Urban Schools

Teaching and Teacher Education 17 (2001) 807–818
A multilevel analysis of the relationship between teacher and
collective efficacy in urban schools
Roger D. Goddarda,*, Yvonne L. Goddardb
a
University of Michigan, 4204 School of Education, 610 East University Avenue, Ann Arbor, MI 48109-1259, USA
b
University of Toledo, 5005 Gillham Hall, College of Education, Toledo, OH 43606, USA
Received 21 September 2000; received in revised form 29 January 2001; accepted 15 March 2001
Abstract
Although a great deal of research has linked both teacher and collective efficacy to student achievement, one
overlooked question concerns the nested association between teacher and collective efficacy. The authors apply social
cognitive theory to offer a theoretical analysis of this relationship. Next, using hierarchical linear modeling, they
empirically test the strength of the relationship between these two theoretically related yet conceptually distinct
constructs. Analysis of data collected from 438 teachers in 47 schools in a large urban school district shows that
collective efficacy predicts variation in teacher efficacy above and beyond the variance explained by a number of school
contextual factors including socioeconomic status and student achievement. The implications of these findings for
future research are discussed. r 2001 Elsevier Science Ltd. All rights reserved.
Keywords: Teacher efficacy; Collective efficacy; Student achievement; Social cognitive theory
1. A multilevel analysis of the relationship between
teacher and collective efficacy in urban schools
Bandura (1997) developed social cognitive
theory to explain that the control humans exercise
over their lives through agentive actions is powerfully influenced by the strength of their efficacy
beliefs. Bandura defines efficacy as ‘‘beliefs in one’s
capabilities to organize and execute the courses of
action required to produce to given attainments’’
(p. 3). Perceptions of efficacy are important to
individual and organizational behavior and
change.
*Corresponding author. Tel.: +1-734-764-7488; fax: +1734-763-1504.
E-mail address: [email protected] (R.D. Goddard).
Notably, there are two distinct, but theoretically-related, types of efficacyFindividual and
collective. For more than two decades researchers
interested in individual teacher efficacy have
investigated its correlates and argued that
teachers’ perceptions of self-capability are important to student learning (e.g., Armor et al., 1976;
Gibson & Dembo, 1984; Ross, 1992). More
recently, researchers have shown that collective
efficacy is also related to student achievement
differences among schools (Bandura, 1993; Goddard, Hoy, & Woolfolk, 2000). But despite the
promise of these constructs when considered
independently, researchers have yet to consider
their interrelationship.
Our focus on teacher and collective efficacy was
also motivated by troubling evidence indicating
0742-051X/01/$ - see front matter r 2001 Elsevier Science Ltd. All rights reserved.
PII: S 0 7 4 2 - 0 5 1 X ( 0 1 ) 0 0 0 3 2 - 4
808
R.D. Goddard, Y.L. Goddard / Teaching and Teacher Education 17 (2001) 807–818
that, depending on personal characteristics and
experiences, teachers new to urban schools sometimes experience declines in teacher efficacy during
their first year (Chester & Beaudin, 1996). Notably, Chester and Beaudin’s study offers insight into
individual-level predictors of variation in teacher
efficacy, but they did not consider variation among
schools in teacher efficacy. This made us curious to
know whether teacher efficacy varied systematically among schools and, if so, to what extent a
school’s collective efficacy in particular was
predictive of this variation.
As we discuss more fully later, collective efficacy
is a potent way of characterizing the social
influence of a school. In their review of the
historical and theoretical development of teacher
efficacy research, Tschannen-Moran, Woolfolk
Hoy, and Hoy (1998) observed that ‘‘the effect of
collective efficacy may be especially pronounced
for novice teachers as they are socialized into the
teaching profession’’ (p. 221). This indicated to us
that other researchers were also interested in the
relationship between teacher and collective efficacy. We decided to examine whether a school’s
collective efficacy is systematically related to
differences in teacher efficacy among schools. To
inform this inquiry, we offer a theoretical analysis
of the relationship between teacher and collective
efficacy. Next, we report the results of a multilevel
analysis of the relationship between teacher and
collective efficacy observed in the elementary
schools of a large urban district. Finally, we
discuss the implications of the important connections we found between teacher and collective
efficacy.
2. The effects of teacher and collective efficacy:
research evidence
Over more than 20 years, researchers have
shown that teachers’ perceptions of their selfcapability to educate students are significantly and
positively related to teacher behaviors that promote student achievement. The earliest connection
between teacher efficacy and student achievement
was established by Armor et al. (1976) in a study
of reading gains among students in Los Angeles
schools. Using reading scores obtained from the
1974 and 1975 administrations of the California
Test of Basic Skills, Armor et al. found that the
higher the efficacy of teachers in a special reading
program, the higher the reading gain of their
students. Since then, other researchers have
provided additional support linking teacher efficacy to student achievement (e.g., Anderson,
Greene, & Loewen, 1988; Ashton & Webb, 1986;
Ross, 1992).
Notably, the relationship between teacher efficacy and student achievement appears to be
indirect, with teacher efficacy influencing numerous teacher behaviors that, in turn, promote
student achievement. According to Gibson and
Dembo (1984), teacher efficacy ‘‘may influence
certain patterns of behavior known to influence
achievement gains’’ (p. 579). For example, Gibson
and Dembo found that highly efficacious teachers
tend to persist in helping struggling students arrive
at correct answers rather than simply giving
students answers or allowing others to provide
the correct answers. Similarly, in their study of the
relationship between teacher efficacy and student
achievement, Ashton and Webb (1986) observed
that ‘‘teachers with a high sense of efficacy seemed
to employ a pattern of strategies that minimized
negative affect, promoted an expectation of
achievement, and provided a definition of the
classroom situation characterized by warm interpersonal relationships and academic work’’
(p. 125). Since these earlier observations were
made, several studies have confirmed that teacher
efficacy is indeed associated with a number of
important variables. These variables include organized and planful teaching (Allinder, 1994), the use
of activity-based learning (Enochs, Scharmann,
& Riggs, 1995), and student-centered learning
(Czerniak & Schriver, 1994). Moreover, the higher
teachers’ efficacy, the more humanistic their
approach to pupil control (Woolfolk & Hoy,
1990). Teacher efficacy may also have an indirect
effect on student achievement through its positive
association with trust (Da Costa & Riordan, 1996;
Goddard, Tschannen-Moran, & Hoy, in press),
openness to educational consultation (DeForest &
Hughes, 1992), positive attitudes toward educational reform (DeMesquita & Drake, 1994;
R.D. Goddard, Y.L. Goddard / Teaching and Teacher Education 17 (2001) 807–818
Guskey, 1988; Smylie, 1988), teacher satisfaction
(Lee, Dedrick, & Smith, 1991), and increased levels
of parental involvement in schooling (HooverDempsey, Bassler, & Brissie, 1992, 1987).
In light of the powerful findings about individual teacher efficacy, several researchers have
considered efficacy at the organizational, or
collective, level. For schools, collective efficacy
refers to the perceptions of teachers in a school that
the faculty as a whole can organize and execute the
courses of action required to have a positive effect on
students. Perceptions of efficacy serve to influence
the behavior of individuals and the normative
environment of collectives by providing expectations about the likelihood of success for various
pursuits.
The study of collective efficacy is relatively new.
Indeed, Bandura (1997) observes that ‘‘although
perceived collective efficacy is widely recognized to
be highly important to a full understanding of
organizational functioning, it has been the subject
of little research’’ (p. 468).1 So far, these studies
have investigated the relationship between collective efficacy and student achievement. One of the
earliest collective efficacy studies was by Bandura
(1993), who showed that collective efficacy is
significantly and positively related to school-level
achievement. More recently, two studies by
Goddard and his colleagues (Goddard, 2000;
Goddard, Hoy, & Woolfolk, 2000) showed that
collective efficacy perceptions are an important
predictor of differences among schools in studentlevel achievement.
To summarize, several teacher efficacy studies
and an emerging body of work on collective
efficacy suggest that these constructs are related
positively to student achievement. To date, however, no study has considered how efficacy
perceptions interact across levelsFindividual and
collective. Since they have the same theoretical
underpinnings, teacher and collective efficacy are
indeed potentially related. We turn now to a
careful consideration of the shared theoretical
1
For a more detailed discussion of research on collective
efficacy in other fields, and social cognitive theory at the group
level, readers are referred to Chapter 11 of Bandura (1997) and
Goddard, Hoy, and Woolfolk Hoy (2000).
809
underpinnings and then a discussion of the
relationship between teacher and collective
efficacy.
3. Shared theoretical underpinnings
As explained earlier, Bandura’s social cognitive
theory (1997) provides the theoretical framework
underlying both teacher and collective efficacy. A
fundamental assumption of social cognitive theory
is human agency. When humans and organizations
(through the collective actions of group members)
make choices, they exhibit agency. According to
social cognitive theory, efficacy is key to the
operation of agency because individuals and
collectives are more likely to pursue activities for
which they believe they have the capability to
succeed.
In addition to the assumption of agency,
Bandura (1997) postulated four sources of information that influence individual efficacy: mastery
experience, vicarious experience, social persuasion,
and affective states. Bandura (1997) also emphasized that whether efficacy beliefs are enhanced or
diminished after a given level of enactive experience is not simply an artifact of the performance;
efficacy beliefs are created when individuals weigh
and interpret their performance relative to other
information. According to Bandura, ‘‘changes in
perceived efficacy result from cognitive processing
of the diagnostic information that performances
convey about capability rather than the performances per se’’ (p. 81). The same is true for all four
sources of efficacy informationFthe role of
cognition is critical. Stated another way, individuals’ perceptions of self-efficacy for various
pursuits arise from cognitive and metacognitive
processing of relevant information.
Importantly, Goddard, Hoy, and Woolfolk Hoy
(2000) argue that the sources of efficacy information postulated by Bandura operate at both the
individual and collective levels. In fact, Goddard
(2000) showed that mastery experience explained
variation among schools in collective efficacy
above and beyond that accounted for by school
SES or minority proportion. Similarly, just as
schools learn vicariously from other schools, they
810
R.D. Goddard, Y.L. Goddard / Teaching and Teacher Education 17 (2001) 807–818
are also influenced by leaders (social persuasion)
and affective states that result from collective
conditions such as successes or tragedies that
impact all school members.
4. Theoretical rationale and hypothesis
The findings reviewed above are evidence of a
growing body of literature that documents the
importance of both teacher and collective efficacy
to student achievement. But, in our opinion, an
interesting question has not yet been addressed in
the research literature. Specifically, how is collective efficacy (an emergent property of schools)
related to teacher efficacy (a teacher-level
attribute)?
Since social cognitive theory specifies that
teachers’ perceptions of self and group capability
influence their actions, it follows that these actions
will be judged by the group relative to group
norms such as those set by collective efficacy
beliefs. According to Coleman (1985, 1987, 1990),
norms develop to permit group members some
control over the actions of others when those
actions have consequences for the group. When a
teacher’s actions are incongruent with the shared
beliefs of the group, the teacher’s actions will be
sanctioned by group members; in fact, Coleman
argues that the severity of the social sanctions
delivered to those who break norms will be equal
in magnitude to the effect of norm-breaking on the
collective. Thus, if most teachers in a school
believe the faculty can successfully teach students,
the normative and behavioral environment will
press teachers to persist in their educational efforts
so that students achieve to high levels. Moreover,
the press to perform will be accompanied by social
sanctions for those who do not. From a sociocognitive perspective, the power of such normative
press may be understood as the effect of social
persuasion on teachers’ individual efficacy perceptions.
Another way to address the relationship between collective and teacher efficacy from a
theoretical perspective is to consider how schools
have collective mastery experiences. When a
school as a unit experiences genuinely high levels
of student achievement, it is axiomatic to conclude
that one or more teachers were directly successful
with their students. In other words, when a school
has a collective mastery experience, so too do one
or more teachers. Thus, mastery experienceFone
of the most powerful sources of efficacy-shaping
information (Bandura, 1997; Pajares, 1997)Fhas
the potential to operate in concert at both the
individual and collective levels. From this perspective, teacher and collective efficacy may covary
positively in response to a given mastery experience.
In addition, we believe that evidence of a
positive relationship between several school contextual variables and teacher efficacy is important
to understanding the relationship between collective efficacy and teacher efficacy. For example,
Moore and Esselman (1992) showed that teacher
efficacy was positively associated with school
climate, lack of impediments to effective instruction, and teacher empowerment. In addition, Hoy
and Woolfolk (1993) found that teachers’ efficacy
is influenced by the contextual variables of
principal influence with superiors and the academic press of a school. Together, these studies
suggest that school contextual factors are related
to teachers’ perceptions of their self-efficacy for
educating students successfully.
These findings led us to consider whether
collective efficacy, a way of characterizing the
normative and behavioral context (i.e., social
influence) of a school, would also be related to
teacher efficacy. We believe that teachers’ perceptions of self-capability may be either enhanced or
attenuated by perceptions of collective capability
and related group member expectations for
performance. Teachers are aware of and influenced by the social processes and collective beliefs
that characterize a school. Indeed, an individual
with modest teacher efficacy might persist more in
the face of personal obstacles and setbacks in a
school where teachers tend to believe in the
group’s conjoint capability to educate the students
successfully. Conversely, the same individual
might experience a decrease in teacher efficacy
upon joining a faculty that dwells on past group
failures and has little expectation of organizational
improvement. In such a context, even a highly
R.D. Goddard, Y.L. Goddard / Teaching and Teacher Education 17 (2001) 807–818
efficacious teacher may begin to view personal
effort and persistence as inconsequential and hence
experience a decline in teacher efficacy. These
examples are meant to illustrate the theoretical
possibility that collective efficacy influences teacher efficacy through its effect on the normative
and behavioral environment of the school. From a
sociocognitive perspective, the power of such
normative press may be understood as the effect
of a faculty’s social persuasion (one of the four
sources of efficacy-shaping information) on individual teacher efficacy.
Based on the theoretical considerations described above, we hypothesized that collective
efficacy would be a significant positive predictor
of differences among schools in teacher efficacy.
Below we describe the method employed to test
this hypothesis.
5. Method
5.1. Sample
The data for this study were obtained from a
survey of teachers in a large urban school district
located in the mid-western United States. Fiftytwo elementary schools from the district were
randomly selected for inclusion. Of these, three
declined to participate. The decision rule for
including schools in the final sample was to obtain
at least five faculty respondents per school. Since
data beyond the scope of the present study were
also collected, approximately half of the faculty
received a survey assessing teacher and collective
efficacy while the other half received a different
survey. Within each faculty, survey distribution
was randomized. Of the 49 participating schools,
two provided fewer than five faculty respondents.
This yielded a final sample of 452 teachers in 47
schools. Importantly, to guarantee their anonymity the teachers surveyed were not asked to report
any demographic information (e.g., grade-level
taught, gender, etc.) that could be used to identify
them. The school district provided data indicating
students’ free or reduced-price lunch status,
gender, minority status, and academic achieve-
811
ment measured by a mandatory state assessment
administered to 4th grade students.
5.2. Efficacy measures
The key distinction between individual and
collective efficacy involves the objectFself or
groupFof the efficacy perception. Hence, researchers interested in teacher efficacy have developed measures that assess a teacher’s perception of
self-capability to educate students successfully
(e.g., Gibson & Dembo, 1984). Collective efficacy,
on the other hand, differs in that the assessment is
not directed at individual capability. This is
consistent with the approach taken by Bandura
(1997) who observes that ‘‘collective efficacy is
concerned with the performance capability of a
social system as a whole’’ (p. 469). Hence,
collective efficacy measures aggregate teachers’
perceptions of the extent to which the faculty as
a whole can teach successfully. It is also important
to observe the distinction between efficacy perceptions and outcome expectancies. Efficacy constructs measure a person’s belief in his/her ability
to execute the actions required to succeed at a
given task. Outcome expectancies, on the other
hand, indicate a person’s belief that certain
behaviors will lead to desired outcomes. Stated
differently, a person may believe that particular
behaviors will produce certain results (outcome
expectancy), but concurrently believe that she/he is
incapable of initiating those behaviors (low self
efficacy). In such a case, it is one’s efficacy beliefs
as opposed to outcome expectancies that are likely
to be more predictive of whether one will successfully achieve some outcome (Bandura, 1997).
5.3. Teacher efficacy measure
Teacher efficacy was measured using a five-item
personal teacher efficacy scale based on Gibson
and Dembo’s (1984) teacher efficacy scale. Scores
on the five-item teacher efficacy subscale have been
shown to have adequate internal consistency and a
one factor structure in previous research (Hoy &
Woolfolk, 1993; Woolfolk & Hoy, 1990). ‘‘If I
really try hard, I can get through to even the most
812
R.D. Goddard, Y.L. Goddard / Teaching and Teacher Education 17 (2001) 807–818
difficult or unmotivated students’’ is a sample item
from the scale.
5.4. Collective efficacy measure
An important reason for the lack of collective
efficacy studies concerns the need for a measure of
collective efficacy (Bandura, 1997; Pajares, 1996).
This study employed a 21-item collective efficacy
scale recently developed by Goddard, Hoy, and
Woolfolk Hoy (in press). ‘‘Teachers in this school
are able to get through to difficult students’’ is a
sample item from the collective efficacy scale.
5.5. School-level control variables
We were interested in controlling for other
school contextual factors that might, in addition
to collective efficacy, be related to differences
among schools in teacher efficacy. Therefore, we
used data provided by the district to construct
measures of the proportion of students who
received a free or reduced-price lunch and the
minority student concentration. To produce a
mean prior achievement score for each school,
we aggregated 3rd grade students’ normal curve
equivalent scores in mathematics on the 7th
Edition of the Metropolitan Achievement Test
administered one year before we surveyed faculty.
In addition, we obtained a measure of school size
that represented each school’s official student
attendance as reported to the state department of
education for the year.
5.6. Multilevel analyses
An important consideration for any study that
examines the relationship between teacher and
collective efficacy is the unit of analysis problem.
Conventional methods require single-level analysis
which leads to the conceptual and empirical
problems associated with either examining teacher-level perceptions of collective efficacy or
school-level perceptions of teacher efficacy. To
avoid these problems, we employed hierarchical
linear modeling (HLM) (Bryk & Raudenbush,
1992). HLM avoids the aggregation bias, misestimated standard errors, and heterogeneity of
regression problems that sometimes compromise
the results of ordinary least squares regression
analyses of data in which (typically) one or more
individual-level characteristics are aggregated to
the group level.
5.7. Unconditional model
Since we were interested in predicting variation
between schools in teacher efficacy with schoollevel characteristics, we began our analysis with an
unconditional multilevel model of the variation in
teacher efficacy. This analysis allowed us to
determine that there was statistically significant
variation in teacher efficacy among the schools we
sampled. Therefore, we proceeded with our multilevel modeling to explain this variation as a
function of school contextual features and collective efficacy.
5.8. Predicting variation among schools in teacher
efficacy
The multilevel modeling approach employed to
investigate the between-school variation in teacher
efficacy was the means-as-outcomes analysis (Bryk
& Raudenbush, 1992). Using this approach we
were able to test several school contextual features
and collective efficacy as predictors of teacher
efficacy. At the teacher-level, variation within
schools in teacher efficacy was modeled as follows:
ð1Þ Teacher level: YðTEACHER
EFFICACYÞij
¼ B0j þ rij :
Since teachers were anonymous, no teacher-level
variables were included in the level-1 model, hence
level-1 variation in teacher efficacy is estimated by
rij : Since we modeled variation among rather than
within schools, rij was constant throughout our
analyses.
At the school-level, we first separately tested the
effects of each of our school-level contextual
controls (mean SES, minority concentration,
school size, and mean prior achievement) and
collective efficacy. This process allowed us to
identify which of these variables was significantly
related to variation among schools in teacher
efficacy. As an example, the structural equation
specifying variation among schools in teacher
R.D. Goddard, Y.L. Goddard / Teaching and Teacher Education 17 (2001) 807–818
efficacy as a function of mean SES was as follows:
ð2Þ School level: B0j ¼ g00 þ g01 WðMeanSESÞj þ m0j :
Models of the effects of minority concentration,
school size, mean prior achievement, and collective
efficacy were analogous. Next, we combined the
statistically significant predictors identified in the
first five models to produce a parsimonious
combined model that explained variation among
schools in teacher efficacy. The results of these
analyses are reported below.
6. Results
The main question in this research concerned the
effect of collective efficacy on teacher efficacy. We
decided that to address this question, we should
control for school contextual factors other than
collective efficacy that might be related to betweenschool variation in teacher efficacy. Our contextual
control variables included mean SES, mean prior
mathematics achievement, minority concentration,
and school size. Descriptive statistics for these
variables and others relevant to the study are
reported in Table 1. Notably, there is considerable
variation across teachers in teacher efficacy with
scores ranging from 1.40 (low teacher efficacy) to
6.0 (high teacher efficacy). The extent to which this
variability in teacher efficacy is systematically
associated with school membership was the focus
of the multilevel analyses that followed. Also, over
half the students in the schools surveyed were
minority and nearly two-thirds received a free or
reduced price lunch while the average normal curve
equivalent score in mathematics across the 47
sampled schools was slightly below the 44th
percentile. Thus, the urban schools sampled for
this research served a large number of minority and
poor students with slightly below average achievement. Correlations among select school-level variables are reported in Table 2.
6.1. Psychometric analysis and construction of
efficacy scales
Of the 452 sampled teachers, 438 (approximately 97%) returned usable answers to the
813
teacher efficacy questionnaire. Teacher responses
to the teacher efficacy scale items were submitted
to a principal axis factor analysis. One factor was
extracted with an eigenvalue of 2.106 explaining
42.12% of the variance. Factor loadings ranged
from 0.62 to 0.70. The alpha coefficient of
reliability for the teacher efficacy scale was 0.79.
This is consistent with the reliability of scores on
the scale in previous research (e.g., 0.77 in Hoy &
Woolfolk, 1993; 0.81 in Woolfolk, Rosoff, & Hoy,
1990; 0.82 in Woolfolk & Hoy, 1990). The teacher
efficacy score for each teacher was constructed as
the mean of the teacher’s responses to all items in
the teacher efficacy scale.
The collective efficacy score for each school was
obtained from a process that began by calculating
a mean score for each of the collective efficacy
scale items for each school. This produced 21
mean item scores for each school. At the school
level, these 21 items were submitted to a principal
axis factor analysis. A single factor was extracted
with an eigenvalue of 12.00 that explained 57.13%
of the variance. Factor loadings ranged from 0.61
to 0.93, with all but five items loading above 0.70.
The alpha coefficient of reliability for scores on the
collective efficacy scale was 0.96. Next, within each
school the 21 mean scores were averaged to yield
an overall collective efficacy score for each school.2
6.2. Multilevel results
An unconditional analysis (one-way ANOVA
with random effects) indicated significant variation
among schools in teacher efficacy. Specifically,
variation among the school means for teacher
efficacy (Var (B0j )) was 0.04758 (w2=67.33, df=46,
po0:05). This finding confirmed that teacher
efficacy does indeed vary systematically with
school characteristics. Given that teacher efficacy
varied significantly among schools, we continued
our analysis by building a school-level model to
explain that variation.
2
Interested readers are directed to a validation study for a 12
item short form of the collective efficacy scale by Goddard (in
press) that may be used in lieu of the 21 item scale employed for
this research.
814
R.D. Goddard, Y.L. Goddard / Teaching and Teacher Education 17 (2001) 807–818
Table 1
Descriptive statistics for teacher and school variablesa
Mean
Teacher level (n ¼ 452):
Teacher efficacyb
School level (n ¼ 47):
Collective efficacyb
School size
Faculty size
Faculty members surveyed
Proportion of students receiving a free or reduced-price lunch
Proportion of minority students
Prior mathematics achievement
Std dev.
Min.
Max.
4.66
0.77
1.40
6.00
4.31
401.40
21.24
9.62
0.62
0.56
43.86
0.48
107.26
5.51
2.58
0.20
0.28
10.17
3.43
229
13
5
0.10
0.08
29.21
5.30
710
37
15
0.89
1.00
73.40
a
Variables used in multilevel analyses were subsequently standardized to a mean of zero and standard deviation of one.
Teacher efficacy and collective efficacy were measured using Likert-type scales with items scaled from 1 (strongly disagree) to 6
(strongly agree).
b
Table 2
Correlations among school-level variables (n ¼ 47)
Variable
Collective efficacy
School size
SES
Minority concentrn.
Prior math achieve.
Collective efficacy
School size
Proportion low SES
Minority concentration
Prior mathematics achievement
F
0.279
0.726a
0.479a
0.731a
F
0.188
0.058
0.060
F
0.521a
0.867a
F
0.550a
F
a
po0.01.
We began by first separately testing each of the
school-level contextual variables and collective
efficacy as predictors of between-school variation
in teacher efficacy. To facilitate comparison of
their effect sizes, all school-level variables were
standardized to a mean of zero and a standard
deviation of one. These five analyses identified
mean SES, mean prior mathematics achievement,
and collective efficacy as significant predictors
of variation among schools in teacher efficacy
while school size and minority concentration
were
statistically
unrelated
to
teacher
efficacy.
In the final step of the multilevel examination we
considered the effects of the three statistically
significant predictors simultaneously. The findings
of this analysis (Combined Model) confirmed the
main hypothesis of the study. The results indicated
that a one standard deviation increase in collective
efficacy was associated with a 0.248 standard
deviation increase in teacher efficacy. Importantly,
when considered together, only collective efficacy
was a significant predictor of differences between
schools in teacher efficacy; neither mean prior
achievement nor mean SES were significant predictors of teacher efficacy in the combined model.
Thus, variation in collective efficacy explained
variance in teacher efficacy above and beyond that
accounted for by our school contextual controls.
The effect sizes and associated p-values from these
analyses are reported in Table 3.
Another interesting question that we investigated in our multilevel analyses was the extent to
which our various models explained betweenschool variation in teacher efficacy. Table 4
displays the proportion of variance explained by
each model with a statistically significant predictor. Notably, mean SES and mean prior
R.D. Goddard, Y.L. Goddard / Teaching and Teacher Education 17 (2001) 807–818
815
Table 3
Prediction of variation in teacher efficacy among school means with selected school characteristics (n ¼ 438 teachers in 47 schools)
Model
1
Intercept
Proportion minority
Proportion low SES
Number of students
Prior math achievement
Collective efficacy
a
b
0.005
0.082
2
3
0.008
0.011
4
5
Combined
0.009
0.020
0.023
F
0.042
F
0.035
0.247a
0.104b
0.045
0.108b
0.191a
po0.01.
po0.05.
achievement alone each explain just under 25% of
the variation among schools in teacher efficacy.
Moreover, after accounting for the effects of mean
SES (Model 2) and mean prior achievement
(Model 4), in each case the remaining variation
among schools in teacher efficacy is statistically
non-zero. In other words, neither mean SES nor
mean prior achievement explained all of the
systematic variation among schools in teacher
efficacy. The case is different, however, when
assessing the relationship between collective efficacy and teacher efficacy (Model 5). As shown in
Table 4, collective efficacy accounts for nearly
three-fourths of the variation among schools in
teacher efficacy. In addition, when collective
efficacy is considered, the remaining variance is
not statistically different from zero. In other
words, collective efficacy predicts all of the
variation in teacher efficacy detected in our
sample. The results of the combined model are
consistent with those of Models 2, 4, and 5.
Namely, when considered with mean SES and
mean prior achievement, collective efficacy is the
lone statistically significant predictor of variation
among schools in teacher efficacy and the model
explains all of the between-school variation in
teacher efficacy.
7. Discussion
A great number of researchers have investigated
the correlates of teacher efficacy and the relation-
ship of this important construct to student
achievement. Far fewer have considered how
school context is related to teacher efficacy. The
findings of this study indicate that teacher efficacy
does vary systematically among schools, at least in
the urban elementary schools we sampled. Hence,
organizations appear to play a role in teachers’
reported levels of efficacy. Next, our results
provide initial evidence that the variation between
schools in teacher efficacy may be explained by the
collective efficacy of a school. Teacher efficacy was
higher in the schools where collective efficacy was
higher. Indeed, after accounting for the effects of
mean prior achievement and mean SES, a one
standard deviation increase in collective efficacy
was associated with a quarter standard deviation
increase in teacher efficacy. Moreover, collective
efficacy explained all of the variation among
schools in teacher efficacy in the schools we
sampled. When considered together with the
effects of school contextual features such as mean
SES and mean prior achievement, collective
efficacy was the only significant predictor of
teacher efficacy differences among schools.
It is not surprising that there exists a significant
positive relationship between teacher and collective efficacy. To be sure, teachers ‘‘are not social
isolates immune to the influence of those around
them’’ (Bandura 1997, p. 469). As postulated by
social cognitive theory, social influence shapes selfefficacy. Where teachers tend to think highly of the
collective capability of the faculty, they may sense
an expectation for successful teaching and hence
816
R.D. Goddard, Y.L. Goddard / Teaching and Teacher Education 17 (2001) 807–818
Table 4
Proportion of variation in teacher efficacy among schools explained and remaining for models reported in Table 3 (n ¼ 438 teachers in
47 schools)
Model
a,b
Proportion of variation explained
Between-school variation remaininga
1
2
3
4
5
Combined
F
F
0.247
0.03581c
F
F
0.231
0.03657d
0.735
0.01263e
0.823
0.00815f
a
Proportion of variation in teacher efficacy explained and remaining reported only for models with statistically significant predictors.
Proportion calculated as the reduction in variation among schools in teacher efficacy found in the unconditional multilevel analysis
(0.04758, Chi-square=67.33, df=46, po0:05).
c
Chi-square=62.30, df=45, po0:05:
d
Chi-square=62.44, df=45, po0:05:
e
Chi-square=50.48, df=45, p ¼ 0:266:
f
Chi-square=49.13, df=43, p ¼ 0:241:
b
work to be successful themselves. Conversely,
where collective efficacy is low, it is less likely
that teachers will be pressed by their colleagues
to persist in the face of failure or that they will
change their teaching when students do not
learn.
A minor limitation of this study was that
teachers were guaranteed complete anonymity in
the survey process. For this reason, no teacher
demographic data were collected. Other researchers may wish to consider the possibility of
obtaining teacher demographic data to examine
the relative effects of these variables on teacher
efficacy in a multilevel analysis that also includes
collective efficacy at the school level. In addition,
although we achieved statistical control by sampling only one type of school (elementary) within
one urban school district, this also means that the
schools we sampled are probably more homogenous in terms of SES, minority concentration,
and student achievement than would be found in
the general population. Therefore, we are not
convinced by the results reported here that
collective efficacy will explain all of the betweenschool variation in teacher efficacy in all types
of schools. Future researchers might wish to
examine the relationship between teacher and
collective efficacy in other settings (e.g., non-urban
secondary schools) to determine whether the
results extend to demographically dissimilar populations.
Our results do have import, however, for the
great many teachers and students who work and
learn in urban school districts like the one we
sampled. Indeed, our findings suggest that in this
urban district, collective perceptions of faculty
capability were predictive of the differences among
schools in the perceptions that teachers held of
their own self-capability. Such a result helps to
show that collective efficacy is an important school
contextual feature that is systematically related to
teacher efficacy, adding to this line of teacher
efficacy research. Moreover, building collective
efficacy in schools may offer a new possibility for
raising teacher efficacy and perhaps at least
lessening the declines in teacher efficacy that are
sometimes experienced by teachers when they
leave their preservice programs (Chester &
Beaudin, 1996; Tschannen-Moran, Woolfolk
Hoy, & Hoy, 1998). Future researchers might
address whether declines in teacher efficacy are
attenuated for teachers who join schools with
relatively high collective efficacy.
The results are also important for the field of
organizational studies as they suggest one way in
which organizational characteristics may influence
member performance. From the perspective of
organizational improvement, Bandura (1997) has
suggested that a strong leader who can ‘‘unite the
community for common cause’’ (p. 501) and who
empowers the faculty may be able to increase the
collective efficacy of a school. Strong leadership
R.D. Goddard, Y.L. Goddard / Teaching and Teacher Education 17 (2001) 807–818
and empowerment may indeed be aspects of
organizational life that can build collective efficacy. Whether changes in collective efficacy lead to
changes in teacher efficacy is, however, an
unanswered question. Future researchers might
wish to examine this question of causality. One
way to do this is to study schools during a period
of intense change (Bandura, 1997) to determine
whether changes in collective efficacy do cause
changes in teacher efficacy. One possibility that is
consistent with the correlational evidence in this
study is that teacher and collective efficacy have a
reciprocal relationshipFa change in one may lead
to changes in the other. Such a finding would
have import regarding the manner in which
teachers and schools influence one another and
consequently the students they serve. Hoy &
Woolfolk (1993), for example, found that ‘‘schools
promoted personal teaching efficacy when
teachers perceived that their colleagues (a) set
high but achievable goals, (b) create an orderly
and serious learning environment, and (c) respect
academic excellence’’ (p. 365). They also found
that teacher efficacy was higher when school
principals are perceived as having influence
with their superiors. Such findings suggest that
features of school organization and leadership can
serve to develop teacher efficacy. At a minimum,
we believe the findings of this study illuminate
possibilities for building teacher efficacy through
school improvement and avenues for a great
deal of future research that can further our
understanding of both teacher and collective
efficacy.
References
Allinder, R. M. (1994). The relationship between efficacy and
the instructional practices of special education teachers and
consultants. Teacher Education and Special Education, 17,
86–95.
Anderson, R., Greene, M., & Loewen, P. (1988). Relationships
among teachers’ and students’ thinking skills, sense of
efficacy, and student achievement. Alberta Journal of
Educational Research, 34(2), 148–165.
Armor, D., Conroy-Oseguera, P., Cox, M., King, N., McDonnell, L., Pascal, A., Pauly, E., & Zellman, G. (1976).
Analysis of the school preferred reading program in selected
Los Angeles minority schools [Report No. R-2007-LAUSD;
817
ERIC Document Reproduction No. 130 243]. Santa Monica,
CA: Rand Corporation.
Ashton, P. T., & Webb, R. B. (1986). Making a difference:
Teachers’ sense of efficacy and student achievement. New
York: Longman.
Bandura, A. (1993). Perceived self-efficacy in cognitive development and functioning. Educational Psychologist, 28(2),
117–148.
Bandura, A. (1997). Self-efficacy: The exercise of control. New
York: W.H. Freeman and Company.
Bryk, A. S., & Raudenbush, S. W. (1992). Hierarchical linear
models: Applications and data analysis methods. Newbury
Park, CA: Sage Publications, Inc..
Chester, M. D., & Beaudin, B. Q. (1996). Efficacy beliefs of
newly hired teachers in urban schools. American Educational
Research Journal, 33(1), 233–257.
Coleman, J. S. (1985). Schools and the communities they serve.
Phi Delta Kappan, 66, 527–532.
Coleman, J. S. (1987). Norms as social capital. In G.
Radnitzky, & P. Bernholz (Eds.), Economic imperialism:
the economic approach applied outside the field of economics.
New York: Paragon House Publishers.
Coleman, J. S. (1990). Foundations of social theory. Cambridge,
Mass: Harvard University Press.
Czerniak, C. M., & Schriver, M. L. (1994). An examination of
preservice science teachers’ beliefs and behaviors as related
to self-efficacy. Journal of Science Teacher Education, 5(3),
77–86.
Da Costa, J. L., Riordan, G. (1996, April). Teacher efficacy and
the capacity to trust. Paper presented at the annual meeting
of the American Educational Research Association, New
York.
DeForest, P. A., & Hughes, J. N. (1992). Effect of teacher
involvement and teacher self-efficacy on ratings of consultant effectiveness and intervention acceptability. Journal
of Educational and Psychological Consultation, 3, 301–316.
DeMesquita, P. B., & Drake, J. C. (1994). Educational reform
and the self-efficacy beliefs of teachers implementing
nongraded primary school programs. Teaching and Teacher
Education, 10, 291–302.
Enochs, L. G., Scharmann, L. C., & Riggs, I. M. (1995). The
relationship of pupil control to preservice elementary
science teacher self-efficacy and outcome expectancy.
Science Education, 79(1), 63–75.
Gibson, S., & Dembo, M. (1984). Teacher efficacy: A construct
validation. Journal of Educational Psychology, 76(4), 569–
582.
Goddard, R. D. (2000, April). Collective efficacy and
student achievement. Paper presented at the annual meeting
of the American Educational Research Association, New
Orleans.
Goddard, R. D. A theoretical and empirical analysis of the
measurement of collective efficacy: The development of a
short form. Educational and Psychological Measurement, in
press.
Goddard, R. D., Hoy, W. K., & Woolfolk, A. (2000). Collective
teacher efficacy: Its meaning, measure, and effect on student
818
R.D. Goddard, Y.L. Goddard / Teaching and Teacher Education 17 (2001) 807–818
achievement. American Education Research Journal, 37(2),
479–507.
Goddard, R.D., Tschannen-Moran, M., & Hoy, W.K. (in
press). Teacher trust in students and parents: A multilevel
examination of the distribution and effects of teacher trust
in urban elementary schools. Elementary School Journal.
Guskey, T. R. (1988). Teacher efficacy, self-concept, and
attitudes toward the implementation of instructional innovation. Teaching and Teacher Education, 4(1), 63–69.
Hoover-Dempsey, K., Bassler, O. C., & Brissie, J. S. (1987).
Parent involvement: Contributions of teacher efficacy,
school socioeconomic status, and other school characteristics. American Educational Research Journal, 24,
417–435.
Hoover-Dempsey, K. V., Bassler, O. C., & Brissie, J. S. (1992).
Parent efficacy, teacher efficacy, and parent involvement:
Explorations in parent-school relations. Journal of Educational Research, 85(5), 287–294.
Hoy, W. K., & Woolfolk, A. E. (1993). Teachers’ sense of
efficacy and the organizational health of schools. The
Elementary School Journal, 93(4), 355–372.
Lee, V. E., Dedrick, R., & Smith, J. (1991). The effect of the
social organization of schools on teachers’ efficacy and
satisfaction. Sociology of Education, 64, 190–208.
Moore, W., Esselman, M. (1992). Teacher efficacy, power,
school climate and achievement: A desegregating districts
experience. Paper presented at the annual meeting of the
American Educational Research Association, San
Francisco.
Pajares, F. (1996, April). Current directions in self research:
Self-efficacy. Paper presented at the annual meeting of the
American Educational Research Association, New York.
Pajares, F. (1997). Current directions in self-efficacy research.
In M. L. Maehr, & P. R. Pintrich (Eds.), Advances in
motivation and achievement (pp. 1–49). Greenwich, CT: JAI
Press.
Ross, J. A. (1992). Teacher efficacy and the effect of coaching
on student achievement. Canadian Journal of Education,
17(1), 51–65.
Smylie, M. A. (1988). The enhancement function of staff
development: organizational and psychological antecedents
to individual teacher change. American Educational Research Journal, 25, 1–30.
Tschannen-Moran, M., Woolfolk Hoy, A., & Hoy, W. K.
(1998). Teacher efficacy: Its meaning and measure. Review of
Educational Research, 68, 202–248.
Woolfolk, A., & Hoy, W. K. (1990). Prospective teachers’ sense
of efficacy and beliefs about control. Journal of Educational
Psychology, 82, 81–91.
Woolfolk, A. E., Rosoff, B., & Hoy, W. K. (1990). Teachers’
sense of efficacy and their beliefs about managing students.
Teaching and Teacher Education, 6, 137–148.