Identity, Cohesion, and Enmeshment Across Cultures

Florida State University Libraries
Electronic Theses, Treatises and Dissertations
The Graduate School
2014
Identity, Cohesion, and Enmeshment Across
Cultures
Summer Brooke Gómez
Follow this and additional works at the FSU Digital Library. For more information, please contact [email protected]
FLORIDA STATE UNIVERSITY
COLLEGE OF HUMAN SCIENCES
IDENTITY, COHESION, AND ENMESHMENT ACROSS CULTURES
By
SUMMER BROOKE GÓMEZ
A Dissertation submitted to the
Department of Family and Child Sciences
in partial fulfillment of the
requirements for the degree of
Doctor of Philosophy
Degree Awarded:
Spring Semester, 2014
`
Summer Brooke Gómez defended this dissertation on April 7, 2014.
The members of the supervisory committee were:
Ronald L. Mullis
Professor Directing Dissertation
Tomi Gomory
University Representative
Ann Mullis
Committee Member
Lenore McWey
Committee Member
The Graduate School has verified and approved the above-named committee members, and
certifies that the dissertation has been approved in accordance with university requirements.
ii
`
For any kid who ever felt controlled and not supported. And for any kid that has suffered because
their strengths were blithely overlooked or discouraged. For my clients, who have been and
continue to be the best mentors I could ever dream of. And for the ghosts that haunt the Dade
County hallways of my youth.
iii
`
ACKNOWLEDGEMENTS
I appreciatively acknowledge Sigmund Freud, Erik Erikson, James Marcia, Seth Schwartz, and
all dedicated identity scholars for teaching me how to place careful value on the examined lives
of others.
I am grateful to my committee for their unique contributions: Ann Mullis for her frankness and
her dedication to precision; Lenore McWey for her ability to meet each student as a person and
still uncompromisingly command her students’ personal bests; Nick Mazza for his example of
leadership, and his rare ability to function as a scholar-practitioner; Tomi Gomory for his
dedication to critical thought and his strong personal commitment to his students’ professional
development. Thank you all.
I also want to thank my major professor, identity scholar Ron Mullis, first and foremost for his
patience with me with regard to both my sometimes stubborn nature and also with regard to
sticking with me well into his retirement. Thank you, Dr. Ron, for your guidance and for going
the distance, literally, to prove that you believed in me and my work.
I have had the luxury and pleasure of a wonderful education in the great State of Florida.
Highlights for me have included the gifted program in Miami-Dade and Broward County Public
Schools, which kept me interested in learning; the Dade County Police Benevolent Association,
whose scholarship started me on my way not only with financial support but also with a deeply
appreciated vote of confidence and sense of purpose and pride; the Honors College at Florida
International University, which gave me a liberal arts experience and a level of personal
development support I now know I could not have done without; the National Student Exchange
program, which pushed me out of my comfort zone and fueled a sense of daring and adventure
that I am still gleefully cultivating to this day; and of course the Florida State University, forever
my alma mater and the beautiful moss-draped backdrop of my becoming.
I am grateful to so many influential mentors from my past, especially Linda Meehan for
grounding me with an appreciation for the unique ecology of my home state; Shearon Lowery
for helping me to pinpoint my academic and professional passions; Bill Kurtines for introducing
me to the field of identity scholarship and encouraging me to move toward it; Neil Kressel for
inspiring me to ask the questions that I believe matter most; Connor Walters for teaching me
about all humans by starting with children; Judge Kathleen Kearney and Katrina Boone for
preparing me well to deal with the real world; and Larry and Sandy Barlow as well as Lenore
McWey and Robert E. Lee for top notch clinical training.
Finally, I extend my deepest gratitude to my incredible family and friends: my grandparents
Guillermo and Angelina Gómez for funding my undergraduate education, for championing the
overall value of education, and for exemplifying the import of community; my Uncle Charlie for
inspiring me with his bright blend of ambition, creativity, and talent, and for always introducing
me to others in the context of my capability; my Aunt Donna and Uncle Ed for showing me how
to balance sanity and ambition with family and art; my Aunt Sandra, the Chafins, and my cousins
for keeping me in touch with my roots and core values; Nany, Glen, Nelly, Benji, Mardele,
Tony, Ronnie, Sylvia, Kenny, Trevor, Katrina, and Chelsie for being vocally proud of me; and
iv
`
my bold, brave, and uniquely brilliant friends for always asking about me with genuine interest
and unconditionally encouraging me.
Most important, I am grateful to my parents, Joy and Bill Gómez as well as my husband,
Joe Anson. Thank you, Mom and Dad, for giving in every conceivable way from a seemingly
bottomless well fed by faith and love and for teaching me what’s worth fighting for in life. And
thank you, Joe, for teaching me that love is a choice you make each moment every day for over
thirteen years. I would not exist in any recognizable way without the three of you and my
gratitude and love for each of you knows no bounds. Thank you from the bottom of my heart for
every second I am blessed to have with you all.
v
`
TABLE OF CONTENTS
List of Tables ............................................................................................................................x
List of Figures ........................................................................................................................ xi
Abstract ................................................................................................................................. xii
1. INTRODUCTION ...............................................................................................................1
Basis for Research......................................................................................................1
Study Overview .........................................................................................................1
Key Variables.............................................................................................................2
Rationale for Study ....................................................................................................2
Theoretical Underpinnings of the Research ...............................................................4
Research Questions ....................................................................................................5
Hypotheses .................................................................................................................5
2. LITERATURE REVIEW ....................................................................................................7
Theoretical Frameworks ............................................................................................7
Ego Identity Theory .........................................................................................7
Overview of Marcian identity statuses ..................................................8
Identity diffusion ....................................................................................8
Identity foreclosure ................................................................................8
Moratorium ............................................................................................8
Identity achievement ..............................................................................9
Rationale for using the ego identity development framework ...............9
Life Course Theory ........................................................................................10
Norms, position, and roles ...................................................................10
Family stages .......................................................................................11
Transition .............................................................................................11
Family life course ................................................................................11
Multiple dimensions of time ................................................................12
Rationale for using family life course development framework .........13
Family Differentiation: One-dimensional versus
Two-dimensional model ......................................................................13
Critical Review of Theoretical Frameworks ..................................................17
Marcian ego identity theory .................................................................17
Family life course development framework ........................................18
Relations between Family Differentiation and Ego Identity ...................................18
Relations between Family Differentiation and Psychological Well-Being .............19
Relations between Ego Identity and Psychological Well-Being .............................20
The Role of Culture and Ethnicity ...........................................................................21
Importance of Comparing Ethnic Groups ......................................................21
Ethnoracial Groups in the Research .........................................................................23
Rationale for Focus on White Non-Hispanic Participants .............................23
Rationale for Focus on Hispanic Participants ................................................25
vi
`
Research Participants ...............................................................................................29
Summary Descriptions of Variables ........................................................................30
Family Differentiation ...................................................................................30
Cohesion ..............................................................................................30
Enmeshment .........................................................................................30
Ego Identity....................................................................................................30
Commitment ........................................................................................30
Exploration ...........................................................................................31
Psychological Well-Being .............................................................................31
Life Satisfaction ...................................................................................31
Depressive symptoms ..........................................................................31
State anxiety .........................................................................................31
Review of Relationships among Variables in the Research ..........................32
Cohesion and commitment ..................................................................32
Cohesion and exploration ....................................................................32
Enmeshment and commitment .............................................................32
Enmeshment and exploration ...............................................................33
Commitment and psychological well-being ........................................33
Exploration and psychological well-being...........................................34
Family differentiation and psychological well-being ..........................34
Family differentiation, identity formation, and
psychological well-being ................................................................35
3. METHOD ..........................................................................................................................37
Sampling and Participant Recruitment ....................................................................37
Selection of Recruitment Sites .......................................................................37
Age Range of Participants .............................................................................37
Participant Recruitment .................................................................................38
Ethnoracial Criteria for Participants ..............................................................38
Descriptive Statistics and Frequencies...........................................................38
Combined sample.................................................................................38
Hispanic subsample .............................................................................39
Non-Hispanic subsample .....................................................................40
Measures ..................................................................................................................41
Family Differentiation ...................................................................................41
Ego Identity....................................................................................................42
Psychological Well-Being .............................................................................42
Reliability Statistics .................................................................................................43
Participant Demographic Information .....................................................................43
Analysis....................................................................................................................44
Assumptions of Structural Equation Modeling (SEM) ..................................45
Model specification ..............................................................................45
Exogenous variables measured without error ......................................45
Completeness of data set ......................................................................45
Multivariate normality .........................................................................45
vii
`
Overall assessment of assumptions ......................................................46
Structural Equation Model Identification ......................................................46
Structural Equation Models ...........................................................................46
Model 1: Configural model ..................................................................46
Model 2: Measurement invariance model............................................47
Model 3: Hypothesized model .............................................................47
Model 4: Modified model ....................................................................48
Power Analysis, Sample Size, and Missing Data ..........................................48
Statistical Predictions ...............................................................................................49
Ethical Considerations .............................................................................................52
4. RESULTS ..........................................................................................................................54
Structural Equation Models .....................................................................................54
Model 1: Configural Model ...........................................................................54
Model 2: Measurement Invariance Model .....................................................55
Model 3: Hypothesized Model.......................................................................55
Model 4: Modified Model ..............................................................................60
Research Questions & Hypotheses ..........................................................................61
Mean Differences by Gender, College Major, and Data Collection Site .................62
5. DISCUSSION ....................................................................................................................63
Review of Notable Findings ....................................................................................63
Theoretical Implications of Findings .......................................................................63
Ethnoracial Differences .................................................................................64
Implications for the Role of Family Differentiation in
Ego Identity Theory ..................................................................................66
Effect of cohesion on commitment ......................................................67
Effect of cohesion on exploration ........................................................68
Effect of enmeshment on commitment ................................................68
Effect of enmeshment on exploration ..................................................68
Direct Effects of Family Differentiation on Psychological Well-Being ........69
Relational Clinical Implications of Theoretical and Empirical Findings ................70
Limitations of the Research .....................................................................................71
Measurement and Comparability of Family Differentiation across
the Literature .............................................................................................71
Limitations of Sample ....................................................................................71
Exclusion of non-students ....................................................................71
Restriction of age range .......................................................................72
Regional bias ........................................................................................72
Model Fit ........................................................................................................73
Mediation in the Structural Equation Models ................................................73
Future Research .......................................................................................................74
Future Study 1: Confirmatory Factor Analysis (CFA) ..................................74
Future Study 2: The Role of Gender in Ego Identity Development ..............74
viii
`
Future Study 3: Replication with Nationally Representative ........................75
Sample of Hispanic Participants ...............................................................75
Summary and Conclusions ......................................................................................75
APPENDICES ...................................................................................................................78
A. INFORMED CONSENT FOR FLORIDA STATE UNIVERSITY ............................78
B. INFORMED CONSENT FOR FLORIDA INTERNATIONAL UNIVERSITY ........77
C. PARTICIPANT PACKET ...........................................................................................81
D. FACTOR LOADING ESTIMATES FOR MODEL 3:
HYPOTHESIZED MODEL ........................................................................................87
E. FLORIDA STATE UNIVERSITY INSTITUTIONAL REVIEW BOARD
HUMAN SUBJECTS APPROVAL LETTER ............................................................99
F. FLORIDA INTERNATIONAL UNIVERSITY INSTITUTIONAL
REVIEW BOARD HUMAN SUBJECTS APPROVAL LETTER ...........................100
REFERENCES ................................................................................................................101
BIOGRAPHICAL SKETCH ...........................................................................................109
ix
`
LIST OF TABLES
1
Demographic Variables by Ethnoracial Subsample ..........................................................41
2
Statistically Significant Path Coefficients for Model 3: Hypothesized Model ..................57
3
Percentage of Variance Explained by Model 3: Hypothesized Model ..............................58
4
Latent Variables Included in the Structural Equation Models:
Descriptive Statistics and Reliability. Non-Hispanic N = 233; Hispanic N = 330 .......59
5
Measurement Items for Cohesion and Enmeshment .........................................................89
6
Measurement Items for Commitment ................................................................................90
7
Measurement Items for Exploration ..................................................................................91
8
Measurement Items for Depressive Symptoms .................................................................92
9
Measurement Items for Life Satisfaction and Anxiety ......................................................93
10
Cohesion and Enmeshment Factor Loading Estimates
for Model 3: Hypothesized Model ................................................................................94
11
Commitment Factor Loading Estimates for Model 3: Hypothesized Model .....................95
12
Exploration Factor Loading Estimates for Model 3: Hypothesized Model .......................96
13
Depression Factor Loading Estimates for Model 3: Hypothesized Model ........................97
14
Life Satisfaction and Anxiety Factor Loading Estimates for
Model 3: Hypothesized Model.....................................................................................98
x
`
LIST OF FIGURES
1
Structural equation modeling (SEM) diagram reflecting the statistical
predictions of the study .................................................................................................49
2
Structural equation modeling (SEM)
Model 3: Hypothesized Model, with standardized parameters .....................................60
xi
`
ABSTRACT
The impact of family differentiation (cohesion and enmeshment), as measured by the Colorado
Self-Report of Family Functioning Inventory, on ego identity development (exploration and
commitment), as measured by the Ego Identity Process Questionnaire, was examined. This study
also examined the impact of family differentiation on psychological well-being, both directly and
indirectly via ego identity development. The psychological well-being variables were life
satisfaction, measured by the Satisfaction with Life Scale; depressive symptoms, measured by
the Center for Epidemiological Studies Depression Scale (CES-D); and state anxiety, measured
by a short form of the state-trait anxiety inventory (STAI-6). The research questions were: 1) Do
cohesion and enmeshment influence commitment and exploration? 2) Do cohesion and
enmeshment influence psychological well-being (anxiety, depressive symptoms, and life
satisfaction)? 3) Do these relationships differ between non-Hispanic white participants and
Hispanic participants of any race? Participants were undergraduate college students at two public
universities in the southeastern United States. The total sample size was N = 563. Structural
equation modeling (SEM) was used to examine the hypothesized relationships. This study also
examined indirect effects via commitment and exploration, identity formation variables drawn
from the ego identity development paradigm. Results show that, consistent with the study
predictions, cohesion had significant positive effects on commitment and life satisfaction, and
significant negative effects on depressive symptoms and anxiety (both directly and indirectly via
by commitment). Also as predicted, commitment had a significant positive direct effect on life
satisfaction, and significant negative direct effects on depressive symptoms and anxiety.
However, no effects were observed for enmeshment or exploration, and no differences were
observed in these relationships between the Hispanics and non-Hispanics. Implications for ego
identity theory and family relations theory are discussed, as well as limitations of the study
methodology and proposals for future research.
xii
`
CHAPTER ONE
INTRODUCTION
Basis for Research
This study expanded upon a European study by Manzi, Vignoles, Regalia, and Scabini
(2006). Manzi et al. (2006) used confirmatory factor analysis (CFA) to establish the concept of
family differentiation as consisting of two dimensions (cohesion and enmeshment). Manzi et al.
(2006) used structural equation modeling (SEM) to examine the impact of both cohesion and
enmeshment on identity threat and on measures of psychological well-being. Manzi et al. (2006)
also compared a model generated from Italian participants with a model generated from British
participants. They found significant differences in the impact of family differentiation across
cultures.
Study Overview
This study examined the relational concept of family differentiation in the context of ego
identity formation and the relevance and function of the two-dimensional model of family
differentiation. However, instead of investigating the impact of family differentiation on identity
threat, this study examined the effect of family differentiation on the identity development
processes outlined within the Marcian identity status paradigm (Marcia, 1966) by using the
Marcian concepts of commitment and exploration. This study is a response to a long-standing
call within the literature for better understanding of the relational underpinnings of ego identity
development (Marcia, 1989). This study examined the impact of family differentiation on
identity development using the same outcome measures of life satisfaction, depressive symptoms
and anxiety reported by Manzi et al. (2006).
In addition to five race categories, the United States Census Bureau (USCB) recognizes
two ethnic categories: Hispanic and Non-Hispanic (Humes, Jones, & Ramirez, 2011). This study
compares two ethnoracial groups based on racial and ethnic categories used in the 2010 United
States Census: non-Hispanic whites and Hispanics of any race. Thus, the present study serves as
a United States counterpart to the work of Manzi et al. (2006) by highlighting the relevance and,
subsequently, the function of family differentiation across United States ethnoracial groups of
late adolescent and early adult participants in a theoretically comparable life stage.
1
`
Key Variables
The impact of four independent variables was studied. These include two family
differentiation variables: cohesion and enmeshment. The other two independent variables are
identity formation variables: commitment and exploration. However, commitment and
exploration (representing identity formation) are also dependent variables in the study. Cohesion
and enmeshment (representing family differentiation) are the only independent variables that are
exogenous to the study's model. Three dependent variables were used to measure psychological
outcomes: life satisfaction, depressive symptoms, and anxiety. These variables serve as broad
measures of psychological well-being. Moreover, research has shown these three particular
outcome variables to be associated with identity threat, an important variable relevant to the
understanding of ego identity (Manzi et al., 2006).
Rationale for Study
Family and identity scholars have acknowledged family differentiation as a possible key
factor underlying the emotional development of older adolescents and young adults. Family
differentiation is the extent and nature of emotional interdependence among family members.
Scholars have often described family differentiation as a one-dimensional construct, with
estrangement (low interdependence) and enmeshment (high interdependence) as opposite poles
of a single spectrum (e.g., Campbell, Adams, & Dobson, 1984; Minuchin, 1974; Olson, 1982).
Each of these extremes has been theoretically associated with maladaptive family functioning
(e.g., Campbell et al., 1984; Minuchin, 1974; Olson, 1982). Moderate values on this spectrum
correspond to cohesion and have been theoretically associated with more adaptive outcomes.
(e.g., Campbell et al., 1984; Minuchin, 1974; Olson, 1982).
In recent years, scholars (e.g., Manzi et al., 2006) have criticized this one-dimensional
model in two important ways. Some researchers have challenged the generalizability of the onedimensional model of family differentiation across ethnic groups (e.g., Chun & MacDermid,
1997; Fuhrman & Holmbeck, 1995). Other scholars have proposed that a two-dimensional model
(e.g., Barber & Buehler, 1996; Manzi et al., 2006; Scabini, 1985) would be a more accurate way
of describing family differentiation. The two-dimensional model of family differentiation
represents cohesion and enmeshment as distinct dimensions.
Addressing these criticisms, Manzi et al. (2006) compared family differentiation among
Italian and British participants. Manzi et al. (2006) found that cohesion and enmeshment were
2
`
best represented as two distinct dimensions in both British and Italian samples. These two
dimensions were positively correlated in the Italian sample. However, they were uncorrelated for
the British sample. Manzi et al. (2006) also found that higher family cohesion led to decreased
identity threat across groups. This in turn led to higher levels of life satisfaction and lower levels
of depressive symptoms and anxiety.
The findings for family cohesion were very similar in the Italian and British samples.
However, cultural differences were observed for enmeshment. For British participants
enmeshment levels led to significantly increased identity threat. This in turn led to lower levels
of life satisfaction and higher levels of depressive symptoms and anxiety. In contrast, for Italian
subjects, there were no significant effects of enmeshment on identity threat, life satisfaction,
depressive symptoms, or anxiety. This cultural difference is consistent with the stronger
emphasis on individual autonomy in British compared to Italian culture.
The findings of Manzi et al. (2006) provide evidence that the two-dimensional model of
family differentiation more accurately represents the nature of family differentiation. The finding
applies across cultural groups. Thus, this study will assume the two-dimensional model rather
than the one-dimensional model of family differentiation.
A major purpose of this study is to examine the indirect effects of family differentiation
on psychological well-being via identity formation variables drawn from the ego identity
development paradigm (Marcia, 1966). This is justified because Manzi et al. (2006) found that
family differentiation affected psychological outcomes indirectly via identity threat. Manzi et al.
(2006) measured identity threat levels at a time when their older adolescent and young adult
participants were experiencing major life stage transitions. However, this study examines
indirect effects via identity formation variables (cohesion and enmeshment) rather than via
identity threat. Thus, this study will be the first to investigate the indirect effects of family
differentiation on psychological outcomes via cohesion and enmeshment.
Freud (1933) proposed that ego identity development was the product of relational
dynamics in the family of origin. Marcia (1989) has also urged study of the impact of relational
dynamics on ego identity development. There are several attempts in the literature to address the
relationship between family relational dynamics and the Marcian ego identity statuses (e.g.,
Adams, Dyk, & Bennion,1987; Arseth, Kroger, Martinussen, & Marcia, 2009; Campbell et al.,
1984; Faber, Edwards, Bauer, & Wetchler, 2003; Jackson, Dunham, & Kidwell, 1990; Marcia,
3
`
1966; Mullis, Brailsford, & Mullis, 2003; Watson & Protinsky, 1988). However, none of these
studies have yielded decisive findings or succeeded in explaining the variance in ego identity
variables using family dynamics variables. Moreover, no study has used the two-dimensional
model of family differentiation to explain the ego identity development processes of older
adolescents or young adults.
In summary, the present study used structural equation modeling [SEM] to examine the
impact of family differentiation (cohesion and enmeshment) on the ego identity variables
(commitment and exploration). This study also examined the impact of the ego identity variables
on psychological well-being outcomes (anxiety, depressive symptoms, and life satisfaction) and
tested whether these relationships vary across the two ethnoracial groups studied.
Theoretical Underpinnings of the Research
Family life course development theory and ego identity development theory are the two
theoretical frameworks used in the present research. Family life course development theory is a
synthesis of family development theory and the life course perspective (White & Klein, 2002).
Family life course development theory posits that society imposes certain norms governing
behavior within a family. The resulting within-system expectations depend upon system
member's stages of development and positions within the family, as well as upon the entire
system's place in the developmental process. Transitions at the level of the family or the
individual from one stage of development to another are key components of this framework.
Problems are presumed to arise when these transitions are off-time (too early or too late).
The use of the family life course development theoretical framework will enhance the
theoretical generalizability of the findings across cultures and eras (e.g., Hammer, 2003).
Specific norms related to life stage transitions vary across cultures and eras. However, life course
development theory facilitates comparison of the results of this study with those of studies on
other populations. This is because it is possible to discuss psychological outcomes in terms of
life course transitions across cultures.
The Marcian ego identity status paradigm (Marcia, 1966) holds that individuals may be
classified into one of four ego identity statuses: achievement, moratorium, diffusion, and
foreclosure. An individual’s identity status is determined by two underlying dimensions:
commitment and exploration. Commitment is the extent to which an individual’s decisions
reflect a stable pattern of identity. Exploration is the extent to which an individual engages in
4
`
active, reflective construction of their own identity. Achievement status indicates high
commitment and high exploration. Moratorium status indicates low commitment and high
exploration. Diffusion status indicates low commitment and low exploration. Foreclosure status
indicates high commitment and low exploration.
Identity commitment is presumed to provide individuals with a sense of purpose
(Crocetti, Klimstra, Keijsers, Hale, & Meeus, 2009) and thus expected to enhance overall
psychological well-being: Commitment is expected to reduce anxiety because it reduces
uncertainty about life goals. Commitment is expected to reduce depressive symptoms because a
sense of purpose reduces feelings of helplessness. Finally, commitment is expected to increase
life satisfaction because it implies that the individual is living according to their own selfselected standards.
By contrast, exploration is expected to have a positive impact on psychological wellbeing outcomes for white non-Hispanic participants. This is supported by empirical evidence
using various measures of adaptive functioning (e.g., Arseth et al., 2009; Crocetti et al., 2009;
Hardy & Kisling, 2006). However, it is expected that the impact of exploration will be less
positive for Hispanics of any race, due to cultural differences in the function of individual
autonomy (e.g., McAdoo, 1981; Watson & Protinsky, 1988). This prediction is also consistent
with the findings of Manzi et al. (2006) regarding British and Italian cultures; Italian participants
were less affected by exploration when compared to British participants.
Research Questions
The present study addresses the following research questions:
1. Does family differentiation (cohesion and enmeshment) relate to identity formation
(commitment and exploration levels) of adolescents and/or young adults?
2. Does family differentiation predict psychological well-being?
3. Do the findings above differ between non-Hispanic whites and Hispanics of any race?
Hypotheses
1. Family differentiation (cohesion and enmeshment), will relate to the identity formation
(commitment and exploration levels) of adolescents and/or young adults by explaining a
significant amount of variance in ego identity status scores, which will be measured by
the Ego Identity Process Questionnaire (EIPQ; Balistreri, Busch-Rossnagel, & Geisinger,
1995).
5
`
2. Family differentiation (cohesion and enmeshment) will both directly and indirectly
predict psychological well-being (anxiety, depressive symptoms, and life satisfaction).
Indirect effects will occur via ego identity variables (commitment and exploration).
3. Differences between ethnic groups will be found with regard to the determinants of
psychological well-being (anxiety, depressive symptoms, and life satisfaction). For
participants who identify themselves as Hispanic, higher levels of enmeshment will be
less associated with diminished psychological well-being. Psychological well-being was
represented in terms of anxiety, depressive symptoms, and life satisfaction as measured
by the Satisfaction with Life Scale (Diener, Emmons, Larsen, & Griffin, 1985); the
Center for Epidemiologic Studies Depression Scale (CES-D; Radloff, 1977); and the
short form of the Spielberger State-Trait Anxiety Inventory (STAI-6; Marteau & Bekker,
1992), respectively. Similarly, for participants who identify as Hispanic (compared to
those who identify as non-Hispanic whites), higher levels of exploration will be
significantly less associated with diminished psychological well-being (anxiety,
depressive symptoms, and life satisfaction as measured by the instruments mentioned
above).
6
`
CHAPTER TWO
LITERATURE REVIEW
This chapter reviews literature regarding connections among family differentiation,
identity formation (independent variables) and psychological well-being (dependent variables).
In addition, this chapter will focus on evidence for the role of indirect effects via identity status
in explaining the impact of family differentiation on psychological well-being. Theoretical
frameworks used in this study are first discussed to help in the conceptualization of the major
variables.
Theoretical Frameworks
Ego identity theory (e.g., Marcia, 1995; Schwartz, 2001) and family life course
development theory (Elder, 1985; White & Klein, 2002) are used as theoretical frameworks for
this study in conceptually explaining relationships between family differentiation, identity status,
and the psychological well-being of older adolescents and young adults.
Ego Identity Theory
Rooted in the work of Sigmund Freud and Erik Erikson, identity theory is relevant across
social science and humanities disciplines (Bosma, Graafsma, Grotevant, & de Levita, 1994;
Marcia, 1966; Schwartz, 2001; and Society for Research on Identity Formation [SRIF], 2013).
Ultimately basing his foundational theoretical work on the Freudian notion that one’s concept of
self can be traced to early child-parent interaction (Schwartz, 2001), Erikson was the first
researcher to use the phrase “ego identity” in the 1960s in describing the resolution of one of his
most well-known crisis stage; identity vs. role confusion (Kroger, 2003). This crisis stage is
associated with adolescents. Assuming a period of intrapsychic destructuring and, ultimately,
crisis resolution, the crisis stages that theoretically follows is intimacy vs. isolation and it is
associated with young adults (Erikson, 1959). In midcentury, Erikson built on Freudian concepts
by expanding his focus of the then-emerging body of research on personality throughout the life
course, depicting the ideal emergence of adult identity as the result of a series of resolutions to
staged crises (Thomas, 2005). James Marcia, in turn, is the intellectual progenitor of most
modern research on identity development (Schwartz, 2001; Waterman, 1988).
There has been some debate in the literature about how directly compatible Marcia’s
(1966) seminal work on identity status is with the Eriksonian concepts it is presumed to be built
7
`
upon (Blasi & Glodis, 1995; Côté & Levine, 1988; van Hoof, 1999). However, it is undeniable
that Marcia’s work has been influential among identity theorists beyond Erikson for half a
century (Kroger, 2003; Schwartz, 2001; Waterman, 1988). Schwartz (2001) referred to the works
of Freud, Erikson, and Marcia as “the fundamentals of identity theory” (p. 8).
Overview of Marcian identity statuses. Marcia (1966) is best known for establishing
four basic identity statuses: Diffusion, foreclosure, moratorium, and achievement. Marcia’s four
basic identity statues differ from Erikson’s psychological stages in that they are not construed as
stages at all, but ways of summarizing personality traits that may or may not describe a person at
any point in life. Further, these four stages represent a unique combination of either higher or
lower levels of exploration and commitment. Exploration refers to the degree to which an
individual leads an examined life, or makes personal choices with an eye toward the active
construction of the self. Commitment refers to an individual’s ability to consistently identify with
those values and roles one has chosen to represent oneself and to make new choices that reflect a
stable and consistent pattern of self-selected identity.
Identity diffusion. A status of identity diffusion reflects low commitment and low
exploration. That is, a person that could be described as identity diffused would lack both
stable conception of self and a current sense that she ought to be seeking such a conception. An
example of this might be a young adult child, living with her parents, not attending school, not
finding a group of peers that she has much in common with, not pursuing any particular career
path, and not expressing any interest in any change with regard to these matters.
Identity foreclosure. A status of identity foreclosure represents high commitment and
low exploration. That is, a person that could be categorized as identity foreclosed would, without
significant consideration of other possible lifestyles, adopt a role, a set of values, and likely a set
of peers and hobbies consistent with a lifestyle that was modeled for them or expected of them,
often by elders or authority figures. An example of this status might be an adolescent who
decides on a career path that her parents want for her even though she is unsure as to the
importance of this career for herself.
Moratorium. Moratorium is an identity status characterized by low commitment and
high exploration. People best described as having this identity status are currently engaged in
questioning any previous assumptions, are constructing a sense of self, and are not immediately
interested in defining themselves, their most appropriate roles, or the type of lifestyle that best
8
`
fits them. For example, if the adolescent noted above decides that she will go to college to pursue
the career path her parents want for her now, but also plans to explore other careers once she
leaves home for college. In this case, she does not want to challenge her parents now on the issue
of careers and will wait until she is emotionally ready to decide for herself.
Identity achievement. This identity achievement status is unique in that it arguably
implies that a person that could best be described as identity achieved has been through a
moratorium stage and that what she has achieved is the ability to make and consistently stick
with lifestyle and identity choices with awareness. This status implies at least a previous period
of high exploration as well as a high level of commitment.
Rationale for using the ego identity development framework. Given the theoretical
role of family differentiation in foundational descriptions of ego identity formation ranging from
Freud (1933) to Marcia (1966), it is worth specifically considering what the observable
connection might be between the two-dimensional model of family differentiation, including the
dimensions of cohesion and enmeshment, and the identity formation process. In general, it is
appropriate to examine family differentiation in terms of its impact on identity formation because
youth within families are theoretically presumed to be struggling to take on the demands of
adulthood by identifying their own values and needs as separate from parents who socialize them
(Grotevant, 1983).
Further, the use of the identity development as a theoretical framework in the context of
this study also provides an opportunity for a theoretical contribution. This study is intended to
address a long-standing call within the literature for enhanced cultural literacy on the part of
identity scholars (e.g., Côté and Levine, 1988). Scholars have frequently implied or suggested
that identity foreclosure is associated with less favorable outcomes than identity achievement
(e.g., Arseth et al., 2009). Schwartz (2001) combined multiple perspectives on the identity
foreclosure status as part of an effort to offer a broad historical overview of Marcian identity
scholarship. He cited Marcia (1980) in describing the identity foreclosed individual as being
“associated with some degree of closed-mindedness, smug self- satisfaction, and rigidity” and (p.
12), going on to say that “foreclosed individuals tend to become increasingly attached to their
current circumstances and to the individuals who have helped to put those circumstances in
place, and they tend to resist change at almost any cost” (p.12). Schwartz (2001) explains that
the identity foreclosed individual has a high degree of commitment “in the relative absence of
9
`
prior exploration” (p.12), arguably implying some contextual level dysfunction inherent in an
individual that enters into a higher commitment status without exploration. This study will
critically examine the argument that the negative connotation exemplified by this language
reflects culturally relevant assumptions across the identity literature. To this end, this study
examined whether the impact of enmeshment on exploration and also on psychological wellbeing is less negative for Hispanics than for non-Hispanic whites.
Life Course Theory
The family life course development framework represents a synthesis of two theoretical
frameworks: a) Family development theory, and, b) The life course perspective (White & Klein,
2002; see also Bengston & Allen, 1993; Elder, 1998; Elder, Johnson, & Crosnoe, 2003; Rodgers
& White, 1993). Family development theory emerged in the 1950s. It is rooted in an interest in
the family as a unit of analysis. The life course perspective evolved in the 1970s and is based on
the notion of life stages that are tied more to the timing of significant events (e.g., birth of a child
or marriage) than to chronological time. White and Klein (2002) argued that these two theories
can be construed as different aspects of a single framework. The integration of family
development theory with the life course perspective affords the researcher an organized way to
consider the development of not just the individual but the family in the context of history,
society, stage of life, and family-specific life changing events.
Norms, positions, and roles. Norms are societal rules that regulate group and individual
behavior. The family life course development framework acknowledges that while some norms
are constant across the life span, such as the societal prohibition of theft, others are tied to an age
and stage graded continuum. For example, in the U.S. we do not generally expect adolescents to
maintain steady employment; at the onset of adulthood or completion of education, that
expectation changes.
Position refers to an individual's place in a family structure relative to others. This
concept is useful in differentiating families from within and between other groups. For example,
in U.S. society, we generally conceptualize families as being comprised of mothers, fathers,
sons, daughters, sisters, and brothers. These primary positions allow us to conceptualize other
culturally relevant positions, such as grandmothers, grandfathers, aunts, uncles, and cousins.
Thus, a grandmother can be thought of as a mother or father’s mother and an aunt can be thought
of as a mother or father’s sister.
10
`
A role is the set of norms applicable to a family position. Roles, like the norms they are
tied to, can be age and stage graded. For example, in many societies, when people think of the
concept of a son, they think of somebody that is financially dependent on a father or mother.
Indeed, as a child or adolescent, a son may be likely to be largely dependent on his parent(s) for
financial support. However, as that son ages, his role may change in accordance with societal
norms (e.g., adult sons become gainfully employed and achieve financial independence). In this
example, the role of the son position evokes societal norms about gender as well as autonomy.
Family stages. As opposed to Erikson’s (1968) or Marcia’s (1966) identity
developmental categories, which focus more on individual development, a family stage
represents a period of time with distinct expectations for changing interdependent family roles.
Individual roles and role interaction within the family generally change with passing family
stages. For example, after a family has entered the stage of launching children, steady
employment is associated with autonomy for a young adult son. Therefore, steady employment
is an appropriate element for the role a young adult son. In the context of the family stage
immediately prior to this, however, primary identification with the role of an employee during
adolescence is correlated with criminality and antisocial behavior (Wright, Cullen, & Williams,
2002).
Transition. In the family life course development framework, a transition is the passage
from one individual or family stage to another. A transition may be on-time or off-time. Offtime transitions are associated with deleterious consequences for family units and family
members. Timing is regulated by age and stage graded social norms. For example, if the
transition from adolescence into the adjacent phase of adulthood is too early, or off time, due to a
financial necessity for employment during adolescence, antisocial behavior such as theft,
vandalism, or physically violent outbursts may be more likely to follow (Wright et al., 2002).
The family life course development framework holds that transitions, such as the
individual transition from high school or college student to high school or college graduate or
dropout are formally marked by distinct events (e.g., graduation day or day of formal withdrawal
from school), but can be characterized by a prolonged transition process.
Family life course. Family life course is family development, conceptualized as an
unending series of individual and group transitions form one stage to another. The family life
course is organized along the lines of on-time and off-time transitions and roles are regulated by
11
`
societal and cultural norms. Interaction within a family and the interaction of the family and
society keep development continuous.
Multiple dimensions of time. Scholars rooted in the family life course development
recognize that although the measurement of time may be uniformly regimented, the experience
of time is quite the opposite. This framework also posits that people experience time not so much
a function of precisely calculated minutes, hours, weeks and years, but in terms of significant
events. This framework appreciates that the family developmental process is not based solely on
an unfolding of biological events. Development is directly influenced by historical factors as
well as societal norms about the order in which specific events are to take place. Further, this
framework recognizes the tendency of people to intentionally organize their individual and
familial growth along the lines of these sequencing norms. Family stages of development are not
just sequential but they also influence one another. For example, the more time an adolescent
spends engaging in behaviors such as studying and participating in extracurricular activities that
lead her to become embedded in the high school phase of her life, the more likely she is to
graduate high school (Alexander, Entwisle, & Kabbani, 2001). This refers not only to the amount
of total time spent in a phase, but also to the way the behavior and events that occur during a
given stage predict future transitions. Although individuals and families certainly stray from
expected sequences, the life course development framework assumes that this has consequences
that can often be measured and predicted.
In an examination of the effects of unemployment on the transition from adolescence to
adulthood in Norway, Hammer (1996) found that young, unemployed men were more likely than
their female counterparts to remain living at home with their parents and also had more of a
tendency to behave according to traditional gender roles. From a family life course development
perspective, this could be explained by assuming that these young men, perceiving pressure to
conform to a societal norm of adult financial independence and autonomy, behave in ways that
are especially consistent with and age and stage appropriate gender roles in an effort to
compensate and self-regulate the family developmental process. This is consistent with and
relevant to more recent empirical findings (e.g., Haid et al., 2010). Hammer (2003) and his
multinational team followed up with this work, offering a more thorough exploration of the
effects of unemployment on young people in the context of European and Western society.
Employing concepts from the family life course development framework, they were able to
12
`
organize a complex discussion of the potential longitudinal family and individual consequences
of unemployment among 15 to 24 year old males and females within the broader social context
of time-limited opportunities, poverty, and marginalization.
Rationale for using family life course development framework. Employing concepts
from the family life course development framework in this study will afford an opportunity to
organize and address the complexity of relevant societal and cultural expectations for the
purpose grasping as much as possible about the perspectives of participants that live within the
ever-evolving context of U.S. society. This framework will also help to bridge the gap between
the inevitable limitations of a study that only addresses a sample of participants living in Florida
and in U.S. society at a given point in history and the international community of identity
research consumers. For example, by employing this framework in this study to conceptualize
participant age in terms of life stage and generally expected major transitions in socio-economic
and broader societal context, a researcher studying a different population in another time may be
able build on this work by using the life course development framework to also explain those
participants' ages in terms of life stage and generally expected major transitions in
socioeconomic and broader societal context. Employing the family life course development
framework in this research project offers a strong opportunity for theoretical generalizability. In
short, the family life course development framework facilitates comparison of the results of this
study to those of research on other populations, because although specific transition-related
norms vary across cultures, it is possible to discuss the family and individual experience in terms
of life course transitions and family life course transitions regardless of culture or broader
societal setting.
Family Differentiation: One-dimensional versus Two-dimensional Model
The concept of family differentiation, or the extent and nature of emotional
interdependence among family system members, has been extensively addressed by relational
social scientists and researchers. Historically, researchers (e.g., Minuchin, 1974 & Olson, 1982)
have used a single dimension to explain family differentiation where an ideal balance of
autonomy is indexed as a single dimensional construct. This model of differentiation has been
termed the one-dimensional model (Manzi et al., 2006), and historically it has gone relatively
unchallenged by identity and family science scholars.
13
`
In recent years, however, the one-dimensional model of differentiation has been criticized
as not fully capturing family differentiation. For example, some researchers have challenged the
model's generalizability across cultural and ethnic boundaries (e.g., Chun & MacDermid, 1997;
Fuhrman & Holmbeck, 1995). Others have suggested that a two-dimensional model better
accounts for the nature of family differentiation (e.g., Barber & Buehler, 1996; Manzi et al.,
2006; Scabini, 1985). Such a model would reflect two separate dimensions of differentiation
including family cohesion and family enmeshment.
Previous research has further highlighted the distinction between a one- versus twodimensional model. For example, Campbell et al. (1984) conducted a study on cohesion and
enmeshment that has been cited often in identity formation literature. The study explored the
relationships between two independent variables, identity status and gender, and four outcome
variables: connectedness, communication, individuality, and satisfaction with independence.
Campbell et al. asserted that their research was a response to Grotevant's contributions
(Grotevant, 1983; Cooper, Grotevant, & Condon, 1984) regarding the family differentiation
constructs, stating “our interpretation of Grotevant's position is that a moderate degree of
connectedness, reflected through shared affection and an acceptance of individuality, provides
the psychological foundation and security to begin the searching process of self-defined
commitments” (p. 512).
This statement is consistent with the one-dimensional model of cohesion and
enmeshment. In effect, the authors define enmeshment as a high level of cohesion, or
connectedness: they use the term “enmeshment” to explain that “extreme affection
(enmeshment) between adolescents and their parents and limited family-based tolerance for
individuality might enmesh adolescents and encourage foreclosure in identity formation” (p.
512). The authors further concluded that the one- dimensional model of cohesion and
enmeshment is supported.
However, the authors' hypotheses themselves, specifically the supported ones, actually
appear to be more consistent with a two-dimensional model of family differentiation. Consider
what it would have looked like had Campbell et al. (1984) constructed hypotheses designed to
test a two-dimensional model of family differentiation. They might have predicted that highcommitment (foreclosed or identity-achieved) adolescents would have more affectionate
relationships with their parents than those in low commitment statuses (moratorium or diffuse).
14
`
In fact Campbell et al. did predict this; and the hypothesis was supported. They might have also
predicted that high-exploration adolescents (identity-achieved or moratorium) would have more
perceived independence (lower enmeshment) from their parents than their lower exploration
counterparts (foreclosed or diffused). Campbell et al. predicted this as well; and this hypothesis
was also supported.
To focus on both family differentiation and identity development, Watson and Protinsky
(1988) examined family interactions and their relationship to adolescent identity development in
black U.S. families. They assumed a one dimensional model of enmeshment and set out to
explore the extent to which enmeshment predicted identity status. The researchers theorized that
a higher enmeshment status would be predictive of identity foreclosure, and that hypothesis was
supported by their study. However, contrary to their stated expectations, the researchers also
concluded that high enmeshment levels were predictive of identity achievement status.
Support of a two-dimensional model of family differentiation including cohesion and
enmeshment comes from Fuhrman and Holmbeck (1995) who examined relationships between
adolescent emotional autonomy and parent child relationships. Participants included 96
adolescents at an urban Catholic school in a large city in the eastern United States (U.S.). The
mothers and teachers of subjects also participated in the study. The authors employed multiple
regression to test this relationship and to test eight moderating variables. Ultimately, the authors
concluded that the effect of emotional autonomy on the parent-child relationship varies
according to the unique qualities of the relationship, specifying that emotional autonomy
predicted a reduction in adjustment only when the parent-child relationship could be
characterized as warm and low conflict. Emotional autonomy actually predicted higher
adjustment if the parent-child relationship could be considered cold or high conflict.
As a more direct test of both the one-dimensional and two-dimensional models, Barber
and Buehler (1996) examined the viability of the two-dimensional model compared to the onedimensional model of cohesion and enmeshment among U.S. children. In measuring the
independent variables, Barber and Buehler (1996) used the cohesion and enmeshment subscales
of the Colorado Self Report of Family Functioning Inventory [CSRFFI]. Each CSRFFI subscale
consists of five 4-point items. They used exploratory factor analysis with oblimin (nonorthogonal) rotation, an analysis that allows factors to be correlated. The authors' analysis
established a two-factor solution, which explained 54 percent of the variance in item responses.
15
`
In measuring outcomes, the authors utilized four subscales of the Child Behavior Checklist
Youth Self Report (CBC): Withdrawn (α =.57), Anxious-Depressed (α =.87), Delinquent
(α =.62), and Aggressive (α =.79). The researchers further identified the moderating variables of
grade in school and gender. The researchers drew from a population of white, middle income,
fifth, eighth, and tenth grade children in suburban Tennessee. The authors recruited the entire
population of school children within a selection of grade levels and schools, with a 65 percent
response rate (N = 471). In their analysis, the researchers employed multiple regression to
compare both linear (two-dimensional) and curvilinear (one-dimensional) models of cohesion
and enmeshment. Linear models yielded nine significant results related specifically to cohesion
and enmeshment, while the curvilinear models yielded only one significant result, supporting the
conclusion that cohesion and enmeshment are best conceptualized as two distinct dimensions,
each having a unique linear relationship with relevant outcomes.
Barber and Buehler (1996) concluded that a two-dimensional model was supported by
their analysis. In this study, the authors measured cohesion and enmeshment separately and
found differences in the ways that cohesion and enmeshment predicted an established set of
outcomes. In addition, the authors noted that the independent variables, cohesion and
enmeshment, interacted with one another, adding further support to their conclusion that a twodimensional model was a more reasonable and accurate understanding of the nature of cohesion
and enmeshment.
It is noteworthy that Barber and Buehler (1996) made a point of attempting to isolate
white, middle income children within their sample for study to the extent of removing fully 53
potential participants, roughly ten percent of their overall sample, explaining only that they did
this because the vast majority of their sample size was comprised of white, middle income
children. This study cannot be taken to explain the relevance of the cohesion and enmeshment
constructs in a cross cultural or minority culture context, especially with regard to outcomes.
Chun and MacDermid (1997) studied the implications of what they referred to as
differentiation and individuation levels for 170 Korean adolescents in what would be the US
equivalent of ninth or tenth grade. They directionally hypothesized that Korean adolescents'
individuation levels would be negatively correlated with self-esteem. They also hypothesized
that gender would play an important role in interrelationships between individuation and that
differentiation between and adolescent and his or her same-sex parent would be more strongly
16
`
related to adolescent individuation than differentiation among parents or in the adolescent's
relationship with the opposite sex parent. These hypotheses were supported. In addition, the
influence of father-adolescent differentiation on adolescent individualization (beta = .32) was
found to be much stronger than the influence of mother-adolescent differentiation; for girls the
converse was true (beta=.61). Considered in the context of our non-homogenous society, this
study is a powerful argument for studying and understanding the implications of family
differentiation outside of a narrow cultural or ethnic reference group.
Critical Review of Theoretical Frameworks
Marcian ego identity theory. Though Marcia (1966) did not present his ego identity
theory as a stagewise model, he developed his identity statuses by expanding upon Erik
Erikson’s stagewise psychosocial crisis resolution paradigm. Further, the Marcian identity status
of identity achievement arguably implies that an individual has passed through a period of
moratorium on her way to achievement. Therefore, a stagewise element of the theory appears to
be implied. This stagewise element may contribute to a tendency for scholars in the field to
assume a high value of exploration. Empirically, this valuation may not be universal.
Additionally, some findings suggest that commitment may be associated with positive outcomes
even in the absence of exploration. In other words, commitment, inclusive of identity
achievement as well as identity foreclosure, may be associated with positive outcomes (e.g.
Arseth, Kroger, Martinussen, & Marcia, 2009); Crocetti, Klimstra, Keijsers, Hale, & Meeus,
2009; Mullis et al., 2003; Watson & Protinsky, 1988). Yet, due to a theoretically inherent
emphasis on the value of exploration as opposed to commitment, it is possible that the data that
supports this may be at risk for being under-acknowledged.
Much of the research that has been done on ego identity development has not addressed
the inter-related issues of culture, race, and ethnicity. Historically, many studies in this area have
used data from participants that predominantly identify as white (Schwartz, 2005). Côté and
Levine (1988) issue a call for a richer contextual understanding of the impact of culture on the
study of ego identity formation. However, more recent findings support the notion that ethnic
identity is a legitimate and stand-alone domain of ego identity and further suggest that a strong
personal sense of ethnic identity may be associated with positive outcomes (Schwartz et al.,
2010; Umaña-Taylor et al., 2009).
17
`
Family life course development framework. The family life course development
framework has been criticized in important ways (Hutchinson, 2011). The underlying theory may
have difficulty in accommodating heterogeneity within a society. The usefulness of the theory
may be questioned with regard to societies where individuals’ lives follow highly diverse
patterns, such as the United States. This is because the theory presupposes the existence of
patterns in development over the life course. However, life course development theory does not
presuppose that individuals’ lives follow any one particular pattern. Therefore the theory is
flexible enough to be adapted to describe the life course across many cultures. A second, perhaps
more serious criticism of life course development theory is that it does not address the
connections between individuals and large institutions. This may be particularly relevant for
Hispanic participants, whose ethnic identity is potentially being molded in part by institutions
such as the United States Census Bureau. However, although this theory taken alone may not
emphasize these connections, the present study explicitly does accomplish this emphasis without
particular impediment from the family life course developmental framework.
Relations between Family Differentiation and Ego Identity
Regardless of whether the concept is represented as one dimensional or two dimensional,
and regardless of whether the one or two dimensions are termed cohesion, enmeshment,
differentiation, emotional autonomy, attachment, or independence, it is both theoretically and
empirically clear that relational dynamics within the family system of an older adolescent or
young adult are predictive of identity status (Adams et al., 1987; Arseth et al., 2009; Campbell et
al., 1984; Faber et al., 2003; Jackson et al., 1990; Marcia, 1966; Watson & Protinsky, 1988).
Further, a review of the literature and the chronologically progressive nature of
conclusions about the underlying constructs have revealed a compelling case for assuming a twodimensional model of family differentiation. This is underscored by the explicit testing of a twodimensional model (Barber & Buehler, 1996; Manzi et al., 2006).
It is also established that the concept of Marcian identity status is of practical value in
part because the Marcian identity statuses can be predictive of measurably positive outcomes and
prosocial behaviors, especially in older adolescents or younger adults (Arseth et al., 2009;
Crocetti et al., 2009; and Hardy & Kisling, 2006). It would therefore be worthwhile to pursue a
detailed understanding of the potential effect of family differentiation on psychological wellbeing via Marcian ego identity variables.
18
`
Although there are several Marcian identity status-specific examples of scholarly
attempts to isolate these underlying family relational dynamics in the literature (Adams et al.,
1987; Arseth et al., 2009; Campbell et al., 1984; Faber et al., 2003; Jackson et al. 1990; Marcia,
1966; Mullis et al., 2003; Watson & Protinsky, 1988), none to date have assumed or tested a
specific, two-dimensional model of family differentiation in relation to the Marcian identity
statuses.
Further, despite strong support of the contextual relevance and function of a twodimensional model of family differentiation, there is no discernible trend in the literature of
moving toward any universal, operationalized language to describe the relational dynamics that
may quantifiably explain the evolution of commitment and exploration statuses in older
adolescents and young adults. This is the case despite the Freudian notion of ego identity
evolving as a consequence of relational dynamics within an individual's family of origin (Freud,
1933) and the inherent relational processes embedded in the Eriksonian intimacy vs. isolation
crisis associated with adolescents moving toward young adulthood (Erikson, 1959).
Relations between Family Differentiation and Psychological Well-Being
The present study uses three measures of psychological well-being as outcome
(dependent) variables: life satisfaction, as measured by the Satisfaction with Life Scale (SLS;
Diener et al., 1985); depressive symptoms, as measured by Center for Epidemiological Studies –
Depression Scale (CES- D; Radloff, 1977) and anxiety, as measured by a short form of the StateTrait Anxiety Inventory (STAI-6; Marteau & Bekker, 1992).
These variables were selected for several reasons. First, Manzi et al. (2006) used these
same measures to show that these outcomes are related to identity threat. The present research
instead hypothesizes that these psychological well-being outcomes will be related to identity
formation variables (commitment and exploration). Second, Manzi et al. (2006) also used these
same measures to show that these psychological well-being outcomes were related to family
differentiation (cohesion and enmeshment). The present research hypothesizes the same
relationships.
Manzi et al. (2006) found family differentiation to be related to the particular
psychological well-being variables that are to be examined in this study. Specifically, in the
British sample, cohesion was negatively related to depressive symptoms and positively related to
19
`
life satisfaction. Enmeshment was positively related to both anxiety and depressive symptoms. In
the Italian sample, cohesion was negatively related to both anxiety and depressive
symptoms, and was positively related to life satisfaction. Manzi et al.’s (2006) results show the
impact of identity development and psychological well-being. These results also form the basis
of the hypotheses in this study.
In related findings, Fuhrman and Holmbeck (1995) concluded that the effect of emotional
autonomy on the parent-child relationship varies according to the quality of the relationship.
Emotional autonomy predicted a reduction in adjustment only when the parent-child relationship
could be characterized as warm and low conflict. Emotional autonomy actually predicted higher
adjustment if the parent-child relationship could be considered cold or high conflict. Emotional
autonomy is conceptually opposed to enmeshment. If the impact of emotional autonomy depends
on the quality of the relationship, this impact may also depend on cultural differences, as
hypothesized in this study and as actually found by Manzi et al. (2006).
Relations between Ego Identity and Psychological Well-Being
Manzi et al. (2006) used the same measures of psychological well-being selected for this
study to show that these outcomes are related to identity threat. The present research builds on
this finding by hypothesizing that these psychological well-being outcomes will be related to
identity formation variables (commitment and exploration).
Barber and Beuhler (1996) found that higher enmeshment predicted less desirable
outcomes for a combined depressive symptoms-anxiety variable as well as for withdrawal and
aggression. Regression slopes ranged from .14 to .26. In contrast, higher cohesion predicted
more desirable outcomes for depressive symptoms-anxiety, withdrawal, aggression, and
delinquency. Regression slopes ranged from -.24 to -.30.
Furthermore, researchers used growth mixture modeling (GMM) to show that adolescents
could be categorized into two classes: those whose anxiety began high and increased; and those
whose anxiety began low and declined. Those in the high anxiety class tended to show lower
levels of commitment over time (Crocetti et al., 2009).
More broadly, ego identity status has been found to predict adaptive psychological
functioning and prosocial behaviors in young adults as well as older adolescents. A metaanalysis of fourteen studies found that secure attachment (the most adaptive attachment style) is
more frequently observed among individuals with identity-achieved status than foreclosed or
20
`
diffuse status (Arseth et al., 2009). Securely attached individuals “are comfortable depending on
others and having others depend on them.” (Arseth et al., 2009, p. 5). Another meta-analysis of
twenty-one studies found that individuals with high-exploration identity statuses also
experienced greater levels of intimacy (Arseth et al., 2009).
Hardy and Kisling (2006) found three measures of prosocial behavior to correlate with
ego identity status. “Community service, prosocial activities, and prosocial behavioral
tendencies” each correlated positively with identity achievement (p. 363). Each of these three
measures also correlated negatively with identity diffusion. Moreover, the higher commitment
statuses (achievement and foreclosure) were associated with higher levels of affection between
parent and child (Campbell et al., 1984).
The Role of Culture and Ethnicity
Côté and Levine (1988) called for a richer contextual understanding of the impact of
culture and ethnicity on the study of ego identity formation. This study represents a relatively
early appeal for identity scholars to acknowledge the perils of an inherent culture of reference
bias across the literature. It also argues that cultural and ethnic literacy incorporated into identity
research as well as related intervention programs could ultimately benefit clinicians and clients.
Importance of Comparing Ethnic Groups
One goal of this study is the preliminary isolation of a possible culture of reference
within the family differentiation and identity literature. The United States is different from
Europe in its governmental structure: States have some independence from the Federal
government, and the United Kingdom of Great Britain and Northern Ireland [U.K.] and Italy are
a part of the European Union, but states in the U.S. are not separate, sovereign, or formal
countries with unique, established cultural roots, norms, and expectations extending back in
excess of a thousand years (Côté & Levine, 1988). The modern civic and legal nationality of a
Floridian, for example, is identical to that of a Georgian or a Californian. For this reason, it
would not be sufficient or justifiable to simply compare participants from one state versus
another state in the present study, because to do so would not capture the contrast in cultural
norms and expectations as described by Manzi et al. (2006). The present study will accomplish
this by focusing on participants who identify themselves as non-Hispanic Whites and Hispanics
of any race.
21
`
Field (2002) describes the melting pot perspective as optimistic, but as Glazer (1993) and
Vertovec (2010) have asserted, one important criticism of the melting pot perspective is that it
was inherently ethnocentric in prioritizing the ultimately British Colonial cultural values of the
early U.S. power elite over the values of residents from other backgrounds. Another criticism of
the melting pot perspective is that even when its proponents did place significant value on the
contributions of early European immigrants, they still often ignored the contributions of other
minority groups in the U.S., notably African-Americans (Baldwin, 1955; DuBois, 1925; and
Glazer, 1993).
Glazer (1993) explained that as professionals continued to explore the value of the
melting pot perspective, eventually the trend in the literature moved away from a debate about
the ideal balancing point on an assimilation continuum ranging from amalgamation to
acculturation, or Americanization (e.g., Hollinger, 2006; Vought, 2004), aided by the metaphor
of the melting pot, to advocacy for cultural pluralism (Kallen, 1956), or the idea that it is
worthwhile for persons of varied ethnic backgrounds to hold on to their unique cultural
identities.
Hollinger (2006) is a post-multiculturalist scholar that has articulated the concept of postethnicity and argued that U.S. citizens should ideally retain and receive respect for expressing a
cultural identity but, in case of immediate conflict, prioritize national, non-ethnic, ideologicallybased norms and expectations where possible in an effort to preserve peaceful societal progress.
Hollinger's post-ethnicity concept is open to criticism about the risk of unwitting, myopically
inherent ethnocentrism in defining those supposedly non-ethnic, ideologically-based norms and
expectations.
As Baldwin (1955) noted, whether one is exploring the benefits of a more assimilationist
or more multiculturalist worldview, separateness can be forcefully imposed on the individual by
the reaction of others to physiological differences in the power elite as distinct from
ethnominorities, despite any intellectual or individual desire to transcend ethnicity. Thus, as
Vertovec (2010) explained, begins another, more modern post-multiculturalist cycle of debate
between the optimistic idealism originally present in the melting pot perspective and the risks of
excessive emphasis on cultural separateness.
It is in this post-multiculturalist context, then, that the present study examined the value
of family differentiation and identity development in furthering a broad academic understanding
22
`
of cross-cultural relations. This will be accomplished by the explicit and contextual justification
of the two comparison groups. This represents an important methodological advance, but it is in
line with contemporary professional standards because in the context of understanding cultural
and ethnic diversity, identity scholars have given greater consideration of postmulticulturalism. For example, Glazer (1993) credited the Eriksonian concept of ego identity
with giving diversity scholars and historians a language with which to explore the experience of
the individual as a separate entity and as a member of a group. In addition, Glazer (1993)
attributed the notion of modern “American identity and its avoidance of explicit ethnic
reference” (p. 125) in part to Erikson's own experiences with Americanization as something that
Erikson acknowledged helped him in his initial conceptualization of ego identity.
Ethnoracial Groups in the Research
The present study featured a comparison of two ethnoracial groups based on the racial
and ethnic categories used in the 2010 United States Census: non-Hispanic whites and Hispanics
of any race (Humes et al. 2011). This methodological choice stands in contrast to comparing
groups from different countries or states (Manzi et al., 2006). The goal of the present study is to
capture potential differences between a societal culture of reference and a culture, ethnicity,
ethnominority, ethnoracial group, or as Emerson (Field, 2002) described it, a U.S. group that is
not yet fully assimilated into ethnocultural majority norms.
Rationale for Focus on White Non-Hispanic Participants
British-Americans are not necessarily representative of ethnocultural U.S. majority or
reference group norms. Vertovec (2010) explained that current immigration and emigration
patterns represent unprecedented, accelerated movement and freedom of movement of different
national and ethnic groups within Western, democratic societies, including in the U.K. Vertovec
(2010) argued that despite this “superdiversity” (p. 86) in the U.K., a power elite-driven culture
of reference and even frank xenophobia currently culturally co-exists with accepted societal
sentiments to the contrary, possibly as a result of the influence of the multiculturalist paradigm.
Thus, it is not currently reasonable to assume anything about the culture of self-identified
British-Americans, in part because they may be part of the new British superdiversity trend and
not steeped in British majority cultural norms, and in part because British citizens are themselves
undergoing a tremendous shift in the definition and implications of majority ethnocultural norms
and roles. Given that the U.S. culture of reference is arguably based on British Colonial majority
23
`
ethnocultural norms from nearly two and a half centuries ago (Glazer, 1993), the argument must
be made that any modern U.S. resident that identifies specifically as British-American might be
more a product of any segment of present-day U.K. culture than a U.S. culture of reference.
Given Manzi, et. al's (2006) decision to study Italian participants and highlight Italian
culture, Italian-Americans in the context of this study represents a convenient, and by no means
exhaustive, example of the ambiguous concept of whiteness in the U.S.: In isolating participants
for the present study that identify per USCB guidelines as white alone, and not ethnically
Hispanic, as a potential representative for the U.S. culture of reference, it is important to consider
the field of academic whiteness studies (e.g., Jacobson, 1999; Kolchin, 2002; Taylor, 2004),
which, in a post-multiculturalist context, implies that U.S. whiteness is more of a complex,
socially constructed privilege status as opposed to a product of an individual's genetic lineage.
For example, the USCB (Humes et al., 2011) allowed census respondents to select one of
five races, or to write in their race in their own words. The USCB then categorized respondents
that asserted via free-response that they were Italian, Italian-American, or even EuropeanAmerican as white. Assuming that the USCB's categorization policies represent one institutional,
bureaucratic, standardized definition of whiteness in the U.S. today, Italian-Americans can in an
important respect be considered white in the present-day U.S. and, therefore, per the assumptions
of whiteness studies (e.g., Jacobson, 1999; Kolchin, 2002; Taylor, 2004), afforded a privileged
social status and by extension the right to further co-construct the meaning of the term white in
the U.S. moving forward, making Italian-Americans an active part of the culture of reference in
the U.S. Of course, this argument could also be made for Irish-American census respondents,
Polish-American census respondents, or any number of groups that the USCB (Humes et al.,
2011) has decided to categorize as white, even if, like Italian-Americans, those groups were not
historically afforded a socially privileged status in the U.S. (e.g., Behdad, 2005; Guglielmo,
2004).
The legacy of discrimination against Italian immigrants, particularly in the twentieth
century U.S. (e.g., Behdad, 2005; Guglielmo, 2004), is one example of the difficulties in cultural
assimilation. Today, people in the U.S. who identify with their Italian heritage have the option of
celebrating their heritage openly and, if they desire to do so, simultaneously working to
incorporate their ideas of what it means to be Italian-American into the notion of what it means
to be part of U.S. culture or white in general. In this way, modern self-identified Italian-
24
`
Americans, like other USCB-defined white groups, are afforded an opportunity to retain and
refine a personal sense of cultural separateness while maintaining an identity as a part of a U.S.
culture of reference and a unified national culture. This process has led scholars to assert that the
Italian- American identity is unique, distinct from the Italian or Italian immigrant identity, and a
recognizable contributor to the overall modern national identity and culture of reference (e.g.,
Guglielmo, 2004; Guglielmo & Salerno, 2003).
The present study featured U.S. participants of different ethnicities, and as such may be
in a unique position to offer insight into that balance. In the context of an evolving postmulticulturalist perspective, this study will focus on all participants that identify or are classified
per USCB (2011) standards as white alone in an effort to capture members of a present-day,
actively co-constructed culture of reference that are afforded the opportunity to keep, define, and
celebrate their separateness while simultaneously contributing to the idea of whiteness in the
U.S.
Rationale for Focus on Hispanic Participants
There are five USCB categories that allow respondents to identify themselves as nonwhite: a) black or African-American; b) American Indian or Alaska Native; c) Asian, or d)
Native Hawaiian or other Pacific Islander groups, or e) some other race (Humes et al., 1999,
p. 2). Researchers have offered recent confirmation that ethnic identity achievement is not only
reasonably conceived of as a dimension of ego identity, it is reliably measurable and even
specifically associated with positive outcomes (e.g. Schwartz, Zamboanga, Weisskirch, & Wang,
2010; Umaña-Taylor, Gonzales-Backen, & Guimond, 2009).
Yet in recent empirical studies, researchers have also found that participants that selfidentify as Hispanic, as opposed to other racial or ethnocultural groups, stand out in the realm of
ethnocultural identity exploration and commitment. For example, Juang and Syed (2010)
compared multiple ethnocultural groups and found that participants that did not identify as white
were more likely than white participants to engage in cultural socialization, or social activities
that increased a sense of belonging to an ethnocultural group, with their families, and that this
type of socialization was particularly likely to predict an increased a sense of ethnocultural
identity for white participants, who, ostensibly, as members of a privileged social group, may not
otherwise have been prompted to think about their cultural uniqueness. However, Juang and
25
`
Syed (2010) also found that self-identified Hispanic participants were the only group that
differed significantly from white participants in that Hispanic participants demonstrate
significantly more cultural identity exploration than their white-identified counterparts.
Schwartz, Mason, Pantin, and Szapocznik (2009) analyzed longitudinal data of Hispanic
early to middle adolescents with a variety of national heritage backgrounds and found that for
participants that tended to experience increasing levels of identity confusion as they moved
toward later adolescence and identity development, family functioning initially predicted identity
confusion at earlier ages, but at older ages, increased levels of identity confusion in turn
predicted lower levels of family functioning. Further, Syed and Azimitia (2008) found that selfidentified Latino/Chicano and self-identified Asian participants tended to have a stronger sense
of ethnocultural identity achievement as opposed to white and mixed-ethnicity participants, and
indicated in open-ended narratives that they had spent more time contemplating an overall sense
of ethnocultural identity in the first place.
Taken in concert, recent research has indicated that that there is a unique component to
the intergenerational transmission of ideas about U.S. Hispanic self-identification (e.g., Juang &
Syed, 2010), that there is evidence that, for Hispanic families specifically, identity confusion of a
mid-adolescent can predict a decrease in family functioning (Schwartz et al., 2009), and that
people that specifically identify as Hispanic in the U.S. have spent more time thinking about
what their ethnocultural identity means to them than their white, mixed-ethnicity, and other nonwhite identified counterparts (Juang & Syed, 2010; Syed & Azimitia, 2008). This is in line with
Phinney's (1990) assertion that ethnocultural identity develops concurrently with ego identity
and, if an individual decides that her ethnocultural identity is an important part of her overall
identity, it becomes an integral part of her overall ego identity, ostensibly due to an intense
period of ethnocultural identity exploration that preceded ethnocultural identity achievement.
Phinney (1990) asserted that this is evident because people with a developed sense
of ethnocultural identity often demonstrate this by raising the issue of ethnocultural identity
themselves, often in everyday social interactions.
Consistent with the post-multiculturalist perspective, evidence suggests that a bicultural
worldview may be beneficial (e.g., Chae & Foley, 2010). In addition, evidence suggests that
ethnocultural ego identity achievement is associated with positive outcomes (e.g., Schwartz et
al., 2010; Umaña-Taylor et al., 2009). Nevertheless, people in the U.S. who identify as Hispanic
26
`
or partly Hispanic tend to struggle with the relevance of taking on a Hispanic identity (e.g., Syed
& Azimitia, 2008), or else expend more effort exploring their ethnocultural identity than other
people in the U.S. (e.g., Juang & Syed, 2010; Schwartz et al., 2009; Syed & Azimitia, 2008).
One possible explanation for this pattern of recently published results is that the
Hispanic-identified people in the U.S. occupy a unique position in the current U.S. ethnoracial
spectrum. Although a Hispanic person in the U.S. may consider themselves white-alone, or some
other race, or a mix of races, they may or may not think of their status as Hispanic, nonHispanic, or mixed Hispanic and non-Hispanic as relevant to their race or ethnocultural identity.
This may create a kind of ethnocultural ego identity cognitive dissonance between an
individual’s cultural views about race that have been handed down through generations to a
Hispanic-identified person and the conflicting messages that U.S. society is currently sending
them about identifying as ethnically Hispanic.
As Boyd (2010) explains, the term Hispanic describes a broad range of people with
varied national, racial, and ethnic origins, most notably outside of the U.S. Further, the person
that is described as Hispanic in the U.S. may have simply identified as white or black in her
home country, or her grandparent's or great-grandparent's home country. Boyd's (2010)
assertions about the strong cultural legacy of European Colonialism, most notably Spanish
Colonialism in Latin America, combined with the legacy of brutality, oppression, and slavery
suffered by persons of African and/or indigenous heritage in the many parts of the world touched
by Spanish Colonialism are an excellent example of why the specific Hispanic racial identity of a
Hispanic-identified person in the U.S. is essential to understanding the rich, multinationally
relevant U.S. ethnocultural Hispanic identity.
In attempting to grasp the possible impact of racial constructs and discrimination and
brutality against Hispanic persons that were not labeled as white before immigrating to the
United States, whether in this generation or a previous generation, it may be helpful to consider
the broader implications of DuBois' (1925) suggestion that:
the race problem is the other side of the labor problem; and the black man's burden is the
white man's burden. At least it will be of absorbing interest, to step within these distant
world shadows, and, looking backward, to view the European and white American labor
problem from this wide perspective, remembering always that empire is the heavy hand
of capital abroad. (p. 386)
27
`
Although the USCB (Humes et al., 2011) allows participants to identify as any race and
either Hispanic or non-Hispanic, conforming respondents are not permitted to identify as, for
example, white-alone without deciding to describe themselves as Hispanic or non-Hispanic. This
is because the USCB defines the term Hispanic as an ethnicity and not a race. This is, in fact, the
only ethnic category that the USCB employed in its most recent census, making a respondent's
self- identification as either a Hispanic or a non-Hispanic a unique experience.
Further, respondents that opted to ignore the five main race categories as well as the
Hispanic ethnicity prompt and simply write in as a free response that they were Hispanic or
Mexican, Mexican-American, Chicano, Puerto Rican, Cuban, Latino, Spanish, or any other preidentified Hispanic group (p. 2), were in effect labeled as non-white, or “some other race” (p. 2),
as opposed to, for example, a person that might have written in that they were Italian-American
and subsequently categorized by the USCB as white. This speaks to a nation that is at odds with
the possibly strong expectation of a Hispanic person in the U.S. that they have a right to describe
themselves as white, black, or any other race without consideration of their status as ethnically
Hispanic or not Hispanic.
People in the U.S. who identify as both Hispanic and any race are essentially being asked
to choose between those two identities in cases where doing so does not necessarily fit with their
conception of ethnocultural identity. Thus, the USCB's decision to recognize Hispanic vs. nonHispanic as a sole ethnic category places respondents that identify as Hispanic in a uniquely
ambiguous category that may in many cases be very much at odds with ethnocultural aspects of a
respondent's established sense of identity and may arguably represent and even partially create a
presently transitional, unique, and ambiguous U.S. ethnoracial status.
Although the discussion here is not empirical proof of the mechanisms that facilitate a
unique ethnocultural status for self-identified Hispanic persons in the U.S., or even proof of the
unique general ambiguity of the Hispanic ethnoracial status, it does lend strong and sufficient
literature support of the methodological choice to target participants that identify as Hispanic.
The goal of the present study is to isolate participants that are less likely to represent an ethnic or
national culture of reference, and self-identified Hispanic participants in the U.S. are empirically,
bureaucratically, and philosophically more likely to capture that dynamic, arguably even more so
than established minority groups, perhaps as a direct result of their currently ambiguous
ethnoracial status.
28
`
Research Participants
Participants for this study were students at one of two large state universities in the
southeastern United States within the age range 18-26. Marcia (1980) posited that a large amount
of ego identity development takes place for older adolescents that ranged from the ages of 18-21,
consistent with the Eriksonian identity vs. role confusion crisis. Marcia (1989) also asserted that
shifts in Eriksonian ego identity crisis stages took place after an interval of intrapsychic
destructuring and crisis resolution. This is in line with the family life course development
framework concept of family stages.
However, it has also been theoretically posited and empirically supported that U.S.
women may be more likely to commence work on the Eriksonian identity vs. role confusion
crisis only after they have resolved the intimacy vs. isolation crisis (e.g., Campbell et al., 1984;
Grotevant, 1983; Schiedel & Marcia, 1985; Mullis et al., 2003), or may possibly work through
both crises concurrently (e.g., Archer & Waterman, 1994).
Moreover, based on the family life course development framework, the status of college
student in itself may represent a transitional stage and further encourage ego identity crisis
development, resolution, and periods of intrapsychic destructuring (Erikson, 1959) as the U.S.
students, like their counterparts in the study by Manzi et al. (2006), prepare for an adult life after
building a foundation for it in college in their late adolescence and/or early adulthood, although it
should be noted that this does not reflect the adulthood preparation experiences of all U.S.
adolescents.
Finally, there are ethnocultural considerations. For example, Manzi et al. (2006) reported
that 71.6 % of British participants in their study planned on living away from home in the near
future and, by contrast, only 8.5% of their Italian counterparts had the same intention (p.
679). There may be a cognate dynamic in Hispanic vs. white participants, and dynamics may
differ in older adolescents vs. young adults. Marcia (1980) asserted that persons ranging in age
from 21-24 may have resolved the identity vs. role confusion Eriksonian crisis. Therefore,
allowing participants' ages to range from 18-26, or in other words, to target older adolescents and
young adults, should sufficiently allow the present study to capture and feature an empirical
exploration of the relevance of all of these factors.
29
`
Summary Descriptions of Variables
Based on the previous research outlined in the literature review, the present study uses a
two-dimensional assessment of family differentiation. Unlike previous research (e.g., Manzi et
al., 2006), the present study examines the effect of family differentiation on psychological wellbeing, both directly and indirectly via identity formation. Moreover, two separate ethnic groups
were used as samples for this study. The variables examined in this research are described below.
Each of these variables and their theoretical underpinnings has been discussed in the preceding
sections of this chapter.
Family Differentiation
Cohesion. Cohesion is the extent to which family members share supportive emotional
bonds. As reflected in the items on the Cohesion subscale of the Colorado Self-report of Family
Functioning Inventory, cohesive families share a “feeling of togetherness,” “help and support
one another,” and “get along well with each other.” (Bloom, 1985, p. 232, Table 3). Members of
cohesive families are able to rely on one another for emotional support. A lack of cohesiveness
results in a sense of isolation which may lead to poor psychological outcomes.
Enmeshment. Enmeshment is the extent to which the family restricts the autonomy of
its members. As reflected in the items on the Enmeshment subscale of the Colorado Self-report
of Family Functioning Inventory, members of enmeshed families feel “pressured to spend most
free time together.” They feel “guilty if they wanted to spend some time alone.” (Bloom, 1985,
p. 234, Table 3). Members of enmeshed families may have difficulty establishing independent
identities, since individual activities and interests must be subordinated to family goals and
values.
Ego Identity
Commitment. Commitment is an individual’s ability to consistently identify with those
values and roles one has chosen to represent oneself and to make new choices that reflect a stable
and consistent pattern of self-selected identity (Marcia, 1966). An individual with a high level of
commitment in a particular domain will pursue relatively consistent goals within that domain.
For example, a person with a high level of commitment within the professional domain will
pursue positions that advance a particular career path, rather than applying for miscellaneous
jobs selected haphazardly. As reflected in the Ego Identity Process Questionnaire, individuals
with a high level of commitment in the professional domain “are unlikely to alter [their]
30
`
vocational goals.” Those with a high level of commitment in the family domain have “firmly
held views concerning [their] role in [their] family.” (Balistreri et al., 1995, p. 192).
Exploration. Exploration is the degree to which an individual leads an examined life, or
makes personal choices with an eye toward the active construction of the self. An individual with
a high level of exploration in a particular domain would have put time and effort into considering
various options within that domain, rather than selecting the most obvious option presented by
their family or other circumstances. As reflected in the Ego Identity Process Questionnaire,
individuals with a high level of exploration in the ideological domain “have considered different
political views thoughtfully.” (Balistreri et al., 1995, p. 192).
Psychological Well-Being
Life satisfaction. Life satisfaction is the cognitive component of subjective well-being,
in contrast with positive affect and negative affect, which are the emotional components (Diener
et al., 1985). Satisfaction with life involves an individual’s evaluation of their own life as whole
by their own personal standards. It is “a global assessment of a person's quality of life according
to his chosen criteria” (Shin & Johnson, 1978, p. 478, cited in Diener et al., 1985, p. 71). The
Satisfaction with Life Scale is “narrowly focused to assess global life satisfaction and does not
tap related constructs such as positive affect...” (Diener et al., 1985, p. 71).
Depressive symptoms. Depressive symptoms are behaviors, emotions, and thoughts
which are associated with a clinical diagnosis of a depressive disorder, but which are also
observed to a lesser extent in individuals who do not meet diagnostic criteria for such a disorder.
As reflected in the CES-D, depressive symptoms include crying spells, feeling sad, poor appetite,
and feelings of inferiority. “The CES-D scale is designed to measure depressive symptomatology
in the general population (Radloff, 1977, p. 385).” This scale is contrasted with other depressive
symptoms scales which are designed for clinical settings. Individuals with a score of 16 or higher
on the CES-D may be at risk for clinical depression. (Lewinsohn, Seeley, Roberts, & Allen,
1997).
State anxiety. State anxiety is an attribute of the individual's current emotional state, in
contrast to trait anxiety, which is an individual's general tendency toward anxiety across a wide
range of situations. As measured by the short-form of the state scale of the STAI-6, state anxiety
involves feeling “tense,” “upset,” and “worried.” (Marteau & Bekker, 1992).
31
`
Review of Relationships among Variables in the Research
This section reviews the predicted relationships among the variables in this study, and
highlights the key theoretical and empirical arguments for these predictions. Relevant theoretical
and empirical arguments have been discussed in detail in earlier sections of this chapter.
Cohesion and commitment. Cohesion was hypothesized to have a positive direct effect
on commitment across ethnoracial groups. Identity theorists, as early as Freud (1933) and
including Marcia (1989), have asserted the impact of family dynamics on identity development.
A cohesive family theoretically provides adolescents with a sense of emotional security that
allows them to feel confident in committing to particular identity choices. Empirical work has
borne this out. Campbell et al. (1984) found that high commitment (identity-achieved or
foreclosed) adolescents had more affectionate relationships with their parents than adolescents in
lower commitment statuses; affectionate relationships are conceptually similar to family
cohesion. Watson and Protinsky (1988) found that cohesion predicted identity achievement and
foreclosure. Both articles asserted a one-dimensional model of family differentiation; this study
will be the first to use the two-dimensional model of family differentiation to empirically
examine the possibility that cohesion influences commitment. This will represent a significant
advance in the literature if the two-dimensional model is supported, as it was in Manzi et al.
(2006).
Cohesion and exploration. Cohesion was hypothesized to have a direct positive effect
on exploration, the same across ethnic groups. Watson and Protinsky (1988) found that high
cohesion levels were predictive of identity achievement status, which is consistent with high
exploration. These authors used a one-dimensional model of family differentiation; the present
research will be the first to use the two-dimensional model to empirically examine the
possibilities that cohesion influences exploration.
Enmeshment and commitment. Enmeshment was hypothesized to have a direct positive
effect on commitment across ethnic groups. Enmeshment may foster commitment by pushing
adolescents and young adults to make identity choices in accordance with family norms. Watson
and Protinsky (1988) found that high enmeshment levels were predictive of identity achievement
and identity foreclosure statuses which are consistent with high commitment. Since these authors
were using the one-dimensional model, high enmeshment levels are equivalent to high cohesion
32
`
levels. The present research will be the first to use the two-dimensional model to empirically
examine the possibility that cohesion influences exploration.
Enmeshment and exploration. For non-Hispanic whites, enmeshment was hypothesized
to have a direct negative effect on exploration. Enmeshment limits the autonomy of adolescents
and young adults; therefore it was expected to reduce opportunities for exploration. Consistent
with this notion, Campbell et al. (1984) found that high-exploration adolescents (identityachieved or moratorium) had greater perceived independence from their parents, corresponding
to lower enmeshment levels, compared to their foreclosed or diffused counterparts. Again, these
authors used a one-dimensional model of family differentiation; the present study will be the first
to use the two-dimensional model to empirically examine the possibility that cohesion influences
exploration.
However, for Hispanics, the effect of enmeshment on exploration was hypothesized to be
less negative (meaning a smaller negative effect, zero effect, or a positive effect) than for whites.
This is because, for Hispanics, enmeshment was expected to be less distinct from cohesion
(consistent with an oblique model of family differentiation) than it is for white non-Hispanics,
perhaps due the value of autonomy in cultural context (e.g., Fuhrman & Holmbeck, 1995; Chun
& MacDermid, 1997; Manzi et al., 2006).
Commitment and psychological well-being. Commitment was hypothesized to have
positive direct effects on each of the three psychological well-being outcomes: anxiety,
depressive symptoms, and life satisfaction. These effects were expected to be the same across
ethnoracial groups. These predictions are based on the premise that identity commitment
provides individuals a sense of purpose (e.g., Crocetti et al., 2009) that enhances psychological
well-being in general, which may be manifest in a wide range of outcome variables.
Commitment was expected to reduce anxiety because it reduces uncertainty about life
pursuits. Commitment was expected to reduce depressive symptoms because a sense of purpose
reduces feelings of helplessness. Finally, commitment was expected to increase life satisfaction
because it implies that the individual is pursuing goals in accordance with the standards he or she
has firmly accepted. With regard to anxiety, there is empirical support for the prediction: in a
large longitudinal study, Crocetti et al., (2009) found that increases in commitment predicted
decreases in anxiety. These particular psychological well-being outcomes have been selected in
part because they are of general interest: depressive symptoms and anxiety are related to
33
`
important clinical diagnoses and are probably the most widely studied affective states;
satisfaction with life is one of the most highly general psychological well-being outcomes. In
addition, the psychological well-being outcomes outcomes have been selected because they are
the same outcomes examined in Manzi et al. (2006), allowing for comparability between the
present study and the study on which it is modeled.
Exploration and psychological well-being. For white non-Hispanic participants,
exploration was hypothesized to have a positive effect on each of the psychological well-being
outcomes: anxiety, depressive symptoms, and life satisfaction. Empirical evidence supports the
positive
effects of exploration on a wide array of measures of adaptive functioning (e.g., Arseth et al.,
2009; Crocetti, et. al, 2009; Hardy & Kisling, 2006). However, none of these studies considered
cultural differences; and all of them included a large majority of participants described as white
or European-American. For Hispanic participants, exploration was hypothesized to have less
positive effect (meaning a smaller positive effect, no effect, or a negative effect) compared to
white non-Hispanic participants. It was expected that due the culturally relevant value of
autonomy, exploration will be less influential outside of the contextual culture of reference (e.g.,
McAdoo, 1981; Watson & Protinsky, 1988).
Family differentiation and psychological well-being. The present study posits that
family differentiation, conceptualized in terms of both cohesion and enmeshment, directly
predicts three facets of psychological well-being. The preceding subsections of this study have
detailed how psychological well-being will be conceptualized in terms of life satisfaction; level
of depressive symptoms; and state anxiety, as measured by the satisfaction with life scale
(Diener et al., 1985); the CES-D scale (Radloff, 1977); and the short-form of the state scale of
the STAI-6 (Marteau & Bekker, 1992). Empirically, each of these predictions is directly
supported (e.g., Manzi et al. 2006). Theoretically, family differentiation was hypothesized to
directly predict these facets of psychological well-being because ego identity status is known to
predict a variety prosocial behaviors and adaptive psychological functioning in older adolescents
and young adults (e.g., Arseth et al., 2009; Crocetti et al., 2009; and Hardy & Kisling, 2006), and
identity theorists have long held that a better understanding of the relational and cultural
dynamics underlying the ego identity development process could only serve to strengthen the
34
`
predictive and practical value of the Marcian ego identity status paradigm (e.g., Marcia, 1989;
Côté & Levine, 1988).
The direct effects of cohesion on each of the three psychological well-being outcomes
(anxiety, depressive symptoms, life satisfaction) were expected to be beneficial, and the same
across ethnic groups. These predictions are based on the premise that cohesiveness in a family
system provides individuals a sense of stability that enhances adaptive functioning in general,
which may be manifest in a wide range of psychological outcome variables. These predictions
are supported theoretically (e.g., Minuchin, 1974; Olson, 1982; Scabini, 1985) as well as
empirically (e.g., Barber and Buehler; Campbell et al., 1984; Manzi et al., 2006; and Watson &
Protinsky, 1988). For white non-Hispanic participants, enmeshment was expected to have a
detrimental direct effect on each of the psychological well-being outcomes: anxiety,
depressive symptoms, and life satisfaction. These predictions are also supported theoretically
(e.g., Minuchin, 1974; Olson, 1982; Scabini, 1985) as well as empirically (e.g., Barber and
Buehler; Campbell et al., 1984; Manzi, et al., 2006; and Watson & Protinsky, 1988).
However, for Hispanic participants, enmeshment was expected to have a less
detrimental effect (meaning a smaller detrimental effect, no effect, or a beneficial effect)
compared to white non-Hispanic participants. For Hispanic participants, it was expected that
enmeshment will be less distinct from cohesion (consistent with an oblique model of family
differentiation) than it is for white non-Hispanics, possibly due to the value of autonomy in
cultural context (e.g., Fuhrman & Holmbeck, 1995; Chun & MacDermid, 1997; Manzi et al.,
2006).
The facets of psychological well-being used in the present study are justified in the
methods chapter of this study on their own merits as established, valid, and reliable ways to
predict practical aspects of adaptive psychological functioning. Furthermore, the measurement of
these variables will yield greater generalizability of results. This study represents only the second
effort in the literature (Manzi, et. al., 2006) to determine the extent to which the two dimensional
model family differentiation directly predicts adaptive psychological functioning and the first
effort to investigate the indirect effect of family differentiation on psychological functioning via
the ego identity development process.
Family differentiation, identity formation, and psychological well-being. The twodimensional model of family differentiation will be examined in this study. The two-
35
`
dimensional model of family differentiation is based upon strong empirical evidence from
scholars that have explicitly tested this model (e.g., Barber & Buehler, 1996; Manzi et al., 2006).
This study will be the first to use the two-dimensional model to empirically examine the
possibility that family differentiation will predict psychological well- being (anxiety, depressive
symptoms, and life satisfaction) both directly and indirectly via ego identity variables. Although
the call for greater evidence of the relational underpinnings of the identity development process
in older adolescents and young adults is as long-standing (e.g., Freud, 1933; Marcia, 1989) as the
scholarly call for a richer contextual understanding of the practical import of complex cultural
literacy on the part of identity scholars, (e.g., Côté and Levine, 1988), no scholars to date have
been able to isolate these underlying relational dynamics. Further, among those who have
conducted research citing this goal explicitly, (e.g., Adams et al., 1987; Arseth et al., 2009;
Campbell et al., 1984; Faber et al., 2003; Jackson et al. 1990; Marcia, 1966; Mullis et al., 2003;
Watson & Protinsky, 1988), none to date have assumed or tested a specific, two-dimensional
model of family differentiation in relation to the Marcian identity statuses.
The present analysis involves considering the direct relationships between family
differentiation and identity development; identity development and various aspects of
psychological well-being; and family differentiation and various aspects of psychological wellbeing because the hypothesized relationships are empirically and theoretically supported. This
approach justifies the question of whether the effect of family differentiation on practical
psychological outcomes may occur, at least in part, indirectly via the ego identity development
process. Comparing two carefully selected ethnic groups will contribute to understanding the
implications of how different ethnic groups reflect the adulthood preparation experiences of all
U.S. adolescents. Finally, there are ethnocultural considerations. For example, Manzi et al.
(2006) reported that 71.6 % of British participants in their study planned on living away from
home in the near future and, by contrast, only 8.5% of their Italian counterparts had the same
intention (p. 679). There may be an analogous dynamic in Hispanic vs. white participants, and
dynamics may differ in late adolescents vs. young adults. Marcia (1980) asserted that persons
ranging in age from 21-24 may have resolved the identity vs. role confusion Eriksonian crisis.
Therefore, allowing participants' ages to range from 18-26, i.e., to target late adolescents and
young adults, should allow the present study to capture and feature an empirical exploration of
the relevance of all of these factors.
36
`
CHAPTER THREE
METHOD
This chapter includes a discussion of the sample, measures, procedures and analyses used
for this study. Statistical predictions for the structural equation modeling (SEM) analysis and
ethical considerations for this study also are presented.
Sampling and Participant Recruitment
This study used purposive, nonprobability sampling of two state universities in the
southeastern United States: the Florida State University (FSU), located in Tallahassee, Florida;
and the Florida International University (FIU), located in Miami, Florida.
Selection of Recruitment Sites
These universities were selected based on the demographic profiles of their enrolled
students to yield adequate numbers of participants in each of the ethnoracial groups examined in
the study. FSU (2012) reports total enrollment of 72.1% white; 13.6% Hispanic; 9.2% black,
2.9% Asian; American Indian 0.4%; two or more races 1.8%. FIU (2012) reports total enrollment
of 61% Hispanic; 15% white non-Hispanic; 13% black; 4% Asian or Pacific Islander; 7% other
minority groups. Thus, FSU’s student body has self-identified as majority white (but also has a
significant minority of Hispanics) and FIU’s student body has self-identified as majority
Hispanic (but also has a significant minority of white non-Hispanics). Therefore, it was expected
that recruitment from these two universities would yield sufficient numbers of participants in
each of the ethnoracial groups to be studied, and indeed this was the case.
Age Range of Participants
Participants for this study were students within the age range 18-26. As discussed in
Chapter Two, this age range is theoretically grounded and is intended to capture participants
whose ages roughly correspond to two Eriksonian crisis states (Erikson, 1959). For that reason,
references to those two subgroups are reflected by the use of the phrases older adolescent and
young adult, in order to distinguish between these two groups. From a family life course
perspective, this entire group might be referred to as emerging adults (e.g., Schwartz et al., 2009;
Schwartz et al., 2010). However, this is potentially subject to debate because, as detailed in
Chapter 2, it remains unclear in what order people confront the Eriksonian crisis states.
37
`
Participant Recruitment
Upon approval from the institutional review boards (IRBs) of both universities,
(Appendices E & F) participants were recruited by reviewing public records to determine the
names and email addresses of university instructors teaching classes at both universities.
Instructors of these classes were asked for their permission to visit their classes in order to
request that their students voluntarily participate in the study. Students in these classes received a
brief oral explanation of the purpose of study and then were given the opportunity to receive a
copy of the informed consent form (Appendix A and B for FSU and FIU, respectively) and a
copy of the participant packet (Appendix C). In order to prevent oversampling, data were
collected in phases, reviewing the number of study relevant participant packets after each phase,
and commencing a new phase of data collection only if necessary.
Ethnoracial Criteria for Participants
Participants were asked (anonymously) to self-identify their race and ethnicity using
United States Census Bureau (Humes et al., 2011) categories. The 2010 Census prompted all
respondents to self-identify as one or more of five races: a) white; b) black or African-American;
c) American Indian or Alaska Native; d) Asian; or e) Native Hawaiian or other Pacific Islander
groups. In addition, respondents were permitted to identify as “some other race” (p. 2).
Respondents were also given the option to write in a race in a free-response fashion; however,
free responses were subsequently categorized by the USCB as falling into one of these six
categories. The present study followed the same procedure.
The USCB allowed respondents to describe a multiple-race background, and categorized
the number of respondents in any particular category as either that category alone or that
category alone-or-in-combination (p.4). In addition to race, the USCB prompted respondents to
choose between two ethnic categories: a) Hispanic; and b) non-Hispanic. Ethnoracial categories
were defined by the combination of racial and ethnic categories. All data analyzed in this study
were from participants that identified as either 1) white alone and non-Hispanic or 2) Hispanic,
regardless of race. Data collected from other participants were not analyzed.
Descriptive Statistics and Frequencies
Combined sample. There were 723 total participants, (individuals that submitted a
signed informed consent form as well as a participant packet that was at least partially
completed): 326 participants were from Florida State University (FSU) in North Florida and 397
38
`
were from Florida International University (FIU) in South Florida. Of these, 131 participant
packets were excluded from the analysis because the participant did not identify themselves as a
member of an ethnoracial category targeted by this study. Of the total participants, 12.03% were
black or African-American; 3.6% were Asian; 2.07% selected multiple races; and 0.41%
identified as some other race. In addition, 11 participants (1.52%) were excluded based on their
age; 4 (0.55%) because they did not report their race, age, or both; and 14 (1.94%) because they
left more than 40% of items blank on at least one measurement scale.
The resulting final sample size of valid cases used for all analyses was N=563 (See Table
1). Of these, 305 (54.17%) packets were collected at FIU; and 258 (45.83%) packets were
collected at FSU. Of the total valid cases, 364 (64.65%) participants were female and 199
(35.35%) were male. Of the total valid cases, 330 (58.61%) participants identified as Hispanic
and 233 (41.39%) identified as non-Hispanic. According to MacCallum, Browne, and Sugawara
(1996), this sample size yields power exceeding .95 for the SEM analyses conducted.
By design, participants who reported any race other than white, or who did not report a
race, were only included in the analysis if they identified as Hispanic. Of the 563 total valid
cases, 481 (85.44%) participants identified as white; 9 (1.60%) as black or African–American; 4
(0.71%) as Native Hawaiian or Pacific Islander; 2 (0.36%) as Asian; 51 (9.06%) as “some other
race;” and 2 (0.36%) selected more than one race. Fourteen participants (2.49%) did not report a
race.
Among the 563 total valid cases, the mean age was 19.96 (SD = 1.97); the median was
19; and the mode (27.89%) was 18. The minimum age for inclusion in the analysis was 18; and
the maximum was 26. Most of the participants in the sample were under 21 years old (68.21%);
93.78% were under 24 years old.
Of the 563 total valid cases, 175 (31.08%) participants listed majors in professional
fields (e.g., law, medicine, or nursing); 121 (21.49%) in psychology; 105 (18.65%) in other
social science fields; 69 (12.26%) in science, technology, engineering, or mathematics; 50
(8.88%) in humanities or arts; 17 (3.02%) had multiple majors that included more than one of
these categories; 23 (4.09%) indicated they had not declared a major; and 3 (0.53%) did not
report their major.
Hispanic subsample. Of the 330 valid cases in the Hispanic subsample, 271 (82.12%)
participant packets were collected at FIU and 59 (17.88%) were collected at FSU. Of the valid
39
`
cases in the Hispanic subsample, 214 (64.85%) participants were female and 116 (35.15%) were
male. Of the Hispanic valid cases, three quarters (248; 75.15%) participants identified as white; 9
(2.73%) as black or African–American; 4 (1.21%) as Native Hawaiian or Pacific Islander; 2
(0.61%) as Asian; 51 (15.45%) as “some other race;” and 2 (0.61%) selected multiple race
categories. Fourteen participants (4.24%) identified as Hispanic, but did not report a race.
Among the 330 valid cases in the Hispanic subsample, the mean participant age was
20.19 (SD = 2.12); the median was 20; and the mode (26.97%) was 18. The minimum age was
18 and the maximum was 26. Most participants (62.73%) were under the age of 21 and 91.82%
of participants were under 24.
Non-Hispanic subsample. Of the 233 valid cases in the non-Hispanic subsample, 199
(85.14%) participant packets were collected at FSU in North Florida and 34 (14.59%) were
collected at FIU in South Florida. Of the valid cases in the non-Hispanic subsample, 150
(64.38%) participants were female and 83 (35.62%) were male. All of the 233 non-Hispanic
valid cases identified as white; this was inevitable due to the study inclusion criteria.
Among the 233 valid cases in the non-Hispanic subsample, the mean age was 19.69 (SD
= 1.76); the median was 19; and the mode (29.18%) was 18. The minimum age was 18 and the
maximum was 26. Three quarters of participants in this subsample were under 21 (75.97%);
96.57% were under 24.
40
`
Table 1
Demographic Variables by Ethnoracial Subsample
Hispanic
Non-Hispanic
Combined
______________________________________________________________________________
Site
FSU
59
(17.88%)
199
(85.14%)
258
(45.83%)
Gender
Race
FIU
271
(82.2%)
34
(14.59%)
305
(54.17%)
Female
214
(64.85%)
150
(64.38%)
364
(64.65%)
Male
116
(35.15%)
83
(35.62%)
199
(35.35%)
White
248
(75.15%)
233
(100%)
481
(85.45%)
Black
9
(2.73%)
0
(0%)
9
(2.73%)
NHPI
4
(1.21%)
0
(0%)
4
(1.21%)
Asian
2
(0.61%)
0
(0%)
2
(0.61%)
Other
51
(15.45%)
0
(0%)
51
(15.45%)
Multiple
2
(0.61%)
0
(0%)
2
(0.61%)
Unreported
14
(4.24%)
0
(0%)
14
(4.24%)
Age (Mean)
20.19 (S.D. = 2.12) 19.69 (S.D. = 1.76) 19.96 (S.D. = 1.97)
______________________________________________________________________________
Note. Data are frequencies except where specified. Percentages shown are percentages of the
ethnoracial group represented in the demographic category at hand. NHPI = Native Hawaiian or
Pacific Islander. By design, 100% of the non-Hispanic subsample identified as white.
Measures
The measurement instruments used for this study are found in the participant packet
(Appendix C). Following the recommendation of Dillman (2007), spacing and similarity of
visual elements are used in the participant packet to enhance readability.
Family Differentiation
Following Barber & Buehler (1996) and Manzi et al. (2006), cohesion and enmeshment
were measured by the appropriate subscales of the Colorado Self- Report of Family Functioning
Inventory (Bloom, 1985). Each subscale contains five 6-point items ranging from 1 (strongly
disagree) to 6 (strongly agree). Manzi et al. (2006) reported acceptable or good reliability for
both scales (cohesion: Italy = .78; U.K.. = .88; enmeshment: Italy = .74; U.K., = .65). An
example of an item measuring cohesion is “family members really helped and supported one
41
`
another.” An example of an item measuring enmeshment is “family members found it hard to get
away from each other.”
Ego Identity
Commitment and exploration were measured by the Ego Identity Process Questionnaire
(EIPQ; Balistreri, et. al, 1995). The present study employed this instrument instead of another
instrument used widely by identity scholars, the Extended Objective Measure of Ego Identity
Status II (Bennion & Adams, 1986), because the latter measures each of the Marcian identity
statuses as a distinct continuous variable, rather than treating commitment and exploration as
variables.
The EIPQ includes 32 items. There are eight domains within the EIPQ, and four items
per domain. Two items per domain measure commitment; the other two items measure
exploration. The eight domains measured by the EIPQ may be broken down into four ideological
domains and four interpersonal domains. The four ideological domains are politics, religion,
occupation, and values. The four interpersonal domains are friendships, dating, gender roles,
and family (Schwartz, 2004).
Each item on the EIPQ presents the respondent with 6 Likert-type points, ranging from
strongly disagree to strongly agree. Twenty of these items are positively worded; twelve are
negatively worded (Balistreri et al., 1995). The EIPQ is typically used to assign individuals to
Marcian identity statuses (achievement, foreclosure, diffusion, and moratorium) using median
splits on the commitment and exploration subscales. The present research does not involve
assignment of identity statuses. Instead, this study treats commitment and exploration as
continuous variables in all SEM analyses.
Schwartz (2004) provided construct validation of the EIPQ by relating it to a measure of
psychological agency, the Multi-Measure Agentic Personality Scale (Côté, 1997). Balistreri et
al., (1995) reported an acceptable EIPQ reliability of α = .75 for commitment and α = .76 for
exploration. Balistreri et al., (1995) also reported one week test-retest reliability of r = .90,
p < 0.01; and a correlation between exploration and commitment of r = - 0.35, p < 0.05.
Psychological Well-Being
Three measures comprise psychological well-being. Life satisfaction was measured by
the Satisfaction with Life Scale (Diener et al., 1985), consisting of five 7-point items. This
42
`
measure is intended, as Manzi et al. (2006) describe it, “to assess global life satisfaction” (p.
679). Manzi et al. (2006) reported good reliability for this measure (Italy, α = .83; U.K., α =.80).
Depression was measured by the CES-D (Radloff, 1977), an instrument that requires
participants to indicate how frequently they have experienced each of twenty depressive
symptoms on a 4-point scale. In Manzi et al., both Italian and British samples had reliability of α
= .89.
Anxiety about an upcoming life stage transition was measured by a short form of the
State-Trait Anxiety Inventory (STAI-6), an instrument developed by Marteau & Bekker (1992).
This measure presents respondents with words that participants are asked to rate for how well or
poorly their mood has been described with regard to their upcoming life stage transition. It has
six 4-point items. Manzi et al. (2006) report good reliability for this measure (Italy, α = .85;
U.K., α =.83).
Reliability Statistics
Alpha reliability statistics were computed for all latent variables. These computations
were performed for the full sample of valid cases (n=563) and also for each of the ethnoracial
groups separately (Hispanic, n=330; non-Hispanic, n=233). Alpha statistics ranged from
adequate to very good for all scales. Differences between the ethnoracial subsamples were small.
Alpha-if-item deleted (AID) scores did not indicate any poorly performing items on any of the
scales.
For the full sample of valid cases, alpha reliability scores were: cohesion, α = .88;
enmeshment, α =.76; commitment, α = .75; exploration, α = 70; life satisfaction, α =.83;
depressive symptoms, α =.89; anxiety, α =.84. For the non-Hispanic subsample, alpha values
were: cohesion, α =.88; enmeshment, α = .77; commitment, α = .73; exploration, α = 73; life
satisfaction, α =.84; depressive symptoms, α = .88; anxiety, α = .84. For the Hispanic
subsample, alpha values were: cohesion, α = .88; enmeshment, α =.73; commitment, α = .75;
exploration, α = .67; life satisfaction, α = .83; depressive symptoms, α = .89; anxiety, α = .84.
Participant Demographic Information
Participants were asked to indicate their gender (male or female); race (white; black or
African-American; American Indian & Alaskan Native; Asian; Native Hawaiian and other
Pacific Islander; or some other race); ethnicity (indicated by a yes or no response to the question
“Are you of Hispanic, Latino, or Spanish origin?”); and age (free response).
43
`
Analysis
The data analysis for this study featured multiple-groups structural equation modeling
(SEM). This study used a full measurement model (including all of the items from the
measurement instruments in the model) rather than a simple path analysis. Scale items were
treated as observed variables whereas the family differentiation, ego identity, and psychological
well-being variables were treated as latent variables. Following the recommendation of Kenny
(2011b), the raw data were inputted rather than analyzing the correlation matrices, because of the
likelihood that groups would differ in their variances. The SEM analysis tested the hypotheses
that cohesion and enmeshment have indirect effects on psychological well-being via
commitment and exploration. Some of the paths were set equal across groups, whereas others
were allowed to vary, in accordance with the study hypotheses detailed later. The path
coefficients and model fit were examined to determine whether the study hypotheses were
supported. Finally, the model was adjusted on the basis of modification indices and the two
models will be compared in terms of fit, parsimony, and theoretical implications. SEM analyses
were conducted using MPlus software (Muthén & Muthén, 2012). Descriptive statistics,
frequencies, and alpha reliability statistics were computed using SPSS.
SEM was used to examine relationships among family differentiation, ego identity, and
psychological outcomes for the two ethnic groups. The diagram representing the structural paths
for the SEM Models is found in Figure 1. The measurement model has been omitted from the
figure for clarity. All measurement items are found in the participant packet (Appendix C). The
model indicates effects of both cohesion and enmeshment on each of the ego identity variables,
commitment and exploration. In turn, commitment and exploration were hypothesized to have
direct effects on each of the psychological outcome variables.
Four structural equation models were analyzed; each was evaluated using the test of close
fit, which tests the null hypothesis that the root mean square error of approximation (RMSEA) is
less than an arbitrary cut-off value (McCullum, Browne, & Sugawara, 1996). The cut-off value
selected for this study was 0.8; values between 0.5 and 0.8 are considered to be a “fair fit” by
these authors.
44
`
Assumptions of Structural Equation Modeling (SEM)
Structural equation modeling (SEM) relies on several assumptions which, if violated,
may undermine the conclusions drawn from the analysis (Kline, 2012). Below, the major
assumptions of SEM are discussed in relation to the data used in this study.
Model specification. SEM assumes that the model is properly specified. This means that
all relevant variables are included in the model. It is never possible to identify all variables that
may affect the outcomes, which renders this assumption is unrealistic. For practical purposes, it
is crucial that the variables included be theoretically justified (Kline, 2012). Chapter 2 explains
the theoretical justification for the variables included in the SEM models of this study. Therefore,
this assumption is met to the extent that could be reasonably expected.
Exogenous variables measured without error. SEM assumes that the exogenous
variables (in this case cohesion and enmeshment) are measured without error: that their
reliability scores are equal to one. This is an inherently unrealistic assumption in the
measurement of latent constructs. However, biases are generally only slight when reliability
scores are greater than .90 (Kline, 2012). The reliability of cohesion was close to this value (α =
.88 for each of the ethnoracial subsamples); however the reliability of enmeshment was α = .76
and α = .73 for the non-Hispanic and Hispanic subsamples respectively. This suggests the
possibility of bias caused by measurement error with regard to enmeshment.
Completeness of data set. If a raw data file is analyzed in SEM (as it is in this study)
there must be no missing data. This assumption was met. As discussed later in this chapter,
participants with excessive missing data were excluded from the study; missing data for other
participants was substituted using the mean-person method (Roth, Switzer, & Switzer, 1999).
Multivariate normality. Endogenous variables must have a multivariate normal joint
distribution. The endogenous variables in this study were Commitment, Exploration, Depression,
Anxiety, and Life Satisfaction. The following steps were taken in order to test the assumption of
multivariate normality (Kline, 2012). Univariate kurtosis and skewness statistics were examined
for each of the endogenous variables. The most kurtotic variable was Depression (1.28, S.E. =
.21). Kurtosis was small for all other variables, with no absolute value exceeding .26. Depression
was also the most skewed variable (1.13, S.E. = .10), with no other absolute value exceeding .77.
These values show only small to moderate violations of multivariate normality (Kline, 1998).
Bivariate scatterplots were examined for each of the ten possible pairings of the endogenous
45
`
variables. None of these scatterplots suggested a non-linear relationship, and none appeared to fit
a curvilinear trend line. Further, none of the scatterplots revealed the cone shape characteristic of
heteroscedasticity. Mardia’s coefficient was calculated using the PAST statistical software
(Hammer, Harper, & Ryan, 2001) to assess the multivariate skewness and kurtosis of the
endogenous variables. Mardia’s coefficient of skewness was 2.633 (247.1, d.f. = 35), below the
threshold of 3 recommended by Kline (1998) for excessive skewness. However, Mardia’s
coefficient of kurtosis was 38.14 (4.453, d.f. = 35), exceeding the threshold of 10 recommended
by Kline (1998).
Overall assessment of assumptions. For the most part, the assumptions of SEM were
met in the current data set. However, the moderately large kurtosis and skewness values for
Depression may be problematic for the interpretation of the results regarding Depression. The
large Mardia’s coefficient for kurtosis may be problematic for the model as a whole. Violations
of multivariate normality tend to result in underestimates of standard errors, increasing the risk
of Type I errors in the testing of path coefficients. On the other hand, these violations also tend to
inflate the chi-square statistic, making the model seem like a worse fit than it actually is.
However, under some circumstances these violations may instead make a model seem like a
better fit than it actually is, and there is no way to determine the direction of the bias. (Kline,
2012). The reliability values for enmeshment may not be large enough to guard against bias
arising from measurement error. This suggests the need for improved measurement of this
construct in future research.
Structural Equation Model Identification
SEM requires that both the structural model and the measurement model be
overidentified in order to obtain meaningful results. Based on the guidelines provided by Kenny
(2011a), the measurement model used in this study is overidentified because for each latent
variable there are at least two observed variables. The structural model is overidentified because,
for each pair of latent variables, no more than one of the following is the case: a) one directly
causes the other; b) the two variables have a correlated disturbance; or c) the two are correlated
exogenous variables.
Structural Equation Models
Model 1: Configural model. A configural model (Kenny, 2011b) was used to determine
whether the hypothesized model would be reasonable with a minimum of constraints. The
46
`
configural model applies the same factor and path structure to the Hispanic and non-Hispanic
groups. In accordance with the hypotheses of the study, this model included direct effects of each
of cohesion and enmeshment on each of commitment and exploration. The model also included
direct effects of each of commitment and exploration on each of anxiety, depressive symptoms,
and life satisfaction.
The model also included indirect effects of each of cohesion and enmeshment on each of
anxiety, depressive symptoms, and life satisfaction levels, via commitment and exploration.
Cohesion and enmeshment were allowed to co-vary. In the configural model, all parameters
were allowed to vary across groups, except that, in both groups, factor means were fixed at zero
and scale variances are fixed at one. These fixed parameters were necessary in order for the
model to be identified. (Kenny, 2011a).
Model 2: Measurement invariance model. Since the configural model was acceptable,
a measurement invariance model was used to determine whether it was reasonable to assume that
the measurement instruments functioned in the same manner for Hispanics and non-Hispanics
(Kenny, 2011b). This model used the same factor and path structure as the configural model, but
held the factor loadings and intercepts constant across groups.
Model 3: Hypothesized model. Model 3 was designed to test the substantive hypotheses
of this study. The path structure was identical to that in Models 1 and 2. Factor loadings and
intercepts were held constant across groups, consistent with the measurement invariance found in
Model 2. Structural paths, however, were either set equal or allowed to vary across ethnoracial
groups in accordance with the statistical predictions of the study.
The paths from cohesion to exploration and to commitment were held constant across
groups (Predictions 1 & 2), as was the path from enmeshment to commitment (Prediction 3);
however, the path from enmeshment to exploration was allowed to vary across groups
(Predictions 4 & 5). The paths from commitment to each of the psychological outcomes
(Prediction 6) were held constant across groups. However, the paths from exploration to each of
the psychological outcomes (Predictions 7 & 8) were allowed to vary across groups.
The indirect paths from cohesion to each of the psychological outcomes via commitment
were held constant across groups (Prediction 9). However, the paths from cohesion to each of the
psychological outcomes via exploration were allowed to vary (Predictions 10 & 11). Similarly,
the indirect paths from enmeshment to each of the psychological outcomes via commitment were
47
`
held constant across groups (Prediction 12); whereas the analogous paths via exploration were
allowed to vary (Predictions 13 & 14). The direct paths from cohesion to the psychological
outcomes will be held constant across groups (Prediction 15); the direct paths from enmeshment
to these outcomes will be allowed to vary across groups (Predictions 16 & 17).
Model 4: Modified model. Model 4 was identical to Model 3, except that additional
paths were estimated on the basis of modification indices computed from Model 3. Any paths
with modification index (M.I.) of 10 or above in Model 3 were added for Model 4. These were:
1. The direct effect of cohesion on life satisfaction, which was allowed to vary across
ethnoracial groups.
2. 57 inter-item covariances in the non-Hispanic group.
3. The following covariances in the non-Hispanic group:
a. Exploration with commitment.
b. Satisfaction with commitment.
4. 80 inter-item covariances in the Hispanic group.
5. Four intercepts in the non-Hispanic group.
6. Four intercepts in the Hispanic group.
Power Analysis, Sample Size, and Missing Data
MacCallum, Browne, and Sugawara (1996) provide a framework for determining
statistical power for SEM based on sample size and the degrees of freedom of the model. The
sample of the present study (N = 563, df = 35) provided power > .95 for rejecting the null
hypothesis that RMSEA < .08 in the test of close fit.
Mean-person substitution was used for any missing data points on any of the
measurement scales. This method substitutes for each missing data point the mean value for that
item across all study participants. A Monte Carlo simulation study found this method to be less
biased than other methods for addressing missing data (Roth, Switzer, & Switzer, 1999). These
results held for situations where up to 40% of items were missing for each scale. Therefore, any
participant for whom more than 40% of the items were missing for any measurement scale was
excluded from the analysis. Finally, any participant who did not report their race or did not report
their ethnicity was excluded from the analysis. This was necessary because the design required
that participants be divided into ethnoracial groups for the analysis.
48
`
Figure 1. Structural equation modeling (SEM) diagram reflecting the statistical predictions of the
study. Cohesion and enmeshment are allowed to co-vary. Cohesion and enmeshment (family
differentiation) directly predict commitment and exploration (identity status components);
commitment and exploration (identity status components) directly predict anxiety, depressive
symptoms, and life satisfaction levels. Cohesion and enmeshment (family differentiation)
directly affect anxiety, depressive symptoms, and life satisfaction levels, and indirectly affect
these variables via commitment and exploration. (Direct paths from cohesion and enmeshment to
the psychological variables are represented by dashed lines for clarity.) In order to test the study
hypotheses regarding differences between the two ethnoracial groups (white non-Hispanic;
Hispanic), the paths from cohesion to each of the ego identity variables, as well as the paths from
commitment to each of the psychological well-being variables, were set equal across groups
(Model 3). The paths from enmeshment to each of the ego identity variables, and from
exploration to each of the psychological variables, were allowed to vary across groups.
Statistical Predictions
This section states the statistical predictions tested by the SEM analysis. In all hypotheses
involving psychological well-being variables, life satisfaction was measured by the Satisfaction
with Life Scale (SLS; Diener et al., 1985); depressive symptoms was measured by the Center for
Epidemiological Studies – Depression Scale (CES-D; Radloff, 1977) and anxiety was measured
by a short form of the State-Trait Anxiety Inventory (STAI-6; Marteau & Bekker, 1992).
49
`
1. Cohesion, as measured by the Cohesion subscale of the Colorado Self-Report of Family
Functioning Inventory (Bloom, 1985) will have a positive direct effect on commitment,
as measured by the Commitment subscale of the Ego Identity Process Questionnaire
(Balistreri, Busch-Rossnagel, & Geisinger, 1995). This effect will be the same across
ethnoracial groups.
2. Cohesion, as measured by the Cohesion subscale of the Colorado Self-Report of Family
Functioning Inventory (Bloom, 1985), will have a positive direct effect on exploration, as
measured by the Exploration subscale of the Ego Identity Process Questionnaire
(Balistreri, et. al, 1995). This effect will be the same across ethnoracial groups.
3. Enmeshment, as measured by the Enmeshment subscale of the Colorado Self-Report of
Family Functioning Inventory (Bloom, 1985) will have a positive direct effect on
commitment, as measured by the Commitment subscale of the Ego Identity Process
Questionnaire (Balistreri, et. al, 1995). This effect will be the same across ethnoracial
groups.
4. For white non-Hispanic participants, enmeshment, as measured by the Enmeshment
subscale of the Colorado Self-Report of Family Functioning Inventory (Bloom, 1995),
will have a negative direct effect on exploration, as measured by the Exploration
subscales of the Ego Identity Process Questionnaire (Balistreri, et. al, 1995).
5. For Hispanic participants, the effect of enmeshment, as measured by the Enmeshment
subscale of the Colorado Self-Report of Family Functioning Inventory (Bloom, 1985), on
exploration, as measured by the Exploration subscale of the Ego Identity Process
Questionnaire (Balistreri, et. al, 1995), will be less negative (i.e., a smaller negative
effect, zero effect, or a positive effect) than for non-Hispanic whites.
6. Commitment, as measured by the Commitment subscale of the Ego Identity Process
Questionnaire (Balistreri, et. al, 1995), will have a positive direct effect on life
satisfaction and negative direct effects on anxiety and depressive symptoms. These
effects will be the same across ethnoracial groups.
7. For white non-Hispanic participants, exploration, as measured by the Exploration
subscale of the Ego Identity Process Questionnaire (Balistreri, et. al, 1995), will have a
positive direct effect on life satisfaction and negative direct effects on anxiety and
depressive symptoms.
50
`
8. For Hispanic participants (as compared to white non-Hispanic participants), the direct
effect of exploration, as measured by the Exploration subscale of the Ego Identity
Process Questionnaire (Balistreri, et. al, 1995), will be less positive on life satisfaction
and less negative on depressive symptoms and anxiety.
9. Cohesion, as measured by the Cohesion subscale of the Colorado Self-Report of Family
Functioning Inventory (Bloom, 1985) will have a positive indirect effect on life
satisfaction and a negative indirect effect on depressive symptoms and anxiety, via
increases in commitment, as measured by the Commitment subscale of the Ego Identity
Process Questionnaire (Balistreri, et. al, 1995). These effects will be the same across
ethnoracial groups.
10. For white non-Hispanic participants, cohesion, as measured by the Cohesion subscale of
the Colorado Self-Report of Family Functioning Inventory (Bloom, 1985), will have a
positive indirect effect on life satisfaction and a negative indirect effect on depressive
symptoms and anxiety, via increases in exploration, as measured by the Exploration
subscale of the Ego Identity Process Questionnaire (Balistreri, et. al, 1995).
11. For Hispanic participants, in comparison to white non-Hispanics, cohesion, as measured
by the Cohesion subscale of the Colorado Self-Report of Family Functioning Inventory
(Bloom, 1985), will have a less positive indirect effect on life satisfaction and a less
negative indirect effect on depressive symptoms and anxiety, via increases in exploration,
as measured by the Exploration subscale of the Ego Identity Process Questionnaire
(Balistreri, et. al, 1995).
12. Enmeshment, as measured by the Enmeshment subscale of the Colorado Self-Report of
Family Functioning Inventory (Bloom, 1985), will have a positive indirect effect on life
satisfaction and a negative indirect effect on depressive symptoms and anxiety, via
increases in commitment, as measured by the Commitment subscale of the Ego Identity
Process Questionnaire (Balistreri, et. al, 1995). This effect will be the same across
ethnoracial groups.
13. For white non-Hispanic participants, enmeshment, as measured by the Enmeshment
subscale of the Colorado Self-Report of Family Functioning Inventory (Bloom, 1985),
will have a negative indirect effect on life satisfaction and a negative indirect effect on
depressive symptoms and anxiety, via reductions in exploration.
51
`
14. For Hispanic participants, in comparison to non-Hispanic white participants,
enmeshment, measured by the Enmeshment subscale of the Colorado Self-Report of
Family Functioning Inventory (Bloom, 1985), will have a less negative indirect effect on
life satisfaction and a less positive indirect effect on depressive symptoms and anxiety,
via reductions in exploration, as measured by the Exploration subscale of the Ego Identity
Process Questionnaire (Balistreri, et. al, 1995).
15. Cohesion, as measured by the Cohesion subscale of the Colorado Self-Report of Family
Functioning Inventory (Bloom, 1985) will have a positive direct effect on life satisfaction
and negative direct effects on anxiety and depressive symptoms. These effects will be the
same across ethnoracial groups.
16. For white non-Hispanic participants, enmeshment, as measured by the Enmeshment
subscale of the Colorado Self-Report of Family Functioning Inventory (Bloom, 1995),
will have a negative direct effect on life satisfaction and positive direct effects on anxiety
and depressive symptoms.
17. For Hispanic participants, in comparison to non-Hispanic whites, enmeshment,
measured by the Enmeshment subscale of the Colorado Self-Report of Family
Functioning Inventory (Bloom, 1985), will have a less negative direct effect on life
satisfaction and less positive direct effects on anxiety and depressive symptoms.
Ethical Considerations
The present study required the participation of human subjects in the collection of survey
data. This research was conducted with the intention of future publication. In accordance with
university rules for both data collection sites, the researcher obtained permission from the
Institutional Review Boards (IRBs) of both universities where participants were recruited (FSU
and FIU), before beginning data collection. Individuals were required to sign an informed
consent form in order to participate in the study. There was one informed consent form for FSU
(Appendix A) and another for FIU (Appendix B); each of these forms was approved by the IRB
of the university for which it was used.
This study required the creation of a new data set, and although this study employed
purposive sampling, the researcher did collect some data from participants outside of the target
populations. The researcher's intention was to collect data from entire university classes, to avoid
any potential undue stress to participants that would have occurred had they been informed their
52
`
data would not be used for this study if they fell outside of the target population demographics
(e.g., age 27 or older; non-targeted racial/ethnic demographic responses). The time and efforts of
these participants will not be wasted; all data were compiled into a larger data set that may be
used for further study. The only people that were asked to refrain from study participation during
data collection are students that are under the age of majority in the state of Florida (age 18).
53
`
CHAPTER FOUR
RESULTS
This chapter reports the results of the present study. The first section reports the results
from structural equation modeling (SEM). Four models are reported: 1) a configural model; 2) a
measurement invariance model; 3) a hypothesized model (based on the statistical predictions of
the study); and 4) a modified model (based on the addition of parameters to the hypothesized
model based on modification indices). The second section evaluates each of the study
hypotheses. The third section reports statistically significant mean differences in the latent
variables by gender and college major, and by data collection site with each ethnoracial group.
This study focused primarily on substantive questions rather than on measurement issues. Factor
loadings for all measurement items are reported in Appendix D (Tables 10-14). The text of the
measurement items is cross-referenced with the item numbers in Appendix D (Tables 5-9).
Structural Equation Models
Model 1: Configural Model
A configural model (Kenny, 2011b) was used to determine whether the hypothesized
model would be reasonable with a minimum of constraints. The configural model applies the
same factor and path structure to the Hispanic and non-Hispanic groups. In accordance with the
hypotheses of the study, this model included direct effects of each of cohesion and enmeshment
on each of commitment and exploration. The model also included direct effects of each of
commitment and exploration on each of anxiety, depressive symptoms, and life satisfaction.
The model also included indirect effects of each of cohesion and enmeshment on each of
anxiety, depressive symptoms, and life satisfaction levels, via commitment and exploration.
Cohesion and enmeshment were allowed to co-vary. In the configural model, all parameters
were allowed to vary across groups, except that, in both groups, factor means were fixed at zero
and scale variances were fixed at one. These fixed parameters are necessary in order for the
model to be identified. (Kenny, 2011a).
The root mean square error of approximation (RMSEA) for the configural model was
0.054 (90% CI = 0.052, 0.055). This allows rejection of the null hypothesis for the test of close
fit, since the confidence interval falls entirely below the cut-off value of 0.8. Thus, Model 1 was
acceptable. The RMSEA is however above the 0.05 level generally considered the cutoff for a
54
`
“good” fit.(MacCullum, Browne, and Sugawara, 1996). The other fit indices for the configural
model were: χ2 = 8908.551, df = 4928, p < .0001; Akaike (AIC) = 119705.685; CFI = 0.707;
SRMR = 0.083.
Model 2: Measurement Invariance Model
Since the configural model was acceptable, a measurement invariance model was used to
determine whether it was reasonable to assume that the measurement instruments functioned in
the same manner (Kenny, 2011b) for Hispanic and non-Hispanic identified participants. This
model used the same factor and path structure as the configural model, but held the factor
loadings and intercepts constant across groups.
RMSEA for the measurement invariance model was 0.054 (90% CI=0.052, 0.056).
Again, this allows rejection of the null hypothesis for the test of close fit; thus Model 2 was also
acceptable, and measurement invariance was supported. The other fit indices for Model 2 were:
χ2 = 9173.753, df = 5058, p < .001; Akaike (AIC) = 119710.888; CFI = 0.697; SRMR = 0.085.
Model 3: Hypothesized Model
Model 3 was designed to test the substantive hypotheses of the research. The path
structure was identical to that in Models 1 and 2. Factor loadings and intercepts were held
constant across groups, consistent with the measurement invariance found in Model 2. Structural
paths, however, were either set equal or allowed to vary across ethnoracial groups in accordance
with the statistical predictions of the study.
The paths from cohesion to exploration and to commitment were held constant across
groups (Predictions 1 & 2), as was the path from enmeshment to commitment (Prediction 3);
however, the path from enmeshment to exploration was allowed to vary across groups
(Predictions 4 & 5). The paths from commitment to each of the psychological outcomes
(Prediction 6) were held constant across groups. However, the paths from exploration to each of
the psychological outcomes (Predictions 7 & 8) were allowed to vary across groups.
The indirect paths from cohesion to each of the psychological outcomes via commitment
were held constant across groups (Prediction 9). However, the paths from cohesion to each of the
psychological outcomes via exploration were allowed to vary (Predictions 10 & 11). Similarly,
the indirect paths from enmeshment to each of the psychological outcomes via commitment were
held constant across groups (Prediction 12); whereas the analogous paths via exploration were
allowed to vary (Predictions 13 & 14). The direct paths from cohesion to the psychological
55
`
outcomes will be held constant across groups (Prediction 15); the direct paths from enmeshment
to these outcomes will be allowed to vary across groups (Predictions 16 & 17).
RMSEA for Model 3 was 0.054 (90% CI = 0.052, 0.055). Once again, this allows
rejection of the null hypothesis for the test of close fit, since the confidence interval falls entirely
below the cutoff value of 0.8. Thus Model 3 was acceptable. The RMSEA was however above
the 0.05 level generally considered the cutoff for a good fit. The other fit indices for Model 3
were: χ2 = 9182.115, df = 5070, p < .001; Akaike (AIC) = 119695.249; CFI = 0.697; SRMR =
0.085.
The statistical significance and path coefficients of structural paths in Model 3 were
examined to evaluate the statistical predictions of the study. Although the model represents a
reasonable fit to the data, the results of the model specification were mixed. Table 2 reports the
standardized and unstandardized path coefficients for Model 3; Figure 2 is a diagram of the
structural paths with the standardized direct effects and covariances for Model 3. The results
indicate a positive direct effect of cohesion on commitment (0.175, S.E. = 0.037, p < .001). This
provides support for Hypothesis 1.
The results indicate the predicted direct effects of commitment on each of the
psychological outcomes: depressive symptoms (-0.128, S.E. = 0.047, p < .007); and anxiety (0.182, S.E. = 0.080, p = 0.022); life satisfaction (.518, S.E. = 0.140, p < .001). These findings
provide support for Prediction 6.
The results indicate the predicted direct effects of cohesion on the psychological
outcomes: depressive symptoms (-0.139, S.E. = 0.023, p < .001); anxiety (-0.151, S.E. = 0.036, p
< .001); and life satisfaction (0.533, S.E. = 0.061, p < .001). These findings provide support for
Prediction 15.
Finally, the results of Model 3 indicate the predicted indirect effects of cohesion on each
of the psychological outcomes via commitment: depressive symptoms (-0.024, S.E. = 0.009, p =
.008); and anxiety (-0.032, S.E. = 0.014, p = 0.022); life satisfaction (0.091, S.E. = 0.024, p <
.001). These findings provide support for Prediction 9.
However, the remaining predictions (2-5, 7, 8, 10-14, 16 & 17) were not supported. The
associated paths were non-significant. In the case of paths that were allowed to vary across
ethnoracial groups, the paths were non-significant for both groups.
56
`
Model 3 explained a statistically significant percentage of the variance in commitment as
well as each of the psychological variables (life satisfaction, depressive symptoms, and anxiety)
for each of the ethnoracial groups (Hispanic and non-Hispanic). The percentage of variance in
exploration explained by the model was not statistically significant. Detailed results for
percentages of variance explained are found in Table 3. The factor loadings for Model 3, along
with their standard errors and p values, are reported in Appendix D. All factor loadings were
statistically significant at the .05 level with the exception of one item on the exploration scale.
Table 2
Statistically Significant Path Coefficients for Model 3: Hypothesized Model.
Direct effect of
Unstd.
Unstd. S.E.
Std.
Std. S.E.
Commitment
0.175
0.037
0.339
0.053
< .001
Depressive
-0.139
0.023
-0.354
0.052
< .001
Anxiety
-0.151
0.036
-0.241
0.053
< .001
Satisfaction
0.533
0.061
0.464
0.048
< .001
p
Cohesion on
______________________________________________________________________________
Commitment on
Depressive
-0.128
0.047
-0.177
0.060
.003
Anxiety
-0.182
0.080
-0.149
0.061
.014
Satisfaction
0.518
0.140
0.241
0.057
< .001
______________________________________________________________________________
Indirect effect of
Cohesion
via Commitment on
Depressive
-0.024
0.009
.008
Anxiety
-0.032
0.014
.022
Satisfaction
0.091
0.024
< .001
______________________________________________________________________________
Note. All coefficients shown were held equal across ethnoracial groups; the values in this table
apply to both groups. Unstd. = Unstandardized. Std. = Standardized.
57
`
Table 3.
Percentage of Variance Explained by Model 3: Hypothesized Model.
Non-Hispanic
Variable
Variance Explained
S.E.
p
Commitment
11.5%
3.3%
< 0.001
Exploration
0.0%
0.5%
n.s.
Satisfaction
34.2%
4.9%
< 0.001
Depressive
18.9%
3.9%
< 0.001
Anxiety
10.8%
3.5%
0.002
______________________________________________________________________________
Hispanic
Commitment
12.7%
3.5%
< 0.001
Exploration
0.6%
1.1%
n.s.
Satisfaction
30.3%
4.1%
< 0.001
Depressive
19.9%
3.7%
< 0.001
Anxiety
14.7%
3.8%
0.001
58
`
Table 4.
Latent Variables Included in the Structural Equation Models: Descriptive Statistics and
Reliability. Non-Hispanic N = 233; Hispanic N = 330
Non-Hispanic
Variable
M
SD
Range
α
Cohesion
4.90
0.99
1.40 – 6.00
.88
Enmeshment
2.12
0.84
1.00 – 5.80
.77
Commitment
4.06
0.64
2.31 – 5.69
.73
Exploration
4.02
0.65
2.13 – 6.00
.73
Satisfaction
5.13
1.24
1.40 – 7.00
.84
Depressive
1.70
0.48
1.00 – 3.47
.88
Anxiety
1.89
0.66
1.00 – 4.00
.84
______________________________________________________________________________
Hispanic
Cohesion
4.91
1.04
1.60 – 6.00
.88
Enmeshment
2.51
0.97
1.00 – 6.00
.73
Commitment
4.18
0.66
2.00 – 5.94
.75
Exploration
4.06
0.60
1.81 – 5.81
.67
Satisfaction
4.88
1.29
1.00 – 7.00
.83
Depressive
1.74
0.51
1.00 – 3.79
.89
Anxiety
1.80
0.65
1.00 – 3.83
.84
59
`
Figure 2. S t ruct u ral e quat i on m odel i ng ( SEM) Model 3: Hypothesized Model, with
standardized parameters. Standardized coefficients are shown for statistically significant
structural parameters. Cohesion and enmeshment are allowed to co-vary. Cohesion and
enmeshment (family differentiation) directly predict commitment and exploration (identity
status components); commitment and exploration directly predict anxiety, depressive
symptoms, and life satisfaction levels. Cohesion and enmeshment directly affect anxiety,
depressive symptoms, and life satisfaction levels, and indirectly affect these variables via
commitment and exploration. (Direct paths from cohesion and enmeshment to the
psychological variables are omitted from the figure for clarity.) In order to test the study
hypotheses regarding differences between the two ethnoracial groups (white nonHispanic; Hispanic), the paths from cohesion to each of the ego identity variables, as well as
the paths from commitment to each of the psychological well-being variables, were set equal
across groups. The paths from enmeshment to each of the ego identity variables, and from
exploration to each of the psychological variables, were allowed to vary across groups.
However, none of the parameters that were allowed to vary across groups were
statistically significant.
Model 4: Modified Model
Model 4 was identical to Model 3, except that additional paths were estimated on the
basis of modification indices computed from Model 3. Any paths with a modification index
(M.I.) of 10 or above in Model 3 were added for Model 4. These were: 1) the covariance of
exploration and commitment in the non-Hispanic group; 2) 53 inter-item covariances in the nonHispanic group; 3) 60 inter-item covariances in the Hispanic group; 4) four item intercepts in the
non-Hispanic group; and 5) four item intercepts in the Hispanic group.
60
`
RMSEA for Model 4 was 0.040 (90% CI = 0.038, 0.042). Yet again, this allows rejection
of the null hypothesis for the test of close fit, since the confidence interval falls entirely below
the cutoff value of 0.8. The confidence interval falls entirely below the .05 cut-off for a good fit.
(MacCullum, Browne, and Suguwara, 1996). The other fit indices for Model 4 were:
χ2 = 7137.268, df = 4954, p < .0001; Akaike (AIC) = 117882.402; CFI = 0.839; SRMR = 0.077.
Although the model fit was improved in Model 4 over Model 3, no theoretical basis was
found for any of the modifications made. Therefore it was decided that Model 3 was more
informative and useful, as well as parsimonious, and Model 4 was rejected.
Research Questions & Hypotheses
This study posed three research questions, and posited a hypothesis for each of them. The
first research question asked how family differentiation (cohesion and enmeshment) would relate
to ego identity status. The corresponding hypothesis predicted that family differentiation would
explain a significant amount of variance in ego identity status.
This hypothesis was partly supported: the results of Model 3 indicate a positive direct
effect of cohesion on commitment (0.175, S.E. = 0.037, p < .0001). However, there was no
significant effect of either family differentiation variable on exploration; nor was there any
significant effect of enmeshment on either identity formation variable.
The second research question asked how family differentiation would relate to
psychological well-being (anxiety, depressive symptoms, and life satisfaction). The
corresponding hypothesis predicted that family differentiation would affect psychological wellbeing both directly and indirectly via ego identity variables. This hypothesis was also partly
supported: the results indicate the predicted direct effects of cohesion on the psychological
outcomes: depressive symptoms (-0.139, S.E. = 0,023, p < .001); anxiety (-0.151, S.E. = 0.036, p
< .001); and life satisfaction (0.533, S.E. = 0.061, p < .001).
Further, the results of Model 3 indicate the predicted indirect effects of cohesion on each
of the psychological outcomes via commitment: depressive symptoms (-0.022, S.E. = 0.008, p =
.006); anxiety (-0.032, S.E. = 0.014, p = 0.022); and life satisfaction (0.091, S.E. = 0.024, p <
.001). However, the predicted indirect effects of enmeshment on the psychological variables
were not observed, nor were the predicted direct effects exploration on the psychological
variables observed.
61
`
The final research question asked whether there would be differences between the
ethnoracial groups studied. The study hypothesized that, for Hispanic participants, higher levels
of enmeshment were predicted to be less associated with diminished psychological well-being.
Similarly, for participants who identify as Hispanic (compared to those who identify as nonHispanic whites), higher levels of exploration were predicted to be less associated with increased
psychological well-being (anxiety, depressive symptoms, and life satisfaction). Thus, the third
hypotheses was not supported in that there were no significant differences found between
ethnoracial groups.
Mean Differences by Gender, College Major, and Data Collection Site
Analysis of variance (ANOVA) was used to examine mean differences in each of the
latent variables (cohesion, enmeshment, commitment, exploration, depressive symptoms,
anxiety, and life satisfaction) by gender and college major for the full sample, and by data
collection site within each ethnoracial group. Differences statistically significant at the .05 level
are reported in this section.
Two variables differed significantly by gender: commitment (F = 5.8, 562, p = .02) and
depressive symptoms (F = 3.82, 562, p = .05). Women reported mean commitment of 4.18,
slightly greater than men’s mean commitment of 4.04. Women reported mean depressive
symptoms of 1.75, slightly greater than men’s mean depressive symptoms of 1.67.
Two variables differed significantly by category of college major: commitment (F = 3.34,
562, p < .01) and life satisfaction (F = 2.57, 562, p = .01). Psychology majors reported the
highest mean level of commitment (4.2), followed by humanities and arts majors (4.19) and
professional majors (4.19). Undeclared majors reported the lowest levels of commitment (3.6).
Humanities and arts majors reported the highest levels of life satisfaction (5.13), followed by
non-psychology social science majors (5.12), and professional majors (5.06). The lowest levels
of life satisfaction were reported by students with multiple majors (3.85).
For non-Hispanic white participants, two variables differed significantly by data
collection site: exploration (F = 8.59, 232, p < .01) and anxiety (F = 4.09, 232, p = .04). Among
this ethnoracial group, FIU students reported higher levels of exploration (4.31) than FSU
students (3.96). On the other hand, FSU students reported higher levels of anxiety (1.93) than
FIU students (1.68). For Hispanic participants, no variables differed significantly by data
collection site.
62
`
CHAPTER FIVE
DISCUSSION
The purpose of this study was to apply a two-dimensional model of family differentiation
to understanding the relational underpinnings of ego identity development. The study also
examined the impact of both family differentiation and ego identity development on
psychological well-being of older adolescent and young adult Hispanic and non-Hispanic white
people in the U.S. This chapter discusses the findings of this study in relation to its theoretical
frameworks, previous research on this topic, and implications for clinical practice. Finally, this
chapter will present limitations of the study and discuss future directions for research.
Review of Notable Findings
The third model of the structural equation analysis explained a significant portion of the
variance in four of the five endogenous variables (Table 3). This was true for both the Hispanic
and the non-Hispanic white subsamples. For the Hispanic subsample, the portions of variance
explained were: commitment, 12.7%; satisfaction, 30.3%; depressive symptoms, 19.9%; and
anxiety, 14.7%. For the non-Hispanic white subsample, the portions of variance explained were:
commitment, 11.5%; life satisfaction, 34.2%; depressive symptoms, 18.9%; anxiety, 10.8%. The
largest portion of variance explained in both ethnoracial groups was for life satisfaction 30.3%
for the Hispanic subsample; 34.2% for the non-Hispanic subsample. Exploration was the only
endogenous variable for which no significant portion of variance was explained by the model.
These large portions of variance explained in variables of such practical importance as life
satisfaction, depressive symptoms, and anxiety underscore the value of the model. The largest
standardized effects in the model were the direct effects of commitment on life satisfaction and
cohesion on life satisfaction.
Theoretical Implications of Findings
This study builds upon Freud’s (1933) proposal that ego identity development was the
product of relational dynamics in the family of origin, and also upon Marcia’s (1989) call for
study of the impact of relational dynamics on ego identity development. There have been some
efforts to explore the link between these relational dynamics and Marcia’s ego identity statuses
(e.g., Adams et al., 1987; Arseth et al., 2009; Campbell et al., 1984; Faber et al., 2003; Jackson et
al. 1990; Marcia, 1966; Mullis et al., 2003; and Watson & Protinsky, 1988). However, this study
63
`
is the first effort to explain the variance in ego identity variables using the two-dimensional
model of family differentiation. This study is also the first to compare the nature and function of
family differentiation processes among Hispanic-identified and white non-Hispanic participants
in the U.S.
Ethnoracial Differences
Manzi et al (2006) reported that enmeshment was predictive of anxiety and depressive
symptoms for British participants, but not predictive of psychological well-being for Italian
participants. The researchers argued that these differences were attributable to cultural
differences. Specifically, they asserted that British youth are socialized to place greater value on
individual autonomy than Italian youth, making enmeshment, or limited individual autonomy,
more problematic for British youth. This study subsequently predicted that, in these respects,
non-Hispanic white youth would resemble British youth, while Hispanic-American youth would
resemble Italian youth. However, the results of this study revealed that enmeshment had no
significant impact on psychological well-being for either of these two U.S. ethnoracial groups.
If Manzi et al. (2006) are correct in arguing that a culture’s level of individualism
determines whether enmeshment will be harmful to psychological well-being, then the results of
this study, contrasted with those of Manzi et al. (2006) suggest that the culture of non-Hispanic
white people in the U.S. is at present less individualistic than British culture. This statement is
subject to several caveats. First, there may be factors other individualism that determine the
relationship between enmeshment and psychological well-being. Second, it is not clear in the
Manzi et al. (2006) study what percentage of their participants would have identified as white, or
what cultural groups may have been represented in their sample. Third, non-Hispanic white
people in the U.S. are themselves a diverse group: they include many groups (such as ItalianAmericans) who, based on their cultural heritage, may have less individualistic attitudes than
people in the U.S. of Anglo heritage, and may therefore experience less adverse consequences
from enmeshment. Further, there may be relevant cultural differences by region within the
United States, or relevant socio-economic differences that were not captured in this study. This
research does not include data on participants’ cultural heritage beyond race and ethnicity (with
the latter being limited to Hispanic vs. non-Hispanic), nor on participants’ regional origin or
regional cultural influences. Finally, as a heterogenous society, the dominant culture of the U.S.
64
`
may be strongly influenced by its British Colonial history, but that clearly does not imply that the
multiple cultures that contribute to the culture of majority are without influence.
Hispanic-identified people in the U.S. are a culturally diverse group. They include people
of dozens of national origins; various levels of acculturation; speakers of English, Spanish, and
Portuguese; recent immigrants and those whose families have lived in the United States for
generations. A large majority of Hispanic-identified people in the U.S. (69%) endorse the view
that “U.S. Hispanics have many different cultures” over the view that “U.S. Hispanics share a
common culture.” (Taylor, Lopez, Martínez, & Velasco, 2012).
The present study did not ask participants to identify their national ancestry, although it is
reasonable to assume that Hispanic participants in the present study were probably
predominantly of Cuban ancestry: the participants in this study are exclusively from two
universities in Florida, yet this study compares Hispanic-identified participants of any race with
White, non-Hispanic-identified participants. The Hispanic-identified participants in this study
therefore are not necessarily representative of the Hispanic population of the United States.
Whereas people of Cuban ancestry are only 3.5% of the Hispanic population nationwide, they
are 46.2% of the total population in Miami-Dade County, where FIU is located. This means that
Cuban-Americans are highly overrepresented among Hispanic-Americans in Miami-Dade
County, making it possible they are overrepresented among the participants in the present study.
In contrast, nearly two thirds (63%) of Hispanic Americans in the United States are of Mexican
ancestry (Ennis, Ríos-Vargas, & Albert, 2011).
Cuban-Americans in Miami-Dade County may be more likely to see themselves as the
culture of reference than Hispanic-Americans elsewhere. Hispanic-Americans in Miami-Dade
may be less likely to encounter routine discrimination, since they are more likely to regularly
interact with persons that to a large extent share are their own ethnicity and culture. The
everyday experience of these Hispanic-identified people in the U.S. may be different in the
following ways: 1) They may live in communities where they are considered a minority group;
2) They may often face discrimination; and 3) They may feel more targeted by anti-immigrant
sentiment in the U.S. and policies to combat illegal immigration when compared with people in
the U.S. of Cuban immigrants. These differing experiences may foster different attitudes among
Cuban-Americans and Mexican-Americans.
65
`
Further, a large majority (85%) of the Hispanic participants in this study identified as
white, compared to just over half (53%) of Hispanic Census respondents nationwide (Humes et
al., 2011). This suggests that the Hispanic participants in the present study may see their ethnic
and cultural identity in a way that is not representative of Hispanics nationwide, and that may be
more consistent with the cultural norms associated with white people in the U.S. in some
important ways.
Further, the literature review of this study revealed that the data from other various other
studies reveals that the predictive value of enmeshment may be culturally bound, and specifically
beneficial or less detrimental to cultures that are not strongly influenced by Western, Anglo, or
ultimately British Colonial norms. For example, in the United States, Watson & Protinsky
(1988) found that enmeshment predicted identity achievement as well as identity foreclosure, or
high commitment statuses, among African-Americans. It is possible that enmeshment may be
beneficial to psychological well-being among some Hispanic-Americans (such as MexicanAmericans) more than others. Additionally, whether or not members of a given demographic see
themselves as a minority may be a relevant factor.
Implications for the Role of Family Differentiation in Ego Identity Theory
Previous research has linked identity achievement to psychological well-being (Arseth et
al., 2009; Hardy & Kisling, 2006). Identity scholars (e.g., Arseth et al., 2009) have frequently
stated or suggested that identity foreclosure is a less desirable outcome than identity
achievement. In contrast to this, the results of this study indicate that commitment, which can
include identity foreclosure, has a large, beneficial impact on psychological outcomes. Higher
levels of commitment were associated with reduced anxiety and depressive symptoms and
increased life satisfaction. Model 3, which tested the substantive hypotheses of the study, found
significant direct effects of commitment on each of the psychological outcomes including
depressive symptoms, anxiety and life satisfaction.
Results show that the largest direct effect of commitment was on life satisfaction
followed by depressive symptoms and anxiety. This suggests that commitment to ego identity
roles is a major determinant of whether older adolescent and young adults are satisfied with their
lives. Since the relevant structural paths were held equal ethnoracial groups, this finding is
equally applicable to both Hispanic and non-Hispanic white participants in the study.
66
`
In contrast, there was no link between exploration and psychological well-being. None
of the direct paths from exploration to psychological well-being were significant, nor were there
significant indirect effects via exploration. It may be that, at least for some cultural groups,
commitment is the key process that makes identity achievement beneficial to psychological wellbeing and that many or all of the same benefits can be obtained from identity foreclosure
(commitment without exploration). Consistent with the present results, Crocetti et al. (2009), in a
large study of Dutch adolescents, found that “whereas commitment was negatively related
to…anxiety, in-depth exploration and reconsideration of commitment were positively related to
adolescent anxiety. These findings were consistent across…gender, age, and ethnic sub-samples
(p. 841).”
Effects of family differentiation on psychological well-being via commitment and
exploration Scholars have provided both theoretical (e.g., Minuchin, 1974; Olson, 1982; Scabini,
1985) and empirical (e.g., Barber and Buehler; Campbell et al., 1984; Manzi et al., 2006; and
Watson & Protinsky, 1988) support for the idea that family cohesion enhances psychological
well-being. However, no previous studies have examined the possible role of commitment or
exploration in this relationship.
The present study supports the idea that cohesion has beneficial effects on psychological
outcomes, in part indirectly via commitment. The results indicated the predicted indirect effects
of cohesion on the psychological outcomes via commitment for depressive symptoms, anxiety
and life satisfaction. In contrast, the present study found no indirect effects of cohesion on
psychological well-being via exploration.
Effect of cohesion on commitment. Scholars have long maintained that ego identity
development was profoundly shaped by the dynamics of the family of origin (Freud, 1933;
Marcia, 1989). Campbell et al. (1984) found that adolescents with high commitment identity
statuses (identity achieved or foreclosed) had more affectionate relationships with their parents
than adolescents in the moratorium or diffusion statuses. Similarly, Watson and Protinsky (1988)
found that higher levels of family cohesion were associated with the identity achievement and
foreclosure statuses. However, no study has previously directly examined the relationship
between cohesion and commitment.
The present study predicted that higher levels of cohesion would be associated with
higher levels of commitment and this prediction was supported. The standardized path
67
`
coefficient from cohesion to commitment was significant and suggests that family cohesion is
one of the main determinants of ego identity commitment. Moreover, the model explained 11.5%
of the variance in cohesion. Theorists have argued that a cohesive family environment gives
adolescents and young adults the stability they need to commit to their identity choices. The
present results provide empirical support for that idea for both ethnoracial groups examined.
Effect of cohesion on exploration. Watson and Protinsky (1988) determined that
identity achievement (which involves high exploration as well as high commitment) was
associated with higher levels of cohesion. However, no previous study had directly explored the
possible link between cohesion and exploration. The present study hypothesized that higher
levels of cohesion would be associated with higher levels of exploration. However, the results
did not bear this out in that the path from cohesion to exploration was non-significant. Thus,
although an emotionally supportive family of origin encourages firm decisions regarding identity
choices, it does not appear to have any bearing on whether adolescents examine their identity
choices carefully. This was true for both Hispanic and non-Hispanic participants, though this
path was allowed to vary across ethnoracial groups.
Effect of enmeshment on commitment. Watson and Protinsky (1988) found that high
enmeshment levels were associated with the identity achievement and identity foreclosure
statuses; the common element of these statuses is a high level of commitment. In keeping with
this finding, the present study hypothesized that high levels of enmeshment would be associated
with high levels of commitment. However, this was not found to be the case in that the path from
enmeshment to commitment was non-significant.
Effect of enmeshment on exploration. Enmeshment restricts the autonomy of
adolescents and young adults; therefore it is expected to limit opportunities for exploration.
Campbell et al. (1984) found that adolescents in high-exploration statuses (identity-achieved or
moratorium) saw themselves as more independent from their parents compared to their
foreclosed or diffused counterparts. Therefore, the present study hypothesized that higher levels
of enmeshment would be associated with lower levels of exploration for non-Hispanic white
participants. However, for Hispanics, the effect of enmeshment on exploration was predicted to
be less negative, because enmeshment was expected to be less distinct from cohesion consistent
with an oblique model of family differentiation (Manzi et al., 2006). This prediction was not
68
`
supported in that paths from enmeshment to exploration were non-significant for both
ethnoracial groups.
Direct Effects of Family Differentiation on Psychological Well-Being
Cohesion was predicted to have beneficial effects on each of the three psychological
well-being outcomes (anxiety, depressive symptoms, and life satisfaction), across ethnoracial
groups. Family cohesion was expected to provide a sense of stability that enhances adaptive
functioning in general, which may be manifest in a wide range of psychological outcome
variables (e.g., Minuchin, 1974; Olson, 1982; Scabini, 1985). Empirical findings have
supported this (e.g., Barber and Buehler; Campbell et al., 1984; Manzi et al., 2006; and Watson
& Protinsky, 1988). The present results are consistent with past findings in this regard. The
results yielded the predicted direct effects of cohesion on the psychological outcomes for
depressive symptoms, anxiety and life satisfaction.
The largest impact of cohesion was on life satisfaction indicating that an increase of one
standard deviation in cohesion is associated with an increase in life satisfaction of nearly one half
standard deviation. The standardized path coefficients of cohesion on depressive symptoms and
anxiety indicated that an increase of one standard deviation in cohesion is associated with
reductions of more than one third of a standard deviation in depressive symptoms and nearly one
quarter standard deviation in anxiety. These are very large effects, and suggest that family
cohesion is a major determinant of psychological well-being among older adolescents and young
adults, independent of its impact on commitment.
For white non-Hispanic participants, enmeshment was predicted to have a detrimental
effect on each of the psychological well-being outcomes: anxiety, depressive symptoms, and life
satisfaction. These predictions have theoretical support (e.g., Minuchin, 1974; Olson, 1982;
Scabini, 1985) and empirical support (e.g., Barber & Buehler; Campbell et al., 1984; Manzi, et
al., 2006; Watson & Protinsky, 1988). However, for Hispanic participants, enmeshment was
expected to have a less detrimental effect, meaning a smaller detrimental effect, no effect, or a
beneficial effect, compared to white non-Hispanic participants. Again, this was because, for
Hispanic participants, it was expected that enmeshment would be less distinct from cohesion,
consistent with an oblique model of family differentiation (Manzi et al., 2006). However, this
was not found to be the case: enmeshment had no effect on psychological well-being for either
ethnoracial group.
69
`
There may be two reasons that the predicted ethnoracial differences were not observed.
First, the participants in this study were not necessarily representative of the Hispanic population
of the United States. Hispanic-identified persons arguably occupy a cultural majority status in
Miami-Dade County, where FIU is located, and the majority of the FIU student body, they may
not be subject to the same sociocultural dynamics as are Hispanics in other parts of the United
States. Second, from a co-constructionist perspective (e.g., Kurtines, 1999) as well as in a postmulticulturalist context (Vertovec, 2010), the increasing percentage of Hispanics in the United
States population may be influencing the norms of the broader culture, reducing differences
between white and Hispanic perspectives.
Relational Clinical Implications of Theoretical and Empirical Findings
There is strong and recent evidence in the literature that identity achievement predicts
positive outcomes (e.g., Arseth et al., 2009; Crocetti, et. al, 2009; Hardy & Kisling, 2006). A
prediction to this effect in an empirical study would be consistent with the theoretically inherent
tendency to presume that the therapeutic fostering of and emphasis on identity achievement is a
valuable goal, and these predictions are borne out in the literature. However, research also
supports the therapeutic value of fostering family cohesion (e.g., Mullis et al., 2003; Protinsky &
Shuts, 1990). Further, recent findings support the value of fostering ego identity commitment,
including both identity foreclosure and identity achievement (Arseth et al., 2009; Crocetti et al.,
2009; and Mullis et al., 2003), as opposed to simply fostering identity exploration, including only
identity moratorium and identity achievement. The results of this study are in line with these
trends in the literature. Further, this study underscores the importance of cohesion in predicting
both commitment and the psychological well-being of older adolescents and young adults.
Marcia (1989) encouraged researchers to explore the complex factors that may ultimately
explain the Marcian identity statuses, and imagined that a successful identity-rooted intervention
would likely offer a clear rationale of the predictors for the statuses themselves, as opposed to
just sorting and attempting to understand clients according to the larger commitment and
exploration conceptual components. This study's contributions to the understanding of this
process could be of direct use to a relationally trained clinician with training in a framework
informed by family systems theory or an understanding of family differentiation. Moreover, this
study's findings might be of use in informing a relationally trained scholar-practitioner in
70
`
answering Marcia's (1989) call to develop an identity-rooted intervention informed by the
relevant underlying relational processes.
The results of this study support the notion that family differentiation is best conceived of
as two dimensions. The results further suggest that facilitating the dimension of family cohesion
in particular may be extremely valuable in relational interventions across cultures. These
findings could contribute to future efforts to develop a manualized, evidence-based approach to
training relational clinicians. This possibility need not be limited to theories that rely heavily on
an understanding of family differentiation: an empirically-based understanding of which
relational dynamics are predictive of psychological well-being outcomes would arguably add
value to the education of any relational clinician in training. This holds especially true given the
apparent cultural literacy value of training relational clinicians to foster family cohesion across
cultures.
Limitations of the Research
Measurement and Comparability of Family Differentiation across the Literature
The assumption of the two-dimensional model of family differentiation in this study is
based on a critical review of the literature as well as explicit empirical evidence (Barber &
Buehler, 1996; Manzi et al., 2006). However, the use of the one-dimensional model of family
differentiation in relevant earlier studies is not always explicit or at least is not measured in a
uniform way, because the two-dimensional model was not identified until just before the turn of
the millennium (Barber & Buehler, 1996; Manzi et al., 2006). This study's assumption of the
two-dimensional model could be further supported by an empirical review of all the measures
used in earlier studies that implicitly or more directly measure what Manzi et al. (2006)
described as the one-dimensional model of family differentiation. Such a study might feature an
exploratory factor analysis with the goal of determining whether pooled measure items from
relevant studies were, in fact, loading onto and therefore measuring cognate concepts. This
would make any endeavor to compare studies that implied or assumed a one-dimensional model
with studies that imply or assume a two-dimensional model much more straightforward.
However, at present, no such study has been conducted.
Limitations of Sample
Exclusion of non-students. The sample from this study is comprised of older
adolescents and young adults who were 18-26 years of age and is intended per the theoretical
71
`
contributions of Marcia (1989) and Erikson (1959) to capture participants in a theoretically
grounded stage and state of intrapsychic destructuring. The participants' status as students
similarly allows for life stage comparability to data from student study participants in other
sociohistorical settings (e.g., Manzi, 2006) per the family life course development framework.
However, although this study may be comparable to the participants in many studies of
students within a similar age range, it does not address the possible differences that might be
found in a population of that is inclusive of similarly aged non-student, even if they were still
going through a theoretically relevant and marked period of intrapsychic destructuring. This
distinction is important because the identity development process of students is a unique
experience in part due to the culture and values of an educational environment (e.g.,
Montgomery & Côtè, 2003) and cannot necessarily be considered comparable to the processes of
similarly aged non-students. This limitation becomes even more noteworthy when considered in
the context of the possibilities for using research relevant to the ego identity process to develop
interventions intended to combat marginalization and oppression, as marginalized and oppressed
persons cannot be assumed to have access to higher education.
Restriction of age range. The restriction of age of the study participants is a limitation in
that it prevents generalization to other age groups. However, the age range of the study was
selected for theoretical reasons, as discussed earlier. It would be fruitful to examine identity
development across the lifespan, and to empirically challenge the notion of stagewise
progression.
Regional bias. Participants in this study are exclusively from two public universities in
Florida, yet this study compares Hispanic-identified participants of any race with white, nonHispanic-identified participants. The Hispanic participants in the present study were not
representative of the Hispanic population of the United States. Whereas people of Cuban
ancestry are only 3.5% of the Hispanic population nationwide, they are 46.2% of the total
population in Miami-Dade County, where FIU is located (USCB, 2010). This means that CubanAmericans are highly overrepresented among Hispanic-Americans in Miami-Dade County, and
probably highly overrepresented among the participants in the present study. Generalization to
the broader Hispanic population of the United States would require a sample representative of
that population.
72
`
Model Fit
RMSEA was the primary model fit statistic examined for this study. RMSEA values
suggested fair to good fit for all models examined, based on the guidelines provided by
MacCallum, Browne, and Sugawara (1996). However, Comparative Fit Index (CFI) values did
not reach the .95 level recommended by Hu and Bentler (1999) to indicate good fit. Future
research should consider models that include additional variables to reduce specification error
and improve model fit.
Mediation in the Structural Equation Models
This study examines the effect of family differentiation (cohesion and enmeshment) on
psychological well-being both directly and indirectly via ego identity variables (commitment and
exploration). This raises the question of whether the relationship between family differentiation
and psychological well-being is mediated by the ego identity variables. According to the
seminal work of Baron and Kenny (1986), a series of three regression analyses may be used to
test for mediation; four criteria must be met to establish a mediational effect. First, when the
mediator (e.g., commitment) is regressed on the independent variable (e.g., cohesion), the
independent variable must be found to have an effect on the mediator. Second, when the
dependent variable (e.g., life satisfaction) is regressed on the independent variable, the
independent variable must be found to have an effect on the dependent variable. Third, when the
dependent variable is regressed on both the independent variable and the mediator, the mediator
must have an effect on the dependent variable. Fourth, the inclusion of the mediator in the
regression equation must reduce the effect of the independent variable on the dependent variable.
In order to claim full mediation, this effect must be completely eliminated by the inclusion of the
mediator.
Baron and Kenny’s (1986) regression procedure was not applied in the present study to
assess potential mediational effects. This is a limitation of the present study, since Baron and
Kenny’s procedure is widely accepted among scholars. However, Brown (1997) argues that
structural equation modeling (SEM) may be used in lieu of Baron and Kenny’s procedure. In the
SEM approach to mediation, Baron and Kenny’s first two criteria are met, respectively, by 1) a
significant path coefficient from an exogenous variable (e.g. cohesion) to a mediator (e.g.,
commitment); and 2) a significant path coefficient from the exogenous variable to an
endogenous variable (e.g., life satisfaction). Baron and Kenny’s third and fourth criteria are
73
`
implicitly met by the presence of significant specific indirect effects in the structural equation
model (Brown, 1997). In this study, these criteria for mediation are met for the effects of
cohesion on depressive symptoms, anxiety, and life satisfaction, via commitment. Specific
indirect effects are reported in Table 2. According to Brown (1997) these effects indicate the
magnitude of the mediational effects in the model. However, the application of Baron and
Kenny’s (1986) procedure would provide greater rigor and more persuasive evidence of
mediational effects.
Future Research
Future Study 1: Confirmatory Factor Analysis (CFA)
This study relates the two-dimensional model of family differentiation to the Marcian ego
identity paradigm. The results challenge long-standing assumptions within the fields of ego
identity theory and family relations theory, raising many questions of theoretical and practical
interest. This study assumes an oblique two-dimensional model of family differentiation
(cohesion and enmeshment are allowed to co-vary), building on the theoretical and empirical
work of Manzi et al. (2006). Those researchers used confirmatory factor analysis (CFA) to
compare one-dimensional, orthogonal, and oblique models of family differentiation for British
and Italian youth.
Future Study 1 would do the same for Hispanic-American and non-Hispanic white youth,
using the data already collected for the present study. This could provide further testing of the
two-dimensional model of family differentiation. For each ethnoracial group, one-dimensional,
orthogonal two-dimensional, and oblique two-dimensional models of family differentiation
would be considered. In all models, factor loadings and error variances would be estimated for
the family differentiation scales (cohesion and enmeshment). The one-dimensional model of
family differentiation assumes that cohesion and enmeshment form a single dimension. This
assumption would be reflected by setting the correlation between these variables equal to 1. The
orthogonal model assumes that cohesion and enmeshment are two uncorrelated dimensions;
therefore the correlation between them would be set equal to 0. The oblique model assumes that
cohesion and enmeshment are correlated, but not perfectly, and therefore the correlation between
them would be allowed to vary freely. For each ethnoracial group separately, the three CFA
models described above would be compared to select the best model based on fit and parsimony,
using the test of close fit (MacCallum, Brown, and Sugawara, 1996).
74
`
Future Study 2: The Role of Gender in Ego Identity Development
This study focuses on race and ethnicity, with only minimal attention to the issue of
gender. The second proposed future study would use the data from the present study to explore
the role of gender in ego identity development. Some theorists have argued that women may be
more likely to work toward a resolution of the Eriksonian identity vs. role confusion crisis only
after they have resolved the intimacy vs. isolation crisis, reversing the order that is expected of
men (Campbell et al., 1984; Grotevant, 1983; Schiedel & Marcia, 1985; Mullis et al., 2003).
Others have suggested that women may address both crises concurrently (Archer & Waterman,
1994; Lacombe & Gay, 1998).
Future Study 2 would use the data from the present study to calculate separate scores for
the interpersonal and ideological domains, corresponding to the identity vs. role confusion and
intimacy vs. isolation crises, respectively. The study would use analysis of variance (ANOVA) to
test whether 1) exploration would be higher for the interpersonal domain for women for the
younger age group; and 2) commitment would be lower for the interpersonal domain for women
for the younger age group.
Future Study 3: Replication with Nationally Representative Sample of Hispanic
Participants
As discussed earlier, Cuban-Americans may be overrepresented among the Hispanic
participants in the present study, relative to Cuban-Americans in the general population of the
United States. Hispanic Americans may vary in their patterns of family differentiation and ego
identity development due to cultural differences associated with national origin. Of particular
interest are Mexican-Americans, since they represent 63% of the Hispanic population in the
United States (USCB, 2010).
Future Study 3 would replicate the methodology of the present study with a sample
representative of the Hispanic-identified population of the United States, which would include an
intentionally representative proportion of Mexican-American participants. Participants could be
most easily recruited from a university located in the southwestern United States, or an area of
the country that has a large Mexican-American population.
Summary and Conclusions
This study is the first scholarly attempt to specifically assume and test the impact of a
two-dimensional model of family differentiation in relation to the Marcian identity statuses. The
75
`
results strongly suggest that family cohesion is a major relational dynamic underlying the
identity development process, and that it is both directly predictive of psychological well-being
and indirectly predictive of psychological well-being via commitment.
This study answers the call of Schwartz (2005) in that it carefully takes the role of
ethnoracial identity into account as a part of an effort to understand ego identity in a broader
theoretical context. This analysis found no difference in the two ethnoracial groups studied, and
the different outcomes hypothesized for the two different ethnoracial groups studied were not
borne out in the data. However, the lack of support for these hypothesized relationships may be
telling still.
Due to the possible relevance of the underrepresentation of Mexican-Americans in the
present study, it is reasonable to plan for furthering this line of research by repeating a similar
study with a population more representative of the total Hispanic-identified population of the
U.S. The USCB is currently considering changes to how Hispanic-identified people in the U.S.
are prompted to designate their ethnic and racial identities. This is based in part on their finding
that Mexican-Americans specifically are responsible for shifts in how Hispanics in the U.S. selfidentify (Krogstad & Cohn, 2014).
It is established that people in the contemporary U.S. who identify as partly Hispanic or
Hispanic are likely to go to a greater effort than their non-Hispanic counterparts in considering
their ethnocultural identity (e.g., Juang & Syed, 2010; Schwartz et al., 2009; Syed & Azimitia,
2008). These people may further struggle with the decision of whether or not it is relevant for
them to take on a Hispanic identity at all (e.g., Syed & Azimitia, 2008). Currently, the sole
ethnicity measured by the USCB requires respondents to identify as either Hispanic or nonHispanic. It has been reported that changes to the 2020 Census may involve collapsing this sole
ethnicity into one question that captures this self-identification alongside the designated race
categories (Krogstad & Cohn, 2014). This may ultimately influence nationwide perceptions in
the U.S. about the meaning of self-identifying as Hispanic.
Therefore, the period between the time of publication of this study and the 2020 Census
may represent a unique time period for ego identity scholars to gather data and study Hispanicidentified people in the U.S. and relevant differences between Hispanic-identified people in the
U.S. of varying backgrounds. Further, this period of time may also be an unusual opportunity for
researchers to examine the role and function of the ethnoracial domain of ego identity overall,
76
`
and in a way that is potentially generalizable across eras and cultures given the lens of Family
Life Course Development theory.
This study has contributed important information about the relational dynamic underlying
the identity development process. Further, by offering data that can be compared with data from
subsequent studies of a Hispanic-identified U.S. population in flux, the present study will
contribute to the growing body of identity scholarship that has been carried out in service of
furthering theoretical knowledge in the context of a broader academic understanding of crosscultural relations.
It is clear that identity scholars are still in a position to understand and foster healthy ego
identity development, even in our globally connected, post-multiculturalist world. However, it is
also clear that the benefits of this potential can only be realized when identity scholars recognize
the bidirectional nature of influence inherent in interventions designed to foster this healthy
development, and when those same interventions are rooted in empirical data that, ideally,
represents the real world of adolescents and young adults in all its complexity.
77
`
APPENDIX A
INFORMED CONSENT FOR FLORIDA STATE UNIVERSITY
78
`
APPENDIX B
INFORMED CONSENT FOR
FLORIDA INTERNATIONAL UNIVERSITY
79
`
80
`
APPENDIX C
PARTICIPANT PACKET
81
`
82
`
83
`
84
`
85
`
86
`
87
`
88
`
APPENDIX D
FACTOR LOADING ESTIMATES FOR
MODEL 3: HYPOTHESIZED MODEL
Table 5
Measurement Items for Cohesion and Enmeshment
Cohesion
1. Family members really helped and supported one another.
2. There was a feeling of togetherness in or family.
3. Our family didn’t do things together.*
4. We really got along well with each other.
5. Family members seemed to avoid contact with each other when at home.*
Enmeshment
1. Family members found it hard to get away from one another.
2. It was difficult for family members to take time away from the family.
3. Family members were pressured to spend most free time together.
4. Family member felt guilty if they wanted to spend some time alone.
5. It seemed like there was never any place to be alone in our house.
*Reverse-scored.
89
`
Table 6
Measurement Items for Commitment
Commitment
1. I have definitely decided on the occupation I want to pursue.
2. I don’t expect to change my political principles and ideals.
3. I have considered adopting different kinds of religious beliefs.*
4. I am very confident about what kinds of friends are best for me.
5. I will always vote for the same political party.
6. I have firmly held views concerning my role in the family.
7. My values are likely to change in the future.*
8. I am not sure about what type of dating relationship is best for me.*
9. Regarding religion, my beliefs are likely to change in the near future.*
10. I have definite views regarding the ways in which men and women should behave.
11. I think what I look for in a friend could change in the future.*
12. I am unlikely to alter my vocational goals.
13. My ideas about men’s and women’s roles will never change.
14. I am not sure that the values I hold are right for me.*
15. The extent to which I value my family is likely to change in the future.*
16. My beliefs about dating are firmly held.
*Reverse-scored.
90
`
Table 7
Measurement Items for Exploration
Exploration
1. There has never been a need to question my values.*
2. My ideas about men’s and women’s roles have changed as I got older.
3. I have engaged in several discussions concerning behaviors regarding dating relationships.
4. I have considered different political views thoughtfully.
5. I have never questioned my views concerning what kind of friend is best for me.*
6. When I talk to people about religion, I make sure to voice my opinion.
7. I have not felt the need to reflect on the importance I place on my family.*
8. I have tried to learn about different occupational fields to find the best one for me.
9. I have undergone several experiences that made me change my views on men’s and women’s
roles.
10. I have consistently re-examined many different values in order to find the ones which are
best for me.
11. I have questioned what kind of date is right for me.
12. I have evaluated many ways in which I fit into my family structure.
13. I have never questioned my political beliefs.*
14. I have many experiences that have led me to review the qualities which I would like my
friends to have.
15. I have discussed religious matters with a number of people who believe differently than I do.
16. I have never questioned my occupational aspirations.*
*Reverse-scored.
91
`
Table 8
Measurement Items for Depressive Symptoms
Depressive Symptoms
1.
I was bothered by things that usually don’t bother me.
2.
I did not feel like eating; my appetite was poor.
3.
I felt that I could not shake off the blues even with help from my family or friends.
4.
I felt that I was just as good as other people.*
5.
I felt depressed.
6.
I felt that everything I did was an effort.
7.
I felt hopeful about the future.*
8.
I thought my life had been a failure.
9.
I felt fearful.
10.
My sleep was restless.
11.
I was happy.*
12.
I talked less than usual.
13.
I felt lonely.
14.
People were unfriendly.
15.
I enjoyed life.*
16.
I had crying spells.
17.
I felt sad.
18.
I felt that people dislike me.
19.
I could not get “going.”
*Reverse-scored.
92
`
Table 9
Measurement Items for Life Satisfaction and Anxiety
Life Satisfaction
1.
In most ways, my life is close to ideal.
2.
The conditions of my life are excellent.
3.
I am satisfied with my life.
4.
So far, I have gotten the important things I want in life.
5.
If I could live my life over, I would change almost nothing.
Anxiety
1.
I feel calm.*
2.
I am tense.
3.
I feel upset.
4.
I am relaxed.*
5.
I feel content.*
6.
I am worried.
*Reverse-scored.
93
`
Table 10
Cohesion and Enmeshment Factor Loading Estimates for Model 3: Hypothesized Model.
Estimate
S.E.
Est./S.E.
Item 1*
1.000
0.000
Item 2
1.227
0.045
27.346
0.000
Item 3
0.985
0.049
20.179
0.000
Item 4
0.891
0.050
17.942
0.000
Item 5
0.835
0.049
17.029
p
Cohesion
_____________________________________________________________________________
Enmeshment
Item 1*
1.000
0.000
Item 2
1.071
0.102
10.553
0.000
Item 3
1.164
0.112
10.427
0.000
Item 4
1.208
0.120
10.103
0.000
Item 5
1.072
0.115
9.323
0.000
______________________________________________________________________________
*Factor loading fixed at 1.
94
`
Table 11
Commitment Factor Loading Estimates for Model 3: Hypothesized Model.
Estimate S.E.
Est./S.E.
p
Commitment
Item 1*
1.000
0.000
Item 2
1.105
0.217
5.092
0.000
Item 3
1.081
0.230
4.702
0.000
Item 4
1.157
0.203
5.708
0.000
Item 5
0.881
0.202
4.353
0.000
Item 6
1.053
0.199
5.298
0.000
Item 7
1.598
0.287
5.567
0.000
Item 8
1.068
0.215
4.967
0.000
Item 9
1.083
0.210
5.153
0.000
Item 10
1.245
0.238
5.239
0.000
Item 11
1.105
0.225
4.911
0.000
Item 12
0.888
0.182
4.887
0.000
Item 13
1.301
0.236
5.509
0.000
Item 14
0.957
0.181
5.296
0.000
Item 15
1.056
0.215
4.918
0.000
Item 16
1.454
0.256
5.687
0.000
*Factor loading fixed at 1.
95
`
Table 12
Exploration Factor Loading Estimates for Model 3: Hypothesized Model.
Estimate
S.E.
Est./S.E.
p
Exploration
Item 1*
1.000
0.000
Item 2
3.213
1.070
3.004
0.003
Item 3
2.170
0.777
2.794
0.005
Item 4
2.437
0.837
2.911
0.004
Item 5
1.166
0.466
2.500
0.012
Item 6
0.501
0.349
1.436
0.151
Item 7
1.037
0.454
2.287
0.022
Item 8
1.713
0.608
2.816
0.005
Item 9
3.932
1.308
3.006
0.003
Item 10
3.713
1.228
3.023
0.003
Item 11
2.680
0.911
2.940
0.003
Item 12
2.614
0.902
2.899
0.004
Item 13
1.914
0.658
2.907
0.004
Item 14
1.936
0.682
2.838
0.005
Item 15
1.947
0.697
2.793
0.005
Item 16
0.779
0.367
2.123
0.034
*Factor loading fixed at 1.
96
`
Table 13
Depression Factor Loading Estimates for Model 3: Hypothesized Model.
Estimate
S.E.
Est./S.E.
Item 1*
1.000
0.000
Item 2
0.843
0.111
7.604
0.000
Item 3
1.781
0.162
10.980
0.000
Item 4
0.891
0.125
7.131
0.000
Item 5
1.670
0.150
11.115
0.000
Item 6
0.724
0.122
5.947
0.000
Item 7
1.063
0.118
8.971
0.000
Item 8
0.944
0.096
9.847
0.000
Item 9
1.068
0.116
9.182
0.000
Item 10
1.163
0.137
8.483
0.000
Item 11
1.380
0.130
10.605
0.000
Item 12
0.987
0.117
8.414
0.000
Item 13
1.616
0.151
10.699
0.000
Item 14
0.671
0.097
6.898
0.000
Item 15
1.322
0.130
10.185
0.000
Item 16
0.896
0.098
9.132
0.000
Item 17
1.669
0.150
11.157
0.000
Item 18
1.137
0.117
9.695
0.000
Item 19
1.304
0.130
10.049
0.000
p
Depression
*Factor loading fixed at 1.
97
`
Table 14
Life Satisfaction and Anxiety Factor Loading Estimates for Model 3: Hypothesized Model.
Estimate
S.E.
Est./S.E.
Item 1
1.000
0.000
Item 2
0.958
0.056
17.168
0.000
Item 3
1.071
0.057
18.711
0.000
Item 4
0.963
0.062
15.608
0.000
Item 5
0.979
0.074
13.284
0.000
Item 1
1.000
0.000
Item 2
0.909
0.059
15.534
0.000
Item 3
0.548
0.047
11.718
0.000
Item 4
1.156
0.060
19.321
0.000
Item 5
0.887
0.062
14.392
0.000
Item 6
0.890
0.065
13.739
0.000
p
Satisfaction
Anxiety
*Factor loading fixed at 1.
98
`
APPENDIX E
FLORIDA STATE UNIVERSITY
INSTITUTIONAL REVIEW BOARD
HUMAN SUBJECTS APPROVAL LETTER
99
`
APPENDIX F
FLORIDA INTERNATIONAL UNIVERSITY
INSTITUTIONAL REVIEW BOARD
HUMAN SUBJECTS APPROVAL LETTER
100
`
REFERENCES
Adams, G. R., Dyk, P. A. H., & Bennion, L. D. (1987). Parent–adolescent relationships and
identity formation. Family Perspective, 21, 249-260.
Alexander, K. L., Entwisle, D. R., & Kabbani, N. S. (2001). The dropout process in life course
perspective: Early risk factors at home and school. Teachers College Record, 103(5),
760-822.
Archer, S.L., & Waterman, A.S. (1994). Adolescent identity development: Contextual
perspectives. In C.B. Fisher & R.M. Lerner (Eds.), Applied Developmental Psychology.
New York, NY: McGraw-Hill.
Arseth, A. K., Kroger, J., Martinussen, M., & Marcia, J. E. (2009). Meta-analytic studies of
identity status and the relational issues of attachment and intimacy. Identity: An
International Journal of Theory and Research, 9(1), 1-32.
Baldwin, J. (1955). Notes of a native son. Boston: The Beacon Press.
Balistreri, E., Busch-Rossnagel, N. A., & Geisinger, K. F. (1995). Development and validation of
the Ego Identity Process Questionnaire. Journal of Adolescence, 18, 179-192.
Barber, B. K., & Buehler, C. (1996). Family cohesion and enmeshment: Different constructs,
different effects. Journal of Marriage and the Family, 58, 433-441.
Baron, R.M., & Kenny, D.A. (1986). The moderator-mediator variable distinction in social
psychological research: Conceptual, strategic, and statistical considerations. Journal of
Personality and Social Psychology, 51, 1173-1182.
Behdad, A. (2005). A forgetful nation: On immigration and cultural identity in the United States.
Durham, NC: Duke University Press.
Bengston, V.L., & Allen, K.R. (1993). The life course perspective applied to families over time.
In P. G. Boss, W.J. Doherty, R. LaRossa, W.R. Schumm, & S.K. Steinmetz (Eds.),
Sourcebook of family theories and methods: A contextual approach (pp. 469-499). New
York. NY: Plenum Press.
Bennion, L.D., & Adams, G.R. (1999). A revision of the extended version of the Objective
Measure of Ego Identify Status: An identity instrument for use with late adolescents.
Journal of Adolescent Research, 1, 183-198.
Blasi, A. & Glodis, K. (1995). The development of identity: A critical analysis from the
perspective of the self as subject. Developmental Review, 15, 404-433.
Bloom, B. L. (1985). A factor analysis of self-report measures of family functioning. Family
Process, 24, 225 – 239.
101
`
Bosma, T. L., Graafsma, G., Grotevant, H. D., & de Levita, D. J. (1994). Identity and
development: An interdisciplinary approach. Thousand Oaks, CA: Sage Publications
Boyd, A.O. (2010). The Latin American identity and the African diaspora: Ethnogenesis in
context. Amherst, NY: Cambria Press.
Brown, R.L. (1997). Assessing specific mediational effects in complex theoretical models.
Structural Equation Modeling: A Multidisciplinary Journal, 4, 142-156.
Campbell, E., Adams, G. R., & Dobson, W. R. (1984). Familial correlates of identity formation
in late adolescence: A study of the perceived utility of connectedness and individuality in
family relations. Journal of Youth and Adolescence, 13, 509-525.
Chae, M. H., & Foley, P. F. (2010). Relationship of ethnic identity, acculturation, and
psychological well-being among Chinese, Japanese, and Korean Americans. Journal of
Counseling & Development, 88(4), 466-476.
Chun, Y., & MacDermid, S. M. (1997). Perceptions of family differentiation, individuation, and
self-esteem among Korean adolescents. Journal of Marriage and the Family, 59(2), 451462.
Cooper, C.R., Grotevant, H.D., & Condon, S.M. (1984). Individuality and connectedness: Both
foster adolescent identity formation and role-taking skills. In H. D. Grotevant and C.R.
Cooper (Eds.), Adolescent development in the family: New directions for child
development. San Francisco, CA: Jossey-Bass.
Côté, J.E. (1997). An empirical test of the identity capital model. Journal of Adolescence, 20,
421-137.
Côté, J.E. & Levine, C. (1988). A critical examination of the ego identity status paradigm.
Developmental Review, 8(2), 147-184.
Crocetti, E., Klimstra, T., Keijsers, L., Hale, W., & Meeus, W. (2009). Anxiety trajectories and
identity development in adolescence: A five-wave longitudinal study. Journal of Youth
and Adolescence, 38(6), 839-849.
Diener, E., Emmons, R. A., Larsen, R. J., & Griffin, S. (1985). The satisfaction with life scale.
Journal of Personality Assessment, 49, 71 – 75.
Dillman, D.A. (2007). Mail and Internet surveys: The tailored design method (2nd ed.). Hoboken,
NJ: John Wiley & Sons, Inc.
DuBois, W.E.B. (1925). The negro mind reaches out. In A. Locke (Ed.), The new negro. (pp.
385-414). New York, NY: Albert and Charles Boni
Elder, G.H. (1985). Life course dynamics: Trajectories and transitions, 1968-1980. Ithaca, NY:
Cornell University Press.
102
`
Elder, G.H. (1998). The life course and human development. In W. Damon & R. M. Learner,
(Eds.), Handbook of child psychology (Vol. 1): Theoretical models of human development
(pp. 939-992). New York, NY: John Wiley & Sons, Inc.
Elder, G.H., Johnson, M.K., & Crosnoe, R. (2003). The emergence and development of life
course theory. In J.T. Mortimer & M.J. Shanahan (Eds.), Handbook of the life course (pp.
3-19). New York, NY: Springer.
Ennis, S.R., Ríos-Vargas, M., & Albert, N.G. (2011). The Hispanic population: 2010. Retrieved
from http://www.census.gov/prod/cen2010/briefs/c2010br-04.pdf
Erikson, E.H. (1959). Identity and the life cycle. New York, NY: Norton.
Erikson, E. H. (1968). Identity: Youth and crisis. New York, NY: Norton.
Faber, A. J., Edwards, A. E., Bauer, K. S., & Wetchler, J. L. (2003). Family structure: Its effects
on adolescent attachment and identity formation. The American Journal of Family
Therapy, 31, 243-255.
Field, P.S. (2002). Ralph Waldo Emerson: The making of a democratic intellectual. New York,
NY: Rowman & Littlefield.
Florida International University. (2012). About us. Retrieved from www.fiu.edu/aboutus/index.html.
Florida State University. (2012). Florida State University Factbook 2011-12: Student
Characteristics, Fall 2011. Retrieved from www.ir.fsu.edu/Factbooks/201112/students.pdf.
Freud, S. (1933). New introductory lectures on psychoanalysis. New York, NY: Norton.
Fuhrman, T., & Holmbeck, G. N. (1995). A contextual-moderator analysis of emotional
autonomy and adjustment in adolescence. Child Development, 66(3), 793-811.
Glazer, N. (1993). Is assimilation dead? Annals of the American Academy of Political and Social
Science, 530, 122–136.
Grotevant, H.D. (1983). The contribution of the family to the facilitation of identity formation in
early adolescence. Journal of Early Adolescence, 3, 225-237.
Guglielmo, T.A. (2004). White on arrival: Italians, race, color, and power in Chicago, 18901945. New York, NY: Oxford University Press.
Guglielmo, J., & Salerno, S. (2003). Are Italians white? How race is made in America. New
York, NY: Routledge.
103
`
Haid, M.L., Seiffge-Krenke, I., Molinar, R., Ciairano, S., Karaman, N. G., & Cok, F. (2010).
Identity and future concerns among adolescents from Italy, Turkey and Germany: Intraand between-cultural comparisons. Journal of Youth Studies, 13(3), 369-389.
Hammer, T. (1996). Consequences of unemployment in the transition from youth to adulthood in
a life course perspective. Youth and Society, 27(4), 450-468.
Hammer, T. (2003) Youth unemployment and social exclusion in Europe. Bristol, United
Kingdom: The Policy Press, University of Bristol.
Hammer, O., Harper, D.A.T., & Ryan, P.D. (2001). PAST: Paleontological Statistics software
package for education and data analysis. Palaeontologia Electronica 4(1), 9.
Hardy, S. A., & Kisling, J. W. (2006). Identity statuses and prosocial behaviors in young
adulthood. Identity: An International Journal of Theory and Research, 6, 363-369.
Hollinger, D. A. (2006). Postethnic America: Beyond multiculturalism. New York, NY: Basic
Books.
Hu, L.T. and Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure
analysis: Conventional criteria versus new alternatives. Structural Equation Modeling,
6(1), 1-55.
Humes, K.R., Jones, N.A., & Ramírez, R.R. (2011). Overview of race and Hispanic origin: 2010.
Retrieved from http//www.census.gov/prod/cen2010/briefs/c2010br-02.pdf.
Hutchison, E.D. (2011). A life course perspective. In Dimensions of human behavior: The
changing life course, E.D. Hutchison, ed., pp. 1-38. Sage Publications: Thousand Oaks,
CA.
Jackson, E. P., Dunham, R. M., & Kidwell, J. S. (1990). The effect of family cohesion and
adaptability on identity status. Journal of Adolescent Research, 5, 161-174.
Jacobson, M.F. (1999). Whiteness of a different color: European immigrants and the alchemy of
race. Cambridge, MA: Harvard University Press.
Juang, L., & Syed, M. (2010). Family cultural socialization practices and ethnic identity in
college-going emerging adults. Journal of Adolescence, 33(3), 347-354.
Kallen, H. M. (1956). Cultural pluralism and the American idea. Philadelphia, PA: University of
Pennsylvania Press.
Kenny, D.A. (2011a). Identification: Overview. Retrieved from
www.davidakenny.net/cm/identify.htm
Kenny, D.A. (2011b). Structural equation modeling: Multiple groups. Retrieved from
www.davidakenny.net/cm/mgroups.htm
104
`
Kline, R.B. (1998). Principles and practice of structural equation modeling. New York: Guilford
Press.
Kline, R.B. (2012). Assumptions in structural equation modeling. In R. Hoyle (Ed.), Handbook
of Structural Equation Modeling. New York: Guilford Press.
Kolchin, P. (2002). Whiteness studies: The new history of race in America. The Journal of
American History, 89(1), 154-173.
Kroger, J. (2003). Identity development during adolescence. In G. R. Adams, & M. D.
Berzonsky (Eds), Blackwell handbook of adolescence. Malden, MA: Blackwell
Publishing.
Krogstad, J.M., and Cohn, D. (2014). U.S. Census looking at big changes in how it asks about
race and ethnicity. Retrieved from http://www.pewresearch.org/fact-tank/2014/us-censuslooking-at-big-changes-in-how-it-asks-about-race-and-ethnicity
Kurtines, W.M. (1999). A co-constructivist perspective on human behavior and development.
Manuscript submitted for publication.
Lacombe, A.C. & Gay, J. (1998). The role of gender in adolescent identity and intimacy
decisions. Journal of Youth and Adolescence, 27(6), 795-802.
Lewinsohn, P.M., Seeley, J.R., Roberts, R.E., & Allen, N.B. (1997). Center for Epidemiological
Studies-Depression Scale (CES-D) as a screening instrument for depression among
community-residing older adults. Psychology and Aging, 12, 277-287.
MacCallum, R.C., Browne, M.W., and Sugawara, H.M. (1996). Power analysis and
determination of sample size for covariance structure modeling. Psychological Methods,
1(2), 130-149.
Manzi, C., Vignoles, V. L., Regalia, C., & Scabini, E. (2006). Cohesion and enmeshment
revisited: Differentiation, identity, and well-being in two European countries. Journal of
Marriage and Family, 68(3), 673-689.
Marcia, J.E. (1966). Development and validation of ego identity status. Journal of Personality
and Social Psychology, 3(5), 551-558.
Marcia, J. E. (1980). Identity in adolescence. In J. Adelson (Ed.), Handbook of adolescent
psychology (pp. 109–187). New York, NY: Wiley & Sons.
Marcia (1989). Identity and intervention. Journal of Adolescence, 12, 401-410.
Marcia, J. (1995). The empirical study of ego identity. In H.A. Bosma, L.G. Tobi, H.D.
Grotevant, & D.J. de Levita (Eds.), Identity and development: An interdisciplinary
approach. Thousand Oaks, CA: Sage Publications.
105
`
Marteau, T. M., & Bekker, H. (1992). The development of a six-item short-form of the state
scale of the Spielberger State-Trait Anxiety Inventory (STAI). British Journal of Clinical
Psychology, 31, 301 – 306.
McAdoo, H.P. (1981). Patterns of upward mobility in black families. In H.P. McAdoo (Ed.),
Black families (pp. 155-170). Beverly Hills, CA: Sage Publications.
Minuchin, S. (1974). Families and family therapy. Cambridge, MA: Harvard University Press.
Montgomery, M. J., & Côté, J. E. (2003). The transition to college: Adjustment, development,
and outcomes. In G. R. Adams & M. D. Berzonsky (Eds.), The Blackwell handbook of
adolescence (pp. 179–194). Oxford, United Kingdom: Blackwell Publishers, Ltd.
Mullis, R. L., Brailsford, J. C., & Mullis, A. K. (2003). Relations between identity formation and
family characteristics among young adults. Journal of Family Issues, 24(8), 966-980.
Muthén, L. K., & Muthén, B. O. (2012). Mplus user's guide (6th ed.). Los Angeles, CA: Muthén
& Muthén.
Olson, D. H. (1982). Circumplex model of marital and family system. In F. Walsh (Ed.), Normal
family processes (pp. 115 – 132). New York, NY: Guilford.
Phinney, J.S. (1990). Ethnic identity in adolescents and adults: Review of research.
Psychological Bulletin, 108, 499-514.
Protinsky, H. & Shuts, L. (1990). Adolescent substance use and family cohesion. Family
Therapy: The Journal of the California Graduate School of Family Psychology, 17(2):
173-175.
Radloff, L. S. (1977). The CES-D scale: A self-report depression scale for research in the general
population. Applied Psychological Measurement, 1, 385–401.
Rodgers, R.H. & White, J.M. (1993). Family development theory. In Boss, P. G., Doherty, W. J.,
LaRossa, R., Schumm, W. R., & Steinmetz, S. K. (Eds.), Sourcebook of family theories
and methods: A contextual approach (pp. 225-258). New York, NY: Plenum.
Roth, P.L., Switzer, F.S., and Switzer, D.M. (1999). Missing data in multiple item scales: A
Monte Carlo analysis of missing data techniques. Organizational Research Methods,
2(3); 211-32.
Scabini, E. (1985). Family organization between crisis and development. Milano, Italy: Franco
Angeli.
Schiedel, D. G. , & Marcia, J. E. (1985). Ego identity, intimacy, sex role orientation, and gender.
Developmental Psychology, 21(1), 149-160.
106
`
Schwartz, S. J. (2001). The evolution of Eriksonian and neo-Eriksonian identity research: A
review and integration. Identity: An International Journal of Theory and Research, 1(1),
7-58.
Schwartz, S.J. (2004). Brief report: Construct validity of two identity status measures: the EIPQ
and the EOM-EIS-II. Journal of Adolescence, 27, 477-483.
Schwartz, S.J. (2005). A new identity for identity research: Recommendations for expanding and
refocusing the identity literature. Journal of Adolescent Research, 20(3), 293-308.
Schwartz, S. J., Mason, C. A., Pantin, H., & Szapocznik, J. (2009). Longitudinal relationships
between family functioning and identity development in hispanic adolescents: Continuity
and change. Journal of Early Adolescence, 29(2), 177-211.
Schwartz, S.J., Zamboanga, B.L, Weisskirch, R.S. & Wang, S.C. (2010). The relationships of
personal and cultural identity to adaptive and maladaptive psychological functioning in
emerging adults. Journal of Social Psychology, 150(1), 1-33.
Shin, D.C., & Johnson, D.M. (1978). Avowed happiness as an overall assessment of the quality
of life. Social Indicators Research, 5, 475-492.
Society for Research on Identity Formation (2013). SRIF: Identity resources. Retrieved from
http://w3.fiu.edu/srif/IdentityResources.htm
Syed, M. & Azmitia, M. (2008). A narrative approach to ethnic identity in emerging adulthood:
Bringing life to the identity status model. Developmental Psychology, 44(4): 1012-27.
Taylor, G. (2004). Buying whiteness: Race, culture, and identity from columbus to hip hop. New
York, NY: Palgrave Macmillan.
Taylor, P., Lopez, M.H., Martínez, J., and Velasco, G. (2012). When labels don’t fit: Hispanics
and their views of identity: II. Identity, pan-Ethnicity, and race. Retrieved from
www.pewhispanic.org/2012/04/04/ii-identity-pan-ethnicity-and-race/
Thomas, R.M. (2005). Erikson’s variation on Freud’s theme. In R.M. Thomas, Comparing
theories of child development (pp. 86-117). Belmont, CA: Thomson Wadsworth.
Umaña-Taylor, A. J., Gonzales-Backen, M. A., & Guimond, A. B. (2009). Latino adolescents’
ethnic identity: Is there a developmental progression and does growth in ethnic identity
predict growth in self-esteem? Child Development, 80(2), 391-405.
van Hoof, A. (1999). The identity status approach: In need of fundamental revision and
qualitative change. Developmental Review, 19, 497-556.
Vertovec, S. (2010). Towards post-multiculturalism? Changing communities, conditions and
contexts of diversity. International Social Science Journal, 61(199): 83-95.
Vought, H.P. (2004). The bully pulpit and the melting pot. Macon, GA: Mercer University Press.
107
`
Waterman, A. S. (1988). Identity status theory and Erikson’s theory: Commonalities and
differences. Developmental Review, 8(2), 185-208.
Watson, M. F., & Protinsky, H. O. (1988). Black adolescent identity development: Effects of
perceived family structure. Family Relations, 37, 288-292.
White, J.M & Klein, D.M. (2002). Family theories: An introduction (2nd ed.). Thousand Oaks,
CA: Sage.
Wright, J. P., Cullen, F. T., & Williams, N. (2002). The embeddedness of adolescent
employment and participation in delinquency: A life course perspective. Western
Criminology Review, 4, 1-19.
108
`
BIOGRAPHICAL SKETCH
Summer Brooke Gómez earned her Bachelor of Science degree in Psychology in 2002 from the
Florida State University and her Master of Social Work degree in 2005, also from the Florida
State University. Ms. Gómez is a Florida native and was born and raised in Miami, Florida.
Ms. Gómez has been interested in the study of ego identity development since she began her
academic career as an undergraduate in psychology at Florida International University, studying
under Dr. William M. Kurtines. Ms. Gómez went on to study under identity scholar Dr. Ronald
L. Mullis for her doctorate. Ms. Gómez is a trained clinician and an active mental health
professional and has owned a private practice in Tallahassee, Florida since 2009. It is her
intention to continue functioning as a scholar-practitioner.
109