic school level to encourage not only high
enrollment rates but also regular attendance.
The direction of relationships between some covariates and educational outcomes
contradicts earlier studies. For instance, age has a negatively and statistically significant
association with educational outcomes. On average, older students have lower math and English
scores than their younger peers, which is contrary to results from studies in Ghana and other
countries (Busari, 2012; Yousefi, 2010). However, the finding indicating that younger students
perform better than older ones in English score is consistent with the critical period hypothesis in
language acquisition studies which posits that children are better equipped in acquiring a second
language when they are younger than when older (Ampaabeng & Tan, 2012). It is possible that
as young people mature in age and begin to engage in work or other income-generating
activities, their interest in school wanes. The Ghana Statistical Service (2003) finds that 34.58%
of study participants aged 6–11 and 44.26% of participants aged 12–15 dropped out of school
because they felt it was uninteresting. This suggests that some young people lose interest in
school as they grow. There is also a question of why such older students still attend JHS? Could
it be that older students are repeating grades or have had interruptions in schooling for various
reasons? These needs more investigation to clarify the reasons for the inverse relationship
between age and academic performance at the JHS level in Ghana.
Another unexpected finding is that planned effort—the number of hours students spend
on school work after normal school hours—is negatively related to English scores, and is not
significantly associated with math scores. It is possible that there is noise in the data either
108
because of social desirability bias or inaccurate reporting by students. Nevertheless, research
suggests that although out-of-school time can generally benefit students academically, the quality
of study time may be more important (Beer & Beer, 1992; Gortner Lahmers & Zulauf, 2000).
For instance, a study by Plant, et. al. (2004) finds that the amount of study time only emerged as
a significant predictor of academic performance when the quality of study and other factors were
taken into consideration. Perhaps, parental involvement may be necessary to provide a favorable
learning environments at home and help structure how their children study during out-of-school
hours. Thus far, studies show that Ghanaian parents are not sufficiently engaged in their
children’s schooling (Nyarko, 2011; Pryor & Ampiah, 2003) and this may contribute to the
negative effects of planned effort on English performance.
This study also finds that commitment to school is positively associated with academic
performance, suggesting that youth’s commitment to their academic work may be fundamental
to their academic success. Jenkins (1995) finds that students’ committed to school are less likely
to engage in school crime, misconduct and nonattendance, all of which may affect academic
performance. These findings are consistent with prior research in the United States that have
shown positive association between student commitment and academic performance (Johnson,
Crosnoe, & Elder, 2001).
Challenges and Limitations
This study uses cross-sectional data, which limits ability to make causal inferences.
Ideally, a dataset with two or more time points would be used to trace the causal pathways
through which economic resources affect educational outcomes. Future data will be collected as
part of the YouthSave project to allow for testing of these relationships. In the meantime, the
109
present study is a test of correlational relationship between these concepts, an important first
step.
Participants for this study were sampled randomly from eight out of the 10 regions of
Ghana. Thus, it may not sufficiently represent youth in Ghana because of the exclusion of these
two regions. Likewise, results may not be applicable to other parts of the world. Replication of
this study with data from other regions would help confirm and build on emerging evidence
about the effects of assets on educational outcomes.
In addition, some of the self-reported measures of economic resources might include
unintentional measurement errors. For instance, given the cultural sensitivities surrounding
disclosure of income, this study cannot rule out under-reporting or over-reporting of income.
Similarly, the latent constructs for self-efficacy and parental involvement used in the analyses
may not tap into all possible dimensions of the constructs. If this is the case, it could have
implications for the validity of the measures, even though all constructs had high validity scores.
This is the first time the self-efficacy and parental involvement scales have been adapted to fit
the Ghanaian context. The lack of standardized scales for these constructs presents an
opportunity for future research to build on these adapted scales and eventually standardize them
for use in SSA and other developing countries.
Despite these limitations, this study has a number of methodological strengths that
expand the conceptualization and testing of direct and indirect effects of different types of
economic resources on young people’s educational outcomes. In addition to the large sample size
with adequate representation of males and females, the Multilevel SEM approach is one of the
most rigorous approaches for assessing mediation and hierarchical data (Bowen & Guo, 2011).
This unified framework accommodates multiple mediation pathways with nested data. This study
110
also provokes further investigation into specific types of household assets and their effects on
educational outcomes.
111
Chapter 7: Implications and Conclusion
Implications for Policy
This study provides revealing evidence on the potential relationships between economic
resources and educational outcomes in Ghana. The findings concerning possible gender
differences could inform policy makers on economic resources that may promote parallel
outcomes for boys and girls. Results from this study indicate that the indirect influence of
household economic resources on academic performance is different for boys and girls. The
evidence of gender differences should inform public discourse and policy makers about specific
types of economic empowerment and asset building strategies that promote equal developmental
outcomes for both genders. This could be an important boost for ongoing efforts to promote
gender equality in terms of not only access to resources but also equality of outcomes, such as
academic performance.
Findings from this study also suggest that income and assets do matter, though indirectly
through psychological factors such as academic self-efficacy and expectations. These
psychological factors are more predictive of youth academic performance at the JHS level than
economic resources. Thus, the Ghana Education Service and policy makers should consider
supplementing current education programs at the JHS level with policies and programs that
promote better psychological outcomes. Effective training of teachers and continued professional
development to build their capacity to enhance students' self-esteem will be a good way to
improve students’ psychological well-being. Because of shortage of high-caliber teachers and
inability to retrain trained teachers, many basic schools in rural areas rely on untrained teachers
(i.e., teachers who are not formally trained to teach) (Aboagye, 2008). Untrained teachers at the
JHS level will need upgrading to ensure that all JHS teachers in rural and urban areas are trained
112
to attain the minimum professional qualification. The Ministry of Education in Ghana has
initiated a four-year school-based in-service program to train and offer untrained teachers a
diploma in basic education. The curriculum for this program will have to include education of
psychological well-being to ensure that teachers are not only capable of teaching course content
but are also competent enough to understand classroom climate and offer support to students
struggling with low self-esteem and other psychological challenges. Similarly, more resources
should be invested into school psychologists, school social workers, and guidance counselors to
provide psychological therapies and services in school settings. Government should aim at
providing all schools with counseling units where both students and teachers would benefit from
behavioral consultation, individual and group counseling, or psycho-educational assessment and
receive appropriate and timely referrals.
While focusing on strategies to boost youth psychological outcomes, there should also be
a broad array of development strategies that promote accumulation of household economic
resources for future educational needs because of two reasons. In the first instance, economic
resources in the short term are important for improving psychological outcomes because this
study reveals that assets and income are associated with academic self-efficacy and expectations
in diverse ways. Secondly, household economic resources in the long term would be crucial
when students are transitioning from JHS to SHS. It may be that, because education is free at the
JHS level, household economic resources would become more important to young people when
they move from the JHS to the SHS level where government support is rather limited. The
Government of Ghana may have to broaden the current emphasis on income-based approaches to
include asset development strategies. This may require changes to existing programs and
realignment of government resources from some income-based programs to asset-based
113
programs. One particular social protection program in the Ghanaian context that can be revised
in the long-term is the Livelihood Empowerment against Poverty Program (LEAP). LEAP is a
social cash transfer program that provides income support for extremely poor households across
Ghana. The goal is to alleviate short-term poverty and encourage schooling and long-term human
capital development. This program can be redesigned or upgraded to incorporate asset-based
policies. The inclusion of such policies can help beneficiaries save some of the income as a way
of reducing vulnerability to shocks associated with high tuition fee during transition from JHS to
SHS level.
More importantly, policy makers should pay attention to evidence-based research in the
design of educational policies. It seems that the non-significant finding of direct asset-effect is
suggestive of the success of government policies and programs to equalize access and
opportunities at the JHS level. In other words, having economic resources may have significant
effect on academic performance when there are limited educational infrastructure and
government support. Prior evidence has demonstrated that, from 2005 to 2006, government
programs such as introduction of the capitation grant (i.e., elimination of school fees) lead to
67% increase in enrolment at the Kindergarten level, 9.7% increase at the primary level, and
10% at the JHS level, and it also narrowed the gender gap in enrollment (UNGEI, n.d.). With
these increases, one can safely assert that, prior to the elimination of school fees, household
economic resources were possibly having a significant effect on enrolment because only
households that had economic resources could afford to enroll their children in school. Perhaps,
similar government efforts at the JHS level such as provision of enough schools, infrastructure,
and elimination of fees could be replicated at the SHS level to ease access.
114
Empirical studies on accumulation of economic resource and their effects in SSA are
increasing; however more rigorous studies are needed to clarify asset effects and the policy
options discussed above. The GoG has to do more to support educational research in Ghana. The
Ghana Education Trust Fund (GETFund) was established by the GoG in 2001 to among other
things, provide financial support for research (Atuahene, 2013). Yet research in Ghana has not
received priority funding from the GETFund because of competing demands from infrastructure
development and other sectors of education. The GoG should do more to incentivize and
encourage research through grants and tax incentives. Such investments will generate ample
evidence on how economic development interventions and strategies might foster better
psychological and educational outcomes of school-going youth in Ghana. Other ways
government could support education research is by facilitating researchers’ access to data on
education such as the BECE results and national, regional, and district level data of SHS intake.
Currently, these types of data are hard to access from the Ghana Education Service, thus
hindering researchers’ ability to rigorously assess determinants of academic performance at the
JHS level as well as transition to SHS.
Implications for Future Research
In developed countries, emerging empirical evidence suggests that there is a positive
relationship between specific types of economic resources and educational outcomes. Findings
from the present study only suggest indirect relationships between economic resources and
educational outcomes in Ghana. This area of research however needs replication to confirm the
nature of the relationships. Because the current study and other empirical research in SSA have
had mixed findings, additional research needs to clarify the nature of the relationships between
different types of economic resources and educational outcomes among different genders, age
115
groups, educational levels, and geographical areas, especially the rural-urban divide. It could be
that because rural areas generally have limited educational infrastructure, it would take
significant amount of household economic resources to improve educational outcomes compared
to urban areas where better educational infrastructure exist. The reverse could be that students in
urban areas face transportation hurdles and other challenges which may affect their school
attendance and other educational outcomes. These are all possibilities that remain to be tested.
Similarly, this study finds that a significant proportion of the variation in academic
performance is explained at the school level, yet the school-level variables in the model (i.e.,
class size and teacher qualification) did not explain much of the variation. Perhaps, most of the
unexplained variation at the school-level would be accounted for if the study controls for more
school-level variables. Also, the large intra-class correlation coefficient (ICC) at the school level
raises the question of whether there is self-selection into schools. In other words, are households
with more economic resources attracted to good schools that tend to be in good neighborhoods,
and are households with limited economic resources more attracted to nonperforming schools
often concentrated in poor neighborhoods? Given the extent of homogeneity among public
schools in Ghana and the fact that probability sampling approach was employed in selecting the
sample for this study, the possibility of selection bias is slim. Nevertheless, it would be important
to conduct further investigations to ascertain whether there is evidence of selection bias,
regardless of the cluster randomized sampling strategy employed in selection of study
participants.
As stated in Chapter 6, the cross-sectional design of this study limits its ability to
establish true causal relationships between economic resources and educational outcomes. Future
research will have to use experimental and longitudinal research designs and datasets to test
116
these potential associations. By using baseline data from the Ghana YouthSave Experiment, the
present study is the first of several steps necessary to establish causality in the relationships
between economic resources and educational outcomes. The next step for the larger YouthSave
Experiment is to meet the temporal sequence criteria for causality because it is logically
necessary that economic resources (the cause) is measured before academic performance (the
effect) to be able to establish causality. Beyond that, establishing a mediated causal relationship
ideally would require a longitudinal design with at least three waves of data. Economic resources
(the cause) would have to be measured at wave 1, academic self-efficacy and expectations (the
mediators) at wave 2, and academic performance (the effect) at wave 3. With its rigorous
experimental design, the larger YouthSave Experiment in Ghana is well positioned to help
clarify these potential causal relationships. After post-test data are collected, the YouthSave
research team will be able to test potential causal relationships between specific types of
economic resources—such as youth savings—and educational outcomes.
One key issue that assets and education researchers aim to address is whether assets have
long-term developmental outcomes, including helping young people progress through different
educational stages. Certainly in Ghana, an important question that needs to be addressed is
whether assets contribute significantly to successful transition from JHS to SHS and then to
completion of SHS. Given the need to pay much higher costs at SHS level, it seems likely that
assets will have an effect, at least on access to school. This study’s findings about the indirect
effects of economic resources is an essential first step to a long-term empirical investigation of
ways that certain economic resources foster educational progression in Ghana. If other studies
confirm this relationship, the next logical step would be an investigation into the types and levels
of economic resources that may result in affordability of and successful transition to SHS. To
117
address these important questions about transition to higher education, studies would have to use
youth longitudinal datasets that cover different levels of educational systems, such as the JHS
and SHS levels in Ghana.
The use of appropriate research designs and youth datasets would help to empirically
clarify the potential relationships between economic resources and educational outcomes of
youth in Ghana and other SSA countries. Such clarifications have implication for the growing
interest in asset-based interventions in SSA. The present study proposes and examines potential
direct and indirect relationships between economic resources and educational outcomes in
Ghana. To advance this area of work, rigorous research will have to replicate this study in other
SSA countries. Also, the field may do well to move toward testing possible causal relationships
between specific types of economic resources and specific educational outcomes among different
youth groups. Different age groups may have different experiences with economic resources, the
effects of which may vary by the phases of a young person’s life.
Conclusion
A key insight from this study is that parents’ income and household assets have different
indirect relationships with academic performance of young people. However, the process that
connects these economic resources to academic performance is through the mediating variables
of young people’s academic self-efficacy and expectations. Findings also suggest that some of
these relationships may be moderated by gender. These findings have added to what is known
about the pathways through which parents’ income and household assets affect young people’s
academic performance. This study is one of the first empirical investigations into potential
connections between different kinds of economic resources and educational outcomes for JHS
students in Ghana. These findings could have implications for policy options to improve
118
educational progress and human capital development of young people in Ghana and other SSA
countries. The findings could also have implications on the larger YouthSave Ghana Experiment
in Ghana. Knowing the nature of the relationship between certain types of economic resources
and academic performance at baseline will allow researchers to understand the dynamics of
young people’s own savings and assets better and how they may impact on academic
performance in the long term. Questions such as, what impact could existing economic resources
have on the level of influence of youth savings accounts on academic performance may be
answered when we control for prior economic resources (i.e., parents’ income and household
assets) at endpoint.
119
References
Aaronson, D. (2000). A note on the benefits of homeownership. Journal of Urban Economics, 47
(3), 356-369.
Aboagye, J. K. (2008). Teacher Education in Ghana: Challenges and Prospects. In PRINCOF.
Developments in Basic Teacher Education in Ghana (pp. 141-154). Kumasi: Greenland’s
Concept.
Abraha, S., Beyene, A., Dubale, T., Fuller, B., Holloway, S., & King, E. (1991). What factors
shape girls' school performance? Evidence from Ethiopia. International Journal of
Educational Development, 11, 107-18.
Abukari, Z. (2010). Risk and protective factors associated with academic achievement among
Ghanaian youth. PhD. Dissertation, University of Denver.
Abukari, Z., & Laser, J. A. (2013). Gender differences in academic outcomes among Ghanaian
youth: The role of protective and risk factors. Journal of Community Psychology, 41(1),
117–138.
Acemoglu, D., & Angrist, J. (1999). How large are the social returns to education? Evidence
from compulsory schooling laws (NBER Working Paper No. 7444). The National Bureau
of Economic Research.
Acemoglu, D., & Pischke, S. (2001). Changes in the Wage Structure, Family Income and
Children’s Education. European Economic Review, 45, 890–904.
Admassie, A. (2002). The implication of asset ownership on child work in rural Ethiopia.
Ethiopian Journal of Economics, 11(2), 25-58.
Agbewode, S. (2009, October 13). Ghana: Education Ministry cautions school heads. The
Chronicle. Retrieved from http://allafrica.com/stories/200910160380.html
120
Agyemang, D. K. (1993). Sociology of education for African students. Accra: Black Mask Ltd.
Ajayi, K. F. (2012). School choice and educational mobility: Lessons from secondary school
applications in Ghana. Boston University. Retrieved from
http://people.bu.edu/kajayi/KAjayi_EducationalMobility_Oct2012.pdf
Ajayi, K. O. (2012). The influences of self-concept and academic motivation on students’
attitude to mathematics in selected secondary schools in Ogun State, Nigeria. European
Journal of Scientific Research, 67(3), 444-455.
Akaguri, L. & Akyeampong, K. (2010). Public and private schooling in rural Ghana: Are the
poor being served? Brighton, United Kingdom, Consortium for Research on Education
Access, Transitions and Equity, University of Sussex.
Akhtar, Z., & Aziz, S. (2011). The effect of peer and parent pressure on the academic
achievement of university students. Language in India, 11, 254-265.
Akresh, R., Bagby, E., de Walque, D., & Kazianga, H. (2012a). Child Ability and Household
Human Capital Investment Decisions in Burkina Faso. Economic Development and
Cultural Change, 61(1), 157-186.
Akyeampong, K. (2003). Teacher training in Ghana – Does it count? Multi-Site Teacher
Education Research Project (Country Report One). Centre for International Education,
University of Sussex Institute of Education, Sussex. Retrieved from
http://r4d.dfid.gov.uk/PDF/Outputs/SkillsForDev/Educationalpaper49b.pdf
Akyeampong, K. (2009). Access and Educational Equity in Ghana. Brighton: Centre for
International Education, University of Sussex
Akyeampong, K. (2011). (Re)Assessing the impact of school capitation grants on education
access in Ghana (Research Monograph No. 71). Consortium for Research on Educational
121
Access, Transitions and Equity. University of Sussex Centre for International Education.
Retrieved from http://www.create-rpc.org/pdf_documents/PTA71.pdf
Akyeampong, K., Rolleston, C., Ampiah, J. G., & Lewin, K. M. (2012). Access, Transitions and
Equity in Education in Ghana: Researching Practice, Problems and Policy (Research
Monograph No. 72). The Consortium for Educational Access, Transitions and Equity
(CREATE). http://www.create-rpc.org/pdf_documents/PTA72.pdf
Alexander, K. L., Entwisle, D. R., & Bedinger, S. D. (1994). When expectations work: Race and
socioeconomic differences in school performance. Social Psychology Quarterly, 57, 283–
299.
Alhassan, E. (2010). Socio-economic and cultural determinants of girl-child education in
Gushegu/Karaga district of the northern region of Ghana. Ghana Journal of Development
Studies, 7(1), 50-72.
Allensworth, E., & Easton, J. Q. (2007). What matters for staying on-track and graduating in
Chicago Public High Schools: A close look at course grades, failures, and attendance in
the freshman year. Chicago: Consortium on Chicago School Research.
Amankrah, J. Y. (2007). Youth unemployment in Ghana: Prospects and challenges. Ghana
statistical Service Report on Job Tracking Survey. Ghana Statistical Service
Amoah, A. K. (2013, April 25). Sacking students from exams halls for fees. The Chronicle.
Retrieved from http://thechronicle.com.gh/?p=44683
Ampaabeng, S. K., & Tan, C. M. (2012). The long-term cognitive consequences of early
childhood malnutrition: The case of famine in Ghana. Retrieved from
https://editorialexpress.com/cgibin/conference/download.cgi?db_name=CSAE2013&paper_id=12
122
Ampiah, J., G., & Kwaah, C. (2010). CREATE Ghana ComSS Report Mfantesman District.
University of Cape Coast, Ghana.
Amuah, I. R. (1996, December). Portfolio assessment: New trends in the evaluation of Ghanaian
Junior Secondary School student's musical achievement. The African Music Educator, 8,
14-20.
Ansong, D., Chowa, G., & Grinstein-Weiss, M. (2013). Future orientation as a mediator between
assets and perceived household economic stability: An SEM approach. Social Work
Research.
Appleton, S. (1995b). The interaction between poverty and gender in human capital
accumulation: The case of the primary leaving exam in the Côte d'Ivoire. Journal of
African Economies, 4, 192-224.
Apt, N. A. (2005). A study of child domestic work and fosterage in Northern and Upper East
Regions of Ghana. UNICEF. Retrieved from
http://www.thewayforwardproject.org/file_uploads/G02%20A%20Study%20of%20Child
%20Domestic%20Work%20and%20Fosterage.pdf
Apt, N. A., Blavo, E., & Opoku, S. (1992). Street children in Accra. Accra. Social
Administration Unit, University of Ghana, Legon.
Atta-Quayson, J. (2007, February 21). Education: A total failure? Daily Graphic (Ghana), pp.
16, 33-34.
Atuahene, F. (2013). Charting Higher Education Development in Ghana: Growth,
Transformations, and Challenges, in Alexander W. Wiseman, Charl C. Wolhuter (ed.)
The Development of Higher Education in Africa: Prospects and Challenges (International
123
Perspectives on Education and Society, Vol. 21), Emerald Group Publishing Limited,
pp.215-263
Baffour-Awuah, P. (2011). Supervision of instruction in public primary schools in Ghana:
Teachers’ and headteachers’ perspectives. [Dissertation] School of Education, Murdoch
University.
Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Englewood
Cliffs, NJ: Prentice- Hall, Inc.
Bandura, A. (1997). Self-efficacy: The exercise of control. New York: Freeman
Bandura, A., Barbaranelli, C., Gian, V. C., & Pastoreili, C. (1996). Multifaceted impact of selfefficacy beliefs on academic functioning. Child Development, 67, 1206-1222.
Barro, R. (2006). Education and development: Quality counts. The World Bank. Retrieved from:
http:/www.worldbank.org/education/pdf/Education
Baum, S., & Payea, K. (2005). Education pays 2004: The benefits of higher education for
individual and society. Trends in Higher Education Series. The College Board. Retrieved
from
http://www.collegeboard.com/prod_downloads/press/cost04/EducationPays2004.pdf.
Bean, K. F., & Sidora-Arcoleo, K. (2012). The relationship between environment, efficacy
beliefs, and academic achievement of low-income African American Children in special
education. Journal of Social Work in Disability & Rehabilitation, 11, 268-286.
Beer, J., & Beer, J. (1992). Classroom and home study times and grades while at college using a
singlesubject design. Psychological Reports, 71, 233–234.
Beaulieu, L. J., & Mulkey, D. (1995). Investing in people: The human capital needs of rural
America. Boulder, CO: Westview Press.
Becker, G. S. (1996). Accounting for Tastes. Cambridge, Mass.: Harvard University Press
124
Beegle, K., Dehejia, R. H., & Gatti, R. (2003). Child labor, income shocks, and access to credit.
World Bank Policy Research Working Paper 3075. http://wwwwds.worldbank.org/servlet/WDSContentServer/WDSP/IB/2004/01/30/000265513_20040
130123109/Rendered/PDF/wps3075.pdf
Belley, P. & Lochner, L. (2007). The changing role of family income and ability in determining
educational achievement. Journal of Human Capital, 1(1), 37-89.
Bhalotra, S., & Heady, C. (2003). Child farm labor: The wealth paradox. World Bank Economic
Review, 17, 197−227.
Bilale, F. J. C. (2007). Educational performance in Mozambique: An economic perspective.
Department of Economics, University of Stellenbosch
Blanden, J., & Gregg, P. (2004). Family income and educational attainment: A review of
approaches and evidence for Britain (CMPO Working Paper Series No. 04/101). The
Centre for Economic Performance, London School of Economics.
http://www.bristol.ac.uk/cmpo/publications/papers/2004/wp101.pdf
Bong, M. (2001). Role of self-efficacy and task-value in predicting college students’ course
performance and future orientation intentions. Contemporary Educational Psychology,
26(4), 553-570.
Boshara, R. (2009, Springs). Combating poverty by building assets: Lessons from around the
world. Pathways, 19-23.
Bouman, F. J. (1995). Rotating and accumulating savings and credit associations: A development
perspective. World Development, 23(5), 371-384.
Bowen, N. K., & Guo, S. (2011). Structural Equation Modeling. New York, NY: Oxford.
University Press.
125
Brooks-Gunn, J., Klebanov, P., & Duncan, G. (1996). Ethnic differences in children‘s
intelligence test scores: Role of economic deprivation, home environment, and maternal
characteristics. Child Development, 67, 396 – 408.
Buchmann, C. (2000). Family structure, parental perceptions, and child labor in Kenya: What
factors determine who is enrolled in school? Social Forces, 78(4), 1349-1378.
Buhs, E.S. (2005). Peer rejection, negative peer treatment, and school adjustment: Self-concept
and classroom engagement as mediating processes. Journal of School Psychology,
43(5): 407-424.
Busari, A. O. (2012). Evaluating the relationship between gender, age, depression and academic
performance among adolescents. Scholarly Journal of Education, 1(1), 6-12.
Bynner, J., & Paxton, W. (2001). The asset-effect. London: Institute for Public Policy Research.
Caillaud, F., Experton, W., Fevre, C., Foko, B., Gioan, P. A., Johnstone, B., et al. (2009).
Financing tertiary education in Africa. Montreal: UNESCO.
Chevalier, A., Harmon, C., O’Sullivan, V., & Walker, I. (2005). The impact of parental income
and education on the schooling of their children (WP05/05). The Institute for Fiscal
Studies. Retrieved from http://discovery.ucl.ac.uk/2726/1/2726.pdf
Chowa, G. (2008). The impacts of an asset-building intervention on rural households in SubSaharan Africa. Ph.D. dissertation, Washington University in St. Louis, United States -Missouri. Retrieved December 15, 2010, from Dissertations & Theses @ Washington
University in St. Louis (Publication No. AAT 3332075).
Chowa, G. A. N., Ansong, D., Masa, R., Despard, M., Osei-Akoto, I., Richmond, A.-A., AgyeiHolmes, A., & Sherraden, M. (2012). Youth and saving in Ghana: A baseline report from
126
the YouthSave Ghana Experiment. St. Louis, MO: Washington University, Center for
Social Development.
Chowa, G., Ansong, D., & Masa, R. (2010). Assets and child well-being in developing countries:
A research review. Children and Youth Services Review, 32, 1509-1519.
doi:10.1016/j.childyouth.2010.03.015
Chowa, G., Masa, R., Wretman, C., & Ansong, D. (2013). The Impact of Household Possessions
on Youth's Academic Achievement in the Ghana YouthSave Experiment: A Propensity
Score Analysis. Economics of Education Review, 33, 69-81.
Christy-McMullin, K., Shobe, M. A., & Wills, J. (2009). Arkansas IDA programs: Examining
asset retention and perceptions of well-being. Journal of Social Service Research, 35(1),
65-76. doi:10.1080/01488370802477485
Cisternas, G. (2011). A continuous-time model of career concerns and human capital
accumulation. Department of Economics, Princeton University. Retrieved from
http://www.princeton.edu/econtheorycenter/wps/wp018_2011_Cisternas.pdf
Cockburn, J., & Dostie, B. (2007). Child work and schooling: The role of household asset
profiles and poverty in rural Ethiopia. Journal of African Economies, 16(4), 519-563.
College Board (2005). Education pays: A supplement to Education Pays 2004: The benefits of
higher education for individuals and society. Trends in Higher Education Series.
Retrieved from
http://www.collegeboard.com/prod_downloads/press/cost05/education_pays_05.pdf
Corwyn, R. F., & Bradley, R. F. (2002). Family process mediators of the relation between SES
and child outcomes. Unpublished manuscript, University of Arkansas at Little Rock.
127
Centre for Social Policy Studies (CSPS). (2001). Child domestic workers in Accra. Accra: Centre
for Social Policy Studies, University of Ghana, Legon.
Cudeck, R. (1989). Analysis of correlation matrices using covariance structure models.
Psychological Bulletin, 105, 317-327.
Curley, J., Ssewamala, F. M., & Han, C-K. (2010). Assets and educational outcomes: Child
development accounts (CDAs) for orphaned children in Uganda. Children and Youth
Services Review, 32(11), 1585-1590.
Dahl, G. B., & Lochner, L. (2012). The impact of family income on child achievement: evidence
from the earned income tax credit. American Economic Review, 102(5), 1927-1956
Daily Guide (2012, May 10). Students thrown out of exam hall over nonpayment of fees. Daily
Guide. Retrieved from http://www.dailyguideghana.com/?p=47526
DANIDA, EFW, & The World Bank. (2011). The joint review of public expenditure and
financial management. Retrieved from
http://www.mofep.gov.gh/sites/default/files/reports/Review_of_Public_Expenditure_101
1.pdf
Davis-Kean, P. (2005). The Influence of Parent Education and Family Income on Child
Achievement: The Indirect Role of Parental Expectations and the Home Environment.
Journal of Family Psychology, 19 (2), 294-304
Day, J. C., & Newburger, E. C. (2002). The big payoff: Educational attainment and synthetic
estimates of work-life earnings. US Census Bureau. Retrieved from
http://www.census.gov/prod/2002pubs/p23-210.pdf
Department of Education and Skills (2011). Literacy and numeracy for learning and life: The
National Strategy to Improve Literacy and Numeracy among children and young people
128
2011-2020. Dublin, Ireland: Department of Education and Skills. Retrieved from
http://www.education.ie/en/Publications/Policy-Reports/lit_num_strategy_full.pdf
Dery, R. G., & Addy-Lamptey, W. (2010). Effects of classroom assessment scores on the final
scores used in grading students at senior high school in Ghana. Paper presented at the
36th International Association for Educational Assessment (IAEA) Annual Conference
held in Bangkok Thailand, from Sunday 22 to Friday 27 August 2010. Retrieved from
http://selectscore.com/fullpaper/127.pdf
Deshpande, R., Zimmerman, J. M., & (Eds). (2010). Youth savings in developing countries:
Trends in practice, gaps in knowledge. A report of the YouthSave Consortium.
Washington, D.C.: YouthSave Consortium.
Destin, M., & Oyserman, D. (2009). From assets to school outcomes: How finances shape
children’s perceived possibilities and intentions. Research Report. Psychological Science
Dib, C. Z. (1988). Formal, non-formal and informal education: Concepts/applicability. Published
in “Cooperative Networks in Physics Education - Conference Proceedings 173”,
American Institute of Physics, New York, 1988, pgs. 300-315. Retrieved from
http://www.techne-dib.com.br/downloads/6.pdf
Dickerson, A., McIntos, S., & Valente, C. (2013). Do the maths: An analysis of the gender gap
in mathematics in Africa (IZA DP No 7174). Bonn, Germany: The Institute for the Study
of Labor. Retrieved from http://ftp.iza.org/dp7174.pdf .
DiStefano, C., Zhu, M., & Mindrila, D. (2009). Understanding and using factor scores:
Considerations for the applied researcher. Practical Assessment, Research & Evaluation.
14(20). Available online: http://pareonline.net/getvn.asp?v=14&n=20
129
Donkoh, S. A., & Amikuzuno, J. A. (2011). The determinants of household education
expenditure in Ghana. Educational Research and Reviews, 6(8), 570-579
Dunne, M. & Leach, F. (2005). Gendered school experiences: The impact on retention and
achievement in Botswana and Ghana. London, UK: Department for International
Development.
Easton-Brooks, D., & Davis, A. (2007). Wealth, traditional socio-economic indicators, and the
achievement debt. Journal of Negro Education, 76(4), 530-541.
Education Encyclopedia (n.d.). Ghana - History Background, Constitutional Legal Foundations,
Educational System—overview, Preprimary Primary Education, Secondary Education.
State University. Retrieved from http://education.stateuniversity.com/pages/529/GhanaHISTORY-BACKGROUND.html#ixzz2RjXnVZyG
Egbo, B. (2000). Gender literacy and life chances in Sub-Saharan Africa. Toronto: Multicultural
Matters
El Badaoui, E., & Rebiere, T. (2012). Informal labor market and access to education in
developing economies. Retrieved from http://crem.univrennes1.fr/Documents/Docs_sem_eco_appliquee/2012-2013/13-01-10_Rebiere.pdf
Elliott, W., Kim, K., Jung, H., & Zhan, M. (2010). Asset holding and educational attainment
among African American youth. Children and Youth Services Review, 32,1497-1507.
Elliott, W. (2008). Closing the gap in college enrollment: The potential of children’s college
accounts (CSD Working Papers No. 08-16).
Elliott, W. (2011). When effort and ability are not enough to reduce the college enrollment gap,
does college savings help (CSD Working Paper No. 11-31). St. Louis, MO: Washington
130
University, Center for Social Development. Retrieved from
http://csd.wustl.edu/Publications/Documents/WP11-31.pdf
Elliott, W. (2013). Small-dollar children’s savings accounts and college outcomes (CSD
Working Paper 13-05). St. Louis, MO: Washington University, Center for Social
Development. Retrieved from http://csd.wustl.edu/Publications/Documents/WP13-05.pdf
Elliott, W. (2013). The effects of economic instability on children's educational outcomes.
Children and Youth Services Review, 35, 461-471.
Elliott, W., & Beverly, S. (2010). The role of savings and wealth in reducing “wilt” between
expectations and college attendance (CSD Working Paper No. 10-01). St. Louis, MO:
Washington University, Center for Social Development.
Elliott, W., Choi, E. H., Destin, M., & Kim, K. (2011). The age old question, which comes first?
A simultaneous test of children's savings and children's college-bound identity. Children
& Youth Services Review, 33(7), 1101-1111.
Elliott, W., Chowa, G., & Loke, V. (2011). Toward a children’s savings and college-bound
identity intervention for raising college attendance rates: A multilevel propensity score
analysis. Sociology Mind, 1(4). doi: 10.4236/sm.2011.14025.
Engelman, A., & Hazel, C. (2010). Partnering for youth empowerment in urban middle schools:
An autoethnography examining the synergies and tensions between positive youth
development and youth participatory action research. Information for Action, 21(1), 1-26.
Retrieved from http://www.servicelearningpartnership.org/site/DocServer/PartneringforYouthEmpowerment.Vol2No1.pdf?
docID=3948
131
Essen, J., Fogelman, K., & Head, J. (1978). Childhood housing experiences and school
attainment. Child: Care, Health and Development, 41, 41-58.
Etsey, K. (2005).Causes of low academic performance of primary school pupils in the Shama
Sub-Metro of Shama Ahanta East Metropolitan Assembly (SAEMA) in Ghana. Paper
presented at the Regional Conference on Education in West Africa Dakar, Senegal 1st –
2nd November, 2005.
Etsey, K., Smith, T. M., Gyamera, E. Koka, J., de Boer, J., Havi, E., Heyneman, S. P. (2009).
Basic Education in Ghana: Progress and Problems. U.S. Agency for International
Development: Washington DC.
Fantuzzo, J., McWayne, C., Perry, M., & Childs, S. (2004). Multiple dimensions of family
involvement and their relations to behavioral and learning competencies for urban, lowincome children. The School Psychology Review, 33(4), 467-480.
Fatusi, A. O., & Hindin, M. J. (2010). Adolescents and youth in developing countries: Health
and development issues in context. Journal of Adolescents, 33(4), 499-508.
Fentiman, A., Hall, A., & Bundy, D. (1999). School enrolment patterns in rural Ghana: A
comparative study of the impact of location, gender, and age and health on children’s
access to basic schooling. Comparative Education, 35(3), 331-349.
Filmer, D., & Pritchett, L. (1999). The Effect of Household Wealth on Educational Attainment:
Evidence from 35 Countries. Population and Development Review, 25(1), 85-120.
Filmer, D., & Pritchett, L. (2001). Estimating wealth effects without expenditure data —or tears:
An application to educational enrollment in states of India. Demography, 38(1), 115−132.
Filmer, D., & Pritchett, L. (1999). The effect ofhouseholdwealth on educational attainment:
Evidence from 35 countries. Population and Development Review, 25(1), 85-120.
132
Filmer, D., & Scott, K. (2008). Assessing Asset Indices (World Bank Policy Research Working
Paper 4605). World Bank: Washington D.C.
Fischhoff, B. (1992). Risk taking: A developmental perspective. In J.F. Yates (Ed.), Risk-Taking
Behavior, (pp. 133-162). Chichester, England: John Wiley & Sons
Fischhoff, B., Crowell, N.A., & Kipke, M. (1999). Adolescent Decision Making: Implications for
prevention programs. Summary of a Workshop. Washington, DC: National Academy
Press. (ERIC Document Reproduction Service No. ED441185)
Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention, and behavior: An introduction to
theory and research. Reading, MA: Addison-Wesley.
Foster, A., & Rosenzweig, M. R. (2009). Missing women, the marriage market, and economic
growth. Brown University Working Paper.
Garcia, M., & More, C.M.T. (2012). The cash dividend: The rise of cash transfer programs in
Sub-Saharan Africa. Washington, D.C.: The World Bank.
Gesthuizen, M., Meer, T. van de, & Scheepers, P. (2008). Education and dimensions of social
capital: do educational effects differ due to educational expansion and social security
expenditure? European Sociological Review, 24(5), 617-632.
Ghana Education Service (GES) (2004). The development of education national report of Ghana.
Presented at the 47th session of the International Conference on Education (ICE), Geneva
June 2004. Retrieved from
http://www.ibe.unesco.org/International/ICE47/English/Natreps/reports/ghana.pdf
Ghana Health and Education Initiative (n.d.). Senior high school scholarship. Retrieved from
http://ghei.org/programs/education-programs/senior-high-school-scholarship
133
Ghana News Agency (2009, March 16). Abolish terminal and promotional exams fees. Retrieved
from http://www.modernghana.com/news/206639/1/abolish-terminal-and-promotionalexams-fees-coalit.html
Ghana News Agency (2012, March 14).Ghana commits chunk of National Budget to Education
Sector. Retrieved from
http://www.ghanaweb.com/GhanaHomePage/regional/artikel.php?ID=232680
Ghana News Agency (2013, April 10). Headmasters urged not to sack students owning fees from
writing WASSCE. Retrieved from http://ghananewsagency.org/education/headmastersurged-not-to-sack-students-owing-fees-from-writing-wassce-58694
Ghana News Agency, (2010, July 7). Lack of facilities, cause of poor academic performance.
Modern Ghana News. Retrieved from http://www.modernghana.com/pdf/mg_news.pdf
Ghana Scholarship Fund (n.d.). Education in Ghana. Retrieved from
http://www.ghanascholarship.com/id17.html
Ghana Statistical Service (2003). Child labour survey. Accra: Ghana Statistical Service.
Ghana Statistical Service (2009). Street Children. Retrieved from
http://www.statsghana.gov.gh/surveys/GCLS2001/survey0/dataSet/dataFiles/dataFile3.ht
ml#V372
Ghana Statistical Service (2012). 2010 Population and housing census: Summary report of final
results. Accra, Ghana: Ghana Statistical Service.
Ghana Statistical Service (n.d.). Ghana Living Standards Survey Round Four (GLSS 4) 1998/99.
Data user’s guide. Accra, Ghana: Ghana Statistical Service. Retrieved from
http://siteresources.worldbank.org/INTLSMS/Resources/33589861181743055198/3877319-1190217341170/G4USERSG.pdf
134
Gilman, R. (1984). The idea of owning land: An old notion forged by the sword is quietly
undergoing a profound transformation. Retrieved from
http://www.context.org/iclib/ic08/gilman1/
Gist, M. E. & Mitchell, T. R.(1992). Selfefficacy: A theoretical analysis of its determinants and
malleability. Academy of Management Review, 17(2), 183-211.
Glaser, D. N. (1999). The controversy of significance testing: misconceptions Misconceptions
and alternatives. American Journal of Critical Care, 8(5), 291-296. Retrieved from
http://ajcc.aacnjournals.org.
Glick, P., & Sahn, D. E. (2000). Schooling of girls and boys in a West African country: The
effects of parental education, income, and household structure. Economics of Education
Review, 19(1), 63–87.
Global Partnership for Education & Ghana Development Partner Group (2012). Appraisal of
Ghana’s Education Sector Plan 2010-2020. Retrieved from
http://www.globalpartnership.org/media/GHANA%20Final%20ESP%20Appraisal%20R
eport%20-%20March%202012.pdf
GNA (2012, May 15). Schooling furniture at mercy of the weather at Ho. Ghana News Agency.
Retrieved from
http://www.ghanaweb.com/GhanaHomePage/NewsArchive/artikel.php?ID=239003
Gortner Lahmers, A., & Zulauf, C. R. (2000). Factors associated with academic time use and
academic performance of college students: A recursive approach. Journal of College
Student Development, 41, 544–556.
Grabowski, L. S., Call, K. T., & Mortimer, J. (2001). Global and economic self-efficacy in the
educational attainment process. Social Psychology Quarterly, 64(2), 164-179.
135
Graham, C. K. (1971). The history of Education in Ghana: From the eeliest times to the
declaration of independence. Accra, Ghana: Ghana Publishing Corporation.
Graham, C. K. (1976). The history of Education in Ghana. Accra, Ghana: Ghana Publishing
Corporation.
Green, E. (2013). Explaining African ethnic diversity. International Political Science Review,
34(3). 235-253.
Green, R. K., & White, M. J. (1997). Measuring the benefits of homeowning: Effects on
children. Journal of Urban Economics, 41(3), 441-461.
Greenwald, R., Hedges, L.V., & Laine, R.D. (1996). The effect of school resources on student
achievement. Review of Educational Research, 66, 361-396.
Handa, S., Simler, K., & Harrower, S. (2004). Human capital, household welfare, and children’s
schooling in Mozambique. Research Report. Internatinal Food Policy Research Institute.
Washington DC.
Hanushek, E. (1996). School resources and student performance. In G. Burtless (Ed.), Does
money matter? The effect of school resources on student achievement and adult success
(pp.43-73). Washington, D.C., Brookings institution,
Hanushek, E. (1999). The evidence on class size. In S.E. Mayer & P.E. Peterson (Eds.), Earning
and learning: How schools matter (pp. 131-168). New York: Russell Sage Foundation.
Hau, K.-T., & Marsh, H. W. (2004). The use of item parcels in structural equation modeling:
Non-normal data and small sample sizes. British Journal of Mathematical and Statistical
Psychology, 57, 327-351.
Haveman, R., & Wolfe, B. (1994). Succeeding generations: On the effects of investment in
children. New York: Russell Sage Foundation.
136
Hayami, Y. (2009). Social capital, human capital and the community mechanism: Toward a
conceptual framework for economists. Journal of Development Studies, 45(1), 96-123.
Hayes, D., Huey, E. L., Hull, D. M., & Saxon, T. F. (2012). The influence of youth assets on the
career decision self-efficacy in unattached Jamaican youth. Journal of Career
Development, 39, 407-442.
Heady, C. (2000). What is the effect of child labor on learning achievement? Evidence from
Ghana (Innocenti Working Paper 79). Florence, Italy: UNICEF Innocenti Research
Centre.
Heintz, J. (2005). Employment, poverty and gender in Ghana (Working paper series No. 92).
Political Economy Research Institute, University of Massachusetts Amherst. Retrieved
from
http://scholarworks.umass.edu/cgi/viewcontent.cgi?article=1076&context=peri_workingp
apers
Hideg, É., & Nováky, E. (2002). The future orientation of Hungarian youth in the years of the
transformation, In J. Gidley & S. Inayatullah (eds), Youth futures: Comparative research
and transformative visions (pp. 123- 129). Westport: Praeger.
Higgins, K. (2009). Regional inequality and secondary education in Ghana (ODI Policy Brief
No. 1). Overseas Development Institute. Retrieved from
http://www.odi.org.uk/sites/odi.org.uk/files/odi-assets/publications-opinionfiles/3375.pdf
Hilger, N. G. (2012). How does family income affect college enrollment? Evidence from timing
of parental layoffs. Harvard University. Retrieved from
http://www.people.fas.harvard.edu/~nhilger/JMP_Nate.pdf
137
Hofler, M. (2005). Causal inference based on counterfactuals. BMC Medical Research
Methodology, 5(28), 1-12. doi:10.1186/1471-2288-5-28.
Hong, E. (1998). Differential stability of state and trait self-regulation in academic performance.
The Journal of Educational Research, 91(3), 148-159.
Horstmanshof, L., & Zimitat, C. (2007). Fture time orientation predicts academic engagement
among first-year university students. British Journal of Educational Psychology, 77, 703718.
Howie, S. J. (2005). Contextual factors at the school and classroom level related to pupils’
performance in Mathematics in South Africa. Educational Research and Evaluation,
11(2), 123-140.
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis:
Conventional criteria versus new alternatives. Structural Equation Model, 6, 1-55. doi:
10.1080/10705519909540118.
Huang, J., Guo, B., Kim, Y., & Sherraden, M. (2010). Parental assets, income, and borrowing
constraints and children’s post-secondary education. Children and Youth Services
Review, 32(4), 585-594.
Huebler, F. (2008, October). International education statistics. Retrieved from
http://huebler.blogspot.com/2008/10/ptr.html
Hutchinson, E. (2008). Dimensions of human behavior in the social environment. Thousand
Oaks, CA: Sage Publications, Inc.
Hyde, J. S., & Mertz, J. E. (2009). Gender, culture, and mathematics performance. Proceedings
of the National Academy of Sciences (PNAS), 106(22), 8801-8807.
138
Iyer, R.V., Kochenderfer-Ladd, B., Eisenberg, N., & Thompson, M. (2010). Peer Victimization
and Effortful Control: Relations to School Engagement and Academic Achievement.
Merrill-Palmer Quarterly, 56(3): 361–387
Jacobo, J. (n.d.). The UNESCO Youth Development Index. UNESCO. Retrieved from
http://www.ccivs.org/New-SiteCCSVI/institutions/jpc-youth/youth-openforum/5.Monitoring/UNESCO/YDI-UNESCO-Brazil1.pdf
Jagsi, R., Griffith, K. A., Stewart, A., Sambucco, D., DeCastro, R., & Ubel, P. A. (2012). Gender
differences in the salaries of physician researchers. The Journal of the American Medical
Association, 307(22), 2410-2417. doi:10.1001/jama.2012.6183.
James, M. (2010). Gender, self-efficacy and achievement among South African Technology
teacher trainees. Gender & Education, 22(1), p87-103
Jebuni, C., McKay, A., & Shepherd, A. (2007). Economic Growth in Northern Ghana. A paper
prepared for the international workshop, Understanding and addressing spatial poverty
traps: an international conference’, 29 March 2007, Spier Estate, Stellenbosch, South
Africa.
Jegede, J.O., Jegede, R.T., & Ugodulunwa, C.A. (1997). Effects of achievement motivation and
study habits on Nigerian secondary school students’ academic performance. The Journal
of Psychology: Interdisciplinary and Applied, 131(5), 523-529.
doi.org/10.1080/00223989709603541.
Jensen, P. & Nielsen, H. S. (1997). Child labour or school attendance? Evidence from Zambia.
Journal of Population Economics, 10(4), 407-424.
139
Jensen, R. T. (2010). Economic opportunities and gender difference in human capital:
Experimental evidence for India (NBER Working Paper No. 16021). The National
Bureau of Economic Research.
Johnson, M. K., Crosnoe, R., & Elder, G. H. (2001). Students' attachment and engagement: The
role of race and ethnicity. Sociology of Education, 74,318-340.
Johnson, S., Wolke, D., Hennessy, E., & Marlow, N. (2011). Educational outcomes in extremely
preterm children: neuropsychological correlates and predictors of attainment.
Development Neuropsychology, 36(1), 74-95.
Jutte, D. P., Roos, N. P., Brownell, M. D., Briggs, G., MacWilliam, L., & Roos, L. L. (2010).
The ripples of adolescent motherhood: Social, Educational and medical outcomes for
children of teen and prior teen mothers. Academic Pediatric, 10(5), 293-301.
Kane, J. M. & Mertz, J. E. (2012). Debunking myths about gender and mathematics
performance. Notice of the AMS, 59(1), 10-21.
Keeley, B. (2007). Human capital: How what you know shapes your life. Paris: OECD.
Kenny, D. A. (2012, April 3). Mediation. Retrieved from http://davidakenny.net/cm/mediate.htm
Keteku, N. W. (1999). Educational reform in Ghana: the senior secondary school. Retrieved
from http://www.bibl.u-szeged.hu/oseas_adsec/ghana.htm
Kiamanesh, A.R. (2005). Evaluating students’ achievements: The role of students’ characteristics
and family background in Iranian students’ mathematics achievement. Prospects, 35(2),
161-174. doi.org/10.1007/s11125-005-1819-1.
Kim, Y. (2010). Impacts of parental resources on child educational outcomes: Assets and
mediating pathways (Doctoral Dissertations). St. Louis, MO: Washington University.
140
Kimuyu, P. K. (1999). Rotating saving and credit associations in rural East Africa. World
Development , 27(7), 1299-1308.
Kisselburgh, L. G., Berkelaar, B., & Buzzanell, P. M. (2009). Discourses, gender, and the
meaning of work: Rearticulating science, technology, and engineering careers through
communicative lenses. In C. S. Beck (Ed), Communication Yearbook 33(pp. 259-299).
New York: Routledge.
Kline, R. B. (2005). Principles and practice of structural equation modeling (2nd ed.). New
York: Guilford Press.
Kolavalli, S., Robinson, E., Diao, X., Alpuerto, V., Folledo, R., Slavova, M., Ngeleza, G., &
Asante, F. (2012). Economic transformation in Ghana: Where will the path lead (IFPRI
Discussion Paper 01161)? International Food Policy Research Institute. Retrieved from
http://www.ifpri.org/sites/default/files/publications/ifpridp01161.pdf
Kumar, R. (2011). Research methodology: A step-by-step guide for beginners. Los Angeles, CA:
Sage Publications Inc.
Kwon, D. (2009). Human capital and its measurement. The 3rd OECD World Forum on
“Statistics, Knowledge and Policy” Charting Progress, Building Visions, Improving Life
Busan, Korea 27-30 Oct. 2009. Retrieved from
http://www.oecd.org/site/progresskorea/44111355.pdf
Lance, C. E., Butts, M. M., & Michels, L. C. (2009). The sources of four commonly reported
cutoff criteria: What did they really say? Organizational Research Methods, 9(2), 202220.
141
Law, B., & Hall, C. (2009). Observational learning use and self-efficacy beliefs in adult sport
novices. Psychology of Sport and Exercise, 10(2), 263-270.
doi:10.1016/j.psychsport.2008.08.003
Ledermann, T., & Macho, S. (2009). Mediation in dyadic data at the level of the dyads: A
structural equation modeling approach. Journal of Family Psychology, 23(5), 661-670.
doi: 10.1037/a0016197
Lee, V., Zuze, T., & Ross, K. (2005). School Effectiveness in 14 Sub-Saharan African Countries:
Links With 6th Graders’ Reading Achievement. Studies in Educational Evaluation, 31,
(2-3), 207-246.
Lerman, R. I., & McKernan, S.-M. (2008). Benefits and consequences of holding assets. In S.-M.
McKernan, & M. Sherraden, Asset building and low-income families (pp. 175-206).
Washington, D.C.: The Urban Institute Press.
Lewin, K. M. (2008). Strategies for sustainable financing of secondary education in SubSaharan Africa (Working Paper No 136). Washington, D.C.: World Bank .
Lincove, J.A. (2009). Determinants of schooling for boys and girls in Nigeria under a policy of
free primary education. Economics of Education Review, 28(4), 474-484.
Loke, V. T. (2009). Asset trajectories and child outcomes: Implications for asset-based policies.
(Doctoral Dissertations). St. Louis, MO: Washington University.
Loke, V., & Sacco, P. (2011). Changes in parental assets and children’s educational outcomes.
Journal of Social Policy, 40(2), 351-368.
Loken, K. V. (2010). Family income and children’s education: Using the Norwegian oil boom as
a natural experiment. Labour Economics, 17(1), 118-129.
142
Lovenheim, M. F. (2011). The effect of liquid housing wealth on college enrollment. Cornell
University. Retrieved from
http://bfi.uchicago.edu/humcap/wp/papers/HousingWealthandCollegeEnrollment_JOLEF
inal.pdf
Lucas, K. (2009). No footsteps to follow: how blue-collar kids navigate postindustrial careers.
Unpublished dissertation, Purdue University, West Lafayette, IN.
MacKinnon, D. P., Krull, J.L., & Lockwood, C.M. (2000).Equivalence of the mediation,
confounding and suppression effect. Prevention Science, 1,173-181.
Maliyamkono, T. L., & Ogbu, O. (1999). Introduction. In T. L. Maliyamkono, & O. Ogbu, Cost
sharing in education and health (pp. 9-19). Dar es Salaam: Tema Publishing Company
Ltd.
Marks, G. (2006). Family size, family type and student achievement: Cross-national differences
and the role of socioeconomic and school factors. Journal of Comparative Family
Studies, 37(1), 1-27.
Mayer, S. (2010). Revisiting an old question: How much does parental income affect child
outcomes? Focus, 27(2), 21-26.
Mbebeb, F. E. (2009). Developing productive lifeskills in children: Priming entrepreneurial
mindsets through socialisation in family occupations. International Journal of Early
Childhood, 41(2), 23-34.
McDonnell, S., Crook, D., & Chellman, C. (2012). Residential stability and academic
performance: City University of New York students living in public housing. 2012 APPA
Fall Research Conference November 8-10, 2012, Baltimore, MD.
143
Mensch, B. & Lloyd, C. (1998). Gender differences in the schooling experiences of adolescents
in low-income countries: The case of Kenya. Studies in Family Planning, 29(2), 167-184.
Mereku, K. D. (2003). The BECE grading system committee report: Implications for minimum
educational qualifications for basic education certificate. Mathematics Connection, 3, 2533.
Michaelowa, K. (2001). Primary education quality in francophone Sub-Saharan Africa:
Determinants of learning achievement and efficiency considerations. World
Development, 29, 1699-716.
Michaelowa, K., & Wittmann, E. (2007). The cost, satisfaction, and achievement of primary
education: Evidence from Francophone Sub-Saharan Africa. World Development, 29(10),
1699-1716. doi.org/ 10.1016/S0305-750X(01)00061-4
Miller, K. (2003). Conducting cognitive interviews to understand question‐response limitations
among poorer and less educated respondents. American Journal of Health Behavior,
27(S3), 264‐272
Mincer, J., & Polachek, S. (1974). Family investments in human capital: Earnings of women. In
T. W. Schultz (Ed) Marriage, family, human capital, and fertility (p.76-110). The
National Bureau of Economic Research. retrieved from
http://www.nber.org/chapters/c3685.pdf
Ministry of Education, Science, and Sports (2007). Preliminary Education Sector Performance
Report 2007. Retrieved from
http://moe.gov.gh/docs/Education%20Sector%20PerformanceReport%202007/Preliminar
y%20Education%20Sector%20Annual%20Performance%20Rport%202007.pdf
144
Ministry of Youth and Sports (2010). National Youth Policy of Ghana. Retrieved from
http://www.ghana.gov.gh/documents/nypolicy.pdf
Miranda, H., & Russell, M. (2011). Predictors of teacher-directed student use of technology in
elementary classrooms: A multilevel SEM approach using data from the USEIT study.
Journal of Research on Technology in Education, 43(4), 301-323
Montgomery, M., & Hewett, P. (2005). Poverty and children’s schooling in urban and rural
Senegal (Policy Research Division Working Paper No.196). New York: Population
Council.
Montgomery, M., Grant, M., Mensch, B., & Roushdy, R. (2005). Children's schooling in
developing-country slums: A comparison of Egypt and India. Stony Brook, NY: Stony
Brook University Economics Departmental Working Paper No. 05-07.
Moretti, E. (2004). Workers' education, spillovers, and productivity: Evidence from plant-level
production functions. American Economic Review, 94(3), 656-690.
Moser, C., & Felton, A. (2007). The construction of an asset index measuring asset accumulation
in Ecuador (CPRC Working Paper 87). The Brookings Institute. Retrieved from
http://cprc.abrc.co.uk/pdfs/87Moser_Felton.pdf
Moyi, P. (2007). Child labor, poverty and schooling in Ghana and Kenya: A comparative
analysis. Amherst College. Retrieved from
http://paa2007.princeton.edu/download.aspx?submissionId=71929
Muller, J., & Roberts, J. (2000). The sound and fury of international school reform: A critical
review. Johannesburg — Report prepared for the Joint Education Trust, JET
Muris, P. (2001). A brief questionnaire for measuring self-efficacy in youths. Journal of
Psychopathology and Behavioral Assessment, 23(3), 145-149.
145
Nam, Y., & Huang, J. (2009). Equal opportunity for all? Parental economic resources and
children's educational attainment. Children and Youth Services Review, 31(6), 625-634.
Nam, Y., Huang, J., & Sherraden, M. (2008). Asset definitions. In S.-M. Mckernan, & M.
Sherraden, Asset building and low income families (pp. 1-31). Washington, D.C.: The
Urban Institute Press.
Nasser, F., & Hagtvet, K. A. (2006). Multilevel analysis of the effects of student and
instructor/course characteristics on student ratings. Research in Higher Education, 47(5),
559-590.
National Center for Education Statistics, (n.d.). Trends in international mathematics and science
study (TIMSS). Retrieved from http://nces.ed.gov/timss/table11_3.asp
National Research Council (U.S.) (2005). Growing up global: The changing transitions to
adulthood in developing countries (C. B. Lloyd, Ed). Washington, D.C.: National
Academies Press.
Ng, T. W., & Feldman, D. C. (2010). Human capital and objective indicators of career success:
The mediating effects of cognitive ability and consientiousness. Journal of Occupational
and Organisational Psychology, 82, 207-235.
NIH News (2012, August 28). Stresses of poverty may impair learning ability in young children.
National Institute of Health. Retrieved from
http://www.nih.gov/news/health/aug2012/nichd-28.htm
Norviewu-Mortty, E. K. (2012). Principals’ strategies for improving the academic achievement
of students of disadvantaged rural Junior High Schools in Ghana. (Doctoral
Dissertations). Perth, Western Australia: Edith Cowan University. Retrieved from
http://ro.ecu.edu.au/cgi/viewcontent.cgi?article=1494&context=theses
146
Nureev, R. (2010). Human capital and its development in present-day Russia. Russian Education
and Society, 52(3), 3-29.
Nyagura, L. M., & Riddell, A. (1993). Primary school achievement in English and mathematics
in Zimbabwe: A multi-level analysis. Washington DC: World Bank
Nyarko, K., & Vorgelegt, V. (2007). Parental involvement: A sine qua non in adolescents’
educational achievement. Unpublished doctoral dissertation, Ludwig Maximilians
University. Retrieved from http://edoc.ub.uni-muenchen.de/8384/1/Nyarko_Kingsley.pdf
Nyarko, K. (2011). Parental school involvement: The case of Ghana. Journal of Emerging
Trends in Educational Research and Policy Studies, 2(5), 378-381.
Nye, B., & Hedges, L.V. (2000). The effects of small classes on academic achievement: The
results of the Tennessee class size experiment. American Educational Research Journal,
37, 123-151.
Oduro, A. D. (2000). Basic education in Ghana in the post-reform period. Centre for Policy
Analysis. Retrieved from
http://www.cepa.org.gh/researchpapers/BASIC%20EDUCATION%20IN%20GHANA%
20IN%20THE%20POST1.pdf
Oduro, A. D. (2008). An investigation into poverty, educational attainment and outcomes in
Ghana. Ghana Working Paper (Meta Analysis) Series. Research Consortium on
Outcomes to Education and Poverty (RECOUP). Retrieved from
http://recoup.educ.cam.ac.uk/publications/OduroInvPovtyEduclOutcomesGhana1012081.
pdf
OECD and World Bank (2007). Reviews of National Policies for Education Reviews of National
Policies for Education: Higher Education in Kazakhstan. OECD and The World Bank.
147
Okyerefo, M. P. K., Fiaveh, D. Y. & Lamptey, S. N. L. (2011). Factors prompting pupils’
academic performance in privately owned Junior High Schools in Accra, Ghana.
International Journal of Sociology and Anthropology, 3(8), 280-289.
Olaniyan, D. A., & Okemakinde, T. (2008). Human capital theory: Implications for educational
development. European Journal of Scientific Research, 24(2), 157-162.
Oliver, M. L., & Shapiro, T. M. (1997). Black wealth/white wealth: A new perspective on racial
inequality. New York: Routledge.
Oliver, M., L., & Shapiro, T. M. (2006). Black Wealth/White Wealth, second edition. New York:
Routledge.
Orr, A. J. (2003). Black-White differences in achievement: The importance of wealth. Sociology
of Education, 36, 281-304
Osafo-Acquah, A., & Asamoah-Gyimah, K. (2009). Impact of Quality Improvement in Primary
Schools (QUIPS) programme on the academic performance of pupils in new Edubiase
Methodist primary school in the Adansi East District of Ghana. Edo Journal of
Counselling, 2(2), 142-154.
Osei R.D., Owusu, G.A., Asem F.E. & Afutu-Kotey, R.L. (2009). Effects of Capitation on
education outcomes in Ghana (GDN Working Paper No. 20). New Delhi: Global
Development Network.
Owusu, M. (2012, May 3). Headmaster bans 7 students from writing exams. Daily Guide.
Retrieved from http://www.dailyguideghana.com/?p=46866
Oyserman, D., Bybee, D., & Terry, K. (2006). Possible selves and academic outcomes: How and when
possible selves. Journal of Personality and Social Psychology, 91(1), 188-204.
148
Page-Adams, D. & Sherraden, M. (1996). What we know about effects of asset holding:
Implications for research on asset-based anti-poverty initiatives (Working Paper no. 961). St. Louis: Center for Social Development, Washington University.
Pajares, F., & Schunk, D. (2001). Self-beliefs and school success: Self-efficacy, self-concept,
and school achievement. In R. J. Riding & S. G. Rayner (Eds.), International
perspectives on individual differences (vol. 2): Self-perception, (pp. 239-265). Westport,
CT: Ablex.
Pasquali, V., & Aridas, T. (2013). The world’s richest and poorest countries. Global Finance.
Retrieved from http://www.gfmag.com/component/content/article/119-economicdata/12529-the-worlds-richest-and-poorest-countries.html#axzz2iYnayHS5
Phillips, M., Brooks-Gunn, J., Duncan, G. J., Klebanov, P., & Crane, J. (1998). Family
background, parenting practices, and the black-white test score gap. In C. Jencks & M.
Phillips (Eds.), The black-white test score gap (pp. 273-317). Washington, DC:
Brookings Institution Press.
Preacher, K. J., & Hayes, A. E. (2008). Asymptotic and resampling strategies for assessing and
comparing indirect effects in multiple mediator models. Behavior Research Methods,
40(3), 879-891.
Pryor, J., & Ampiah, J. G. (2003a). Understanding of Education in an African Village: The role
of information and communication technologies. London, Department for International
Development.
Public Agenda (2011, 31 October). Ghana: Senior high school fees go up. Retrieved from
http://allafrica.com/stories/201111010466.html
149
Railsback, J. (2004). Increasing student attendance: Strategies from research and practice.
Portland, Oregon: Northwest Regional Educational Laboratory.
Randolph, K., Fraser, M. & Orthner, D. (2006). A strategy for assessing the impact of timevarying family risk factors on high school dropout. Journal of Family Issues, 27, 933-950.
Regambal, M. J., & Alden, L. E. (2009). Pathways to intrusive memories in a trauma analogue
paradigm: A structural equation model. Depression and Anxiety, 26, 155-166.
doi:10.1002/da.20483
Reynolds, A. J., & Temple, J. A. (1998). Extended early childhood intervention and school
achievement: Age 13 findings from the Chicago Longitudinal Study. Child Development,
69, 231-246.
Reynolds, A.J., Temple, J.A., & Ou, S. (2010). Preschool education, educational attainment, and
crime prevention: Contributions of cognitive and non-cognitive skills. Children and
Youth Services Review, 32(8):1054-1063.
Rolleston, C., Akyeampong, K., Ampiah, J. G., & Lewin, K. M. (2010). Educational access in
Ghana. Country research summary. The Consortium for Educational Access, Transitions
and Equity (CREATE). Retrieved from http://www.createrpc.org/pdf_documents/Ghana_Country_Research_Summary.pdf
Rootman, I. (2001). Evaluation in health promotion: Principles and perspectives. WHO
Regional Publications European Series, No. 92). World Health Organization Regional
Office, Europe.
Rose, P. & Al-Samarrai, S. (1997). Household constraints on schooling by gender: Empirical
evidence from Ethiopia (IDS Working Paper 56). Institute of Development Studies.
150
Ross, P. D., & Roberts, P. (1999). Income and child well-being: A new perspective on the
poverty debate. Ottawa: Canadian Council on Social Development. Retrieved from
http://www.ccsd.ca/pubs/inckids/1.htm
Rutter, D. R., & Bunce, D. J. (1989). The theory of reasoned action of Fishbein and Ajzen: A
test of Towriss's amended procedure for measuring beliefs. British Journal of Social
Psychology, 28(1), 39–46.
Sackey, H. A. (2008). Private returns to education in Ghana: Implications for investments in
schooling and migration (AERC Research Paper 174). African Economic Research
Consortium, Nairobi.
Scanlon, E. & Adams, D. (2009). Do assets affect well-being? Perceptions of youth in a matched
savings program. Journal of Social Service Research, 35, 33-46.
Schneider, M., Dasappa, P., Khan, N., & Khan, A. (2009). Measuring disability in censuses: The
case of South Africa. ALTER, European Journal of Disability Research, 3, 245 – 265
Schultz, P. T. (2003). Human capital, schooling and health returns (Center Discussion Paper No.
853). Economic Growth Center, Yale University. Retreived from
http://ageconsearch.umn.edu/bitstream/28475/1/dp030853.pdf
Schumacker, R. E., & Lomax, R. G. (1996). A beginner’s guide to structural equation modeling.
Mahwah, NJ: Erlbaum.
Schunk, D. H. (1995). Self-efficacy and education and instruction. In J. E.Maddux (Ed.), Selfefficacy, adaptation, and adjustment: Theory, research, and applications (pp. 281-303).
New York: Plenum
Schunk, D. H., & Pajares, F. (2002). The development of self-efficacy. In A. Wigfield, & J.
Eccles, Development of achievement motivation. San Diego: Academic Press.
151
Scientific Software International, Inc. (2013). LISREL.9.10 Student Edition.
Seginer, R. (2009). Future orientation: Developmental and ecological perspectives. New York:
Springer.
Shanahan, M. J., Mortimer, J. T., & Kruger, H. (2002). Adolescence and adult work in the
twenty-first century. Journal of Research on Adolescence, 12(1), 99-120.
Shapiro, T. & Wolff, E. N. (2005) Assets for the poor. The benefits of spreading asset
ownership. New York: Russell Sage Foundation.
Shea, J. (2000). Does parents' money matter? Journal of Public Economics, 77(2), 155-184.
Sherraden, M. (1991). Assets and the poor: A new American welfare policy. Armonk, N.Y.:
M.E. Sharpe.
Sherraden, M., & McKernan, S. (2008). Directions for research. In S. McKernan, & M.
Sherraden, Asset building and low-income families (pp. 207-220).Washington, D.C.: The
Urban Institute Press
Sherraden, M.S., Johnson, L., Elliott, W. III, Porterfield, S., & Rainford, W. (2007). Schoolbased children's saving accounts for college: The I Can Save program. Children and
Youth Services Review, 29, 294-312.
Sherraden, M.S., McBride, A., Johnson, E., Hanson, S., Ssewamala, F., & Shanks, T. (2005).
Saving in low-income households: Evidence from interviews with participants in the
American Dream Demonstration. Washington University, Center for Social
Development.
Shobe, M., & Page-Adams, D. (2001). Assets, future orientation, and well-being: Exploring and
Extending Sherraden's Framework. Journal of Sociology and Social Welfare, 28 (3), 109129.
152
Shrout, P.E. & Bolger, N. (2002). Mediation in experimental and nonexperimental studies: New
procedures and recommendations. Psychological Methods, 7(4), 422-445.
Siziya, S., Muula, A. S., & Rudatsikira, E. (2007). Prevalence and correlates of truancy among
adolescents in Swaziland: Findings from the Global School-Based Health Survey. Child
Adolescent Psychiatry Metal Health, 1(15), 1-8.
Smith, M., & Barrett, A. M. (2010). Social and economic effects on primary pupils’ reading
achievement: Findings from southern and east Africa (EdQual Working Paper No. 23).
University of Bristol, UK. Retrieved from
Sparr, P., & Moser, C. (2007). International NGOs and Poverty Reducation Strategies: The
contribution of an asset-based approach (Global Economic and Development Working
Paper #08). Washington, D.C.: The Brookings Institution.
Spector, P. E., Cooper, C. L., Sanchez, J. I., O’Driscoll, M., Sparks, K., & Bernin, P., et al.
(2002). Locus of control and well-being at work: How generalizable are Western
findings? Academy of Management Journal, 45, 453-470.
Ssewamala, F. M., Alicea, S., Bannon, W., & Ismayilova, L. (2008). A novel economic
intervention to reduce HIV risks among school-going AIDS orphaned children in rural
Uganda. Journal of Adolescent Health, 42(1), 102-104
Ssewamala, F. M., Sperber, E., Zimmerman, J., & Karimli, L. (2010). The potential of assetbased development strategies for poverty alleviation in sub-Saharan Africa. International
Journal of Social Welfare, 19(4), 433-443.
Ssewamala, F., Curley, J. (2005). Asset ownership and school attendance of orphaned children
in Sub-Saharan Africa. Social Development Issues, 28(2), 84-105
153
Stewart, E.B. (2007). School structural characteristics, student effort, peer associations, and
parental involvement: The influence of school- and individual-level factors on academic
achievement. Education and Urban Society, 40, 179-204.
Sun, J., Buys, N., & Wang, X. (2012). Association between low income, depression, selfefficacy and mass-incident-related strains: an understanding of mass incidents in China.
Journal of Public Health, 34(3), 340-347. doi: 10.1093/pubmed/fdr107
Sundstrom, E., Bell, P. A., Busby, P. L., & Asmus, C. (1996). Environmental psychology 19891994. In M.R. Rosenzweig & L. W. Porter (Eds.), Annual Review of Psychology, 47,
485-512.
Suryadarma, D., Suryahadi, A., Sumarto, S., and Rogers, F. H. (2006). Improving student
performance in public primary schools in developing coutries: Evidence from Indonesia.
Education Economics, 14(4), 401-429.
Tamura, R. (2004). Human capital and economic development (Working Paper 2004-34).
Federal Reserve Bank of Atlanta. Retrieved from
https://www.frbatlanta.org/filelegacydocs/wp0434.pdf
Tansel, A. (1997), Education attainment, parental education, and gender in Cot d’Ivoire and
Ghana. Economic Development and Cultural Change, 45(4), 825-856.
Taylor, S. E., & Pham, L. B. (1996). Mental simulation, motivation, and action. In P. M.
Gollwitzer, & J. A. Bargh, The pschology of action: Linking congition and motivation to
behavior (pp. 219-235). New York: The Guilford Press.
Thompson, B. (2004). Exploratory and confirmatory factor analysis: Understanding concepts
and applications. Washington, DC: American Psychological Association.
154
Thornberry, T.P., Lizotte, A.J., Krohn, M.D., Farnworth, M., & Jang, S.J. (1991). Testing
interactional theory: An examination of reciprocal causal relationships among family,
school, and delinquency. Journal of Criminal Law and Criminology, 82, 3-35.
Tonah, S. (2009). The unending cycle of education reform in Ghana. JERA/RARE, 1, 45-52
Topor, D. R., Keane, S. P., Shelton, T. L., & Calkins, S. D. (2010). Parental involvement and
student academic performance: A multiple mediational analysis. Journal of Prev Interv
Community, 38(3), 183-197.
Torres, J. B., and Solberg, V. S. (2001). Role of self-efficacy, stress, social integration, and
family support in Latino college student persistence and health. Journal of Vocational
Behavior, 59(1), 53–63.
UNESCO Institute for Statistics (2012). Opportunities lost: The impact of grade repetition and
early school leaving. Global Education Digest 2012. Montreal: UNESCO Institute for
Statistics. Retrieved from
http://www.uis.unesco.org/Education/GED%20Documents%20C/GED-2012-CompleteWeb3.pdf
UNESCO Institute for Statistics (2013). Education (all levels) profile – Ghana. UIS Statistics in
Brief. Retrieved from
http://stats.uis.unesco.org/unesco/TableViewer/document.aspx?ReportId=289&IF_Langu
age=eng&BR_Country=2880&BR_Region=40540
UNGEI (2006). Call for abolition of school fees. UN Girl’s Education Initiative. Retrieved from
http://www.ungei.org/infobycountry/247_885.html
UNGEI (n.d.). Ghana: Background. UN Girl’s Education Initiative. Retrieved from
http://www.ungei.org/infobycountry/ghana.html?q=printme
155
Wang, J. (2011). Culture difference and English teaching. English Language Teaching, 4(2),
223-230.
Warne, R. T. (2011). An Investigation of Measurement Invariance Across Genders on the
Overexcitability Questionnaire–Two. Journal of Advanced Academics, 22(4), 578-593.
doi:10.1177/1932202X11414821
Watkins, T.J. (1997). Teacher communications, child achievement, and parent traits in parent
involvement models. Journal of Educational Research, 91(1):3-12
Weir, S. (1997, December). The determinants of school enrollment in rural Ethiopia. Mimeo,
Centre for the Study of African Economies, University of Oxford.
Werner, C. M., & Altman, I. (2000). Humans and nature: Insights from a transactional view. In
S. Wapner, J. Demick, T. Yamamoto, & H. Minami (Eds.) Theoretical perspectives in
environment-behavior research: Underlying assumptions, research problems, and
methodologies (pp. 21-37). New York: Plenum Press.
Werner, C. M., Brown, B. B., & Altman, I. (2002). Transactionally-oriented research: Examples
and strategies. In R. B. Bechtel & A. Churchman (Eds.) Handbook of Environmental
Psychology (2nd ed., pp. 203-221). New York: John Wiley & Sons, Inc.
Whittaker, T. A. (2012). Using the modification index and standardized expected parameter
change for model modification. Journal of Experimental Education, 80(1), 26-44.
Williams Shanks, T. R., Kim, Y., Loke, V., & Destin, M. (2010). Assets and child well-being in
developed countries. Children and Youth Services Review, 32(11), 1488−1496.
Willis, G. B. (2005). Cognitive interviewing: A tool for improving questionnaire design.
Thousand Oaks, CA: SAGE Publication, Inc.
156
Wolff, E. (1995). How the pie is sliced: America‘s growing concentration of wealth. The
American Prospect, 22, 58-64.
Woodhall, M. (1992). Student loans in developing countries: Feasibility, experience and
prospects for reform. Higher Education, 23(4), 347-356.
World Bank (2010). Education in Ghana: Improving equity, efficiency and accountability of
education service delivery (Report No. 59755-GH). Washington, D.C.: World Bank.
World Bank & UNICEF (2009). Abolishing school fees in Africa: Lessons from Ethiopia,
Ghana, Kenya, Malawi, and Mozambique. Development Practice in Education.
Washington, DC: The International Bank for Reconstruction and Development / The
World Bank.
http://www.unicef.org/publications/files/Aboloshing_School_Fees_in_Africa.pdf
Wößmann, L., & West, M. (2006). Class-size effects in school systems around the world:
Evidence from between-grade variation in TIMSS. European Economic Review, 50(3),
695-736.
Yadama, G., & Sherraden, M. (1996). Effect of assets on attitudes and behaviors: Advance test
of a social policy proposal. Social Work Research, 20, 3-11.
Yeung, J., & Conley, D. (2008). Black - white achievement gap and family wealth. Child
Development, 79(2), 303-324. doi:10.1111/j.1467-8624.2007.01127.x
Yousefi, F., Redzuan, M., Bte, M., Juhari, R. B., Talib, M. A. (2010). The effects of family
income on test-anxiety and academic achievement among Iranian high school students.
Asian Social Science, 6(6), 89-93
Zajacova, A., Lynch, S. M., & Espenshade, T. J. (2005). Self-efficacy, stress, and academic
success in college. Research in Higher Education, 46(6), 677-706
157
Zhan, M. & Sherraden, M. (2009). Assets and liabilities, educational expectations, and
children’s college degree attainment (Research brief No. 09-63). Center for Social
Development. retrieved from http://csd.wustl.edu/Publications/Documents/RB09-63.pdf .
Zhan, M. (2006). Assets, parental expectations and involvement, and children's educational
performance. Children and Youth Services Reivew, 28(8), 961-975.
Zhan, M., & Sherraden, M. (2003). Assets, expectations, and children's educational achievement
in female-headed households. Social Service Review, 77, 191-211.
Zhan, M., & Sherraden, M. (2010). Assets and liabilities, race/ethnicity, and children's college
education (CSD Working Paper 10-08). St. Louis, MO: Washington University, Center
for Social Development.
Zhan, M., & Sherraden, M. (2011). Assets and liabilities, race/ethnicity, and children's college
education. Children and Youth Services Review, 33(11), 2168-2175.
Zhang, Z., & RiCharde, R. S. (1998). Prediction and Analysis of Freshman Retention.
Minneapolis, MN: Paper presented at the Annual Forum of the Association for
Institutional Research.
Zhou, H. (2002). Determinants of youth earnings: The case of Harare. Zambezia, 29(2), 213-220.
Zimmerman, B. J., Bandura, A., & Martinez-Pons, M. (1992). Self-motivation for academic
achievement: The role of self-efficacy beliefs and personal goal setting. American
Educational Research Journal, 29, 633-676.
Zweig, J. M., Yahner, J., & Rossman, S. B. (2012). Does recent physical and sexual
victimization affect further substance use for adult drug-involved offenders? Journal of
Interpersonal Violence, 27(2), 2348-2372. DOI: 10.1177/0886260511433517
158
Appendix A: Public and private JHS enrollment trends from 2001 to 2010
159
Appendix B: Illustration of the transition from primary school through SHS
90% of primary school
pupils transition to JHS
10% don’t
71% reach JHS 3
29% don’t
61% take exit exam
(BECE)
26% fail at
least 1 course
35% pass 4 core courses
and transition to SHS
160
Appendix C. Illustration of how the current study fits into the larger YouthSave project
Current
study
161
Appendix D: Data Abstraction Form for School Records
Name
Surname
First name
Continuous
assessment score
Attendance
(Number of Math
days)
Exam score
Class teacher’s assessment of student
Student’s class
English Math English participation is:
Student’s conduct
in class is:
❶ Outstanding ❶ Outstanding
❷ Satisfactory
❷ Satisfactory
❸ Unsatisfactory ❸ Poor
162
Student follows
teacher's directions:
❶ Always
❷ Sometimes
❸ Never