Modeling teacher self - Efficacy as a function of peer observation, administrative feedback, job satisfaction, and work enjoyment

Hue University Journal of Sciences: Social Science and Humanities
ISSN 2588–1213
Vol. 128, No. 6B, 2019, Tr. 71–83, DOI: 10.26459/hueuni-jssh.v128i6B.5265
* Corresponding: chau.nguyen@ttu.edu 
Submitted: 24–05–2019; Revised: 05–06–2019; Accepted: 21–06–2019. 
MODELING TEACHER SELF-EFFICACY AS A FUNCTION 
OF PEER OBSERVATION, ADMINISTRATIVE FEEDBACK, 
JOB SATISFACTION, AND WORK ENJOYMENT 
Nguyen Phuoc Hong Chau1, 2, Aaron Samuel Zimmerman1, * 
1 College of Education, Texas Tech University, 3008 18th Street, Lubbock, TX 79409‐1071, USA 
2 Phu Xuan University, 28 Nguyen Tri Phuong, Hue, Vietnam 
Abstract. This study used a large‐scale, international data set – the Organization for Economic Co‐
operation and Development (OECD) – Teaching and Learning International Survey (TALIS) 2013, consist‐
ing of 14,583 teachers from 34 countries – to examine the manner in which feedback from administrators, 
time spent observing colleagues’ classes, job satisfaction, and work enjoyment predicted teachers’ instruc‐
tional self‐efficacy. To guide the present study, Bandura’s (1986, 1997) part of the social cognitive theory – 
that is, self‐efficacy theory – is utilized. We adopted Bandura’s self‐efficacy theory as the framework, for it 
provides a valuable lens through which we could identify the predictors of teacher self‐efficacy to include 
in the model investigated in this study. The results of this study suggest that feedback from administrators 
was not a significant predictor of teacher self‐efficacy for instruction, whilst peer observation, job satisfac‐
tion, and work enjoyment were estimated as being significant predictors. These results have implications 
for practice – specifically, how teachers and school leaders should cultivate teachers’ self‐efficacy for in‐
struction – and future research. 
Keywords: teacher self‐efficacy, administrative feedback, peer observation, job satisfaction, work enjoy‐
ment 
1. Introduction 
Grounded in Bandura’s (1986) social cognitive theory (see next section), Skaalvik and 
Skaalvik (2010) conceptualized teacher self‐efficacy as “individual teachers’ beliefs in their own 
ability to plan, organize, and carry out activities that are required to attain given educational 
goals” (p. 1059). A large body of existing research has shown that teacher self‐efficacy occupies 
a crucial role in enhancing the effectiveness of teacher teaching and student learning. Specifical‐
ly, on the one hand, teacher self‐efficacy correlates positively with teacher well‐being (e.g., 
Brouwers & Tomic, 2000; Zee & Koomen, 2016), teacher professional practices (e.g., Gibson & 
Dembo, 1984), teacher job satisfaction (e.g., Avanzi et al., 2013; Caprara, Barbaranelli, Borgogni, 
& Steca, 2003), teacher commitment to teaching (e.g., Chesnut & Burley, 2015; Coladarci, 1992; 
Nguyen Phuoc Hong Chau, Aaron Samuel Zimmerman Vol. 128, No. 6B, 2019
72 
Evans & Tribble, 1986), student motivation (e.g., Dembo & Gibson, 1985), and student achieve‐
ment (e.g., Anderson, Greene, & Loewen, 1988; Kassen & Tze, 2014; Ross, 1992; Tschannen‐
Moran & Johnson, 2011). On the other hand, teacher self‐efficacy has a negative correlation with 
teacher burnout (e.g., Betoret, 2006; Chwalisz, Altmaier, & Russell, 1992; Evers, Brouwers, & 
Tomic, 2002; Skaalvik & Skaalvik, 2009). Morris and Usher (2011) contended, “Although a 
growing body of research attests to the benefits associated with teaching self‐efficacy, less is 
known about how teaching self‐efficacy is cultivated” (pp. 232–233). Responding to this gap in 
the research literature, the present study was designed to examine the manner in which feed‐
back from administrators, time spent observing colleagues’ classes, job satisfaction, and work 
enjoyment predicted teacher self‐efficacy for instruction. 
2. Theoretical framework 
This study was guided by Bandura’s (1986, 1997) part of the social cognitive theory – that 
is, self‐efficacy theory which is depicted as follows: 
2.1. Definition of self-efficacy 
In his social cognitive theory, Bandura (1986) defines self‐efficacy as “people’s judge‐
ments of their capabilities to organize and execute courses of action required to attain designat‐
ed types of performances” (p. 391). Bandura emphasizes, “[Self‐efficacy] is concerned not with 
the skills one has but with judgements of what one can do with whatever skills one possesses” 
(p. 391). 
2.2. Principal sources of self-efficacy 
Bandura (1997) theorizes that self‐efficacy beliefs are constructed as individuals interpret 
information from four major sources: (a) enactive mastery experience, (b) vicarious experience, 
(c) verbal persuasion, and (d) physiological and affective states. Of these four self‐efficacy 
sources, enactive mastery experience, Bandura postulates, is the most influential since it is de‐
rived from one’s own personal experience, hence is the most authentic evidence of one’s capa‐
bilities. He posits that performance successes would raise one’s self‐efficacy; whereas repeated 
performance failures would undermine it. In respect of the second source, Bandura explains 
that one’s self‐efficacy is also influenced by his/her vicarious experience – that is, visualizing 
other people perform, through which he/she can compare his/her performance with that of oth‐
ers and form a perception of his/her own capabilities. Regarding verbal persuasion, Bandura 
claims one’s self‐efficacy could be bolstered if others verbally persuade him/her that, for in‐
stance, he/she is capable to successfully perform a certain task. It should, however, be noted that 
the effectiveness of verbal persuasion (e.g., praise and evaluative feedback) is mediated not only 
Jos.hueuni.edu.vn Vol. 128, No. 6B, 2019
73 
by who provides it, but also by the way in which the message is constructed (Bandura, 1997; 
Morris, Usher, & Chen, 2017; Pajares, 2006). For example, feedback from peers or inexperienced 
people tends to exert less influence upon one’s efficacy beliefs compared with that from profes‐
sional experts or those who are experienced (Bandura, 1997). Or, as found by Parajes (2016), 
“knee‐jerk praise or empty inspirational homilies” (p. 349) are unlikely to be an effective means 
of nurturing one’s positive beliefs about their capabilities. Finally, one’s physiological and affec‐
tive states are also important indicators one uses to judge his/her efficacy. Such states comprise, 
for example, “stress, fatigue, anxiety, and mood” (Morris et al., 2017, p. 798). For instance, when 
one is suffering from fatigue or pains, his/her self‐efficacy tends to be lowered compared with 
when he/she is in a good mood. We summarize Bandura’s theoretical framework for sources of 
self‐efficacy as follows: 
Figure 1. Theoretical framework for sources of self‐efficacy (Bandura, 1997) 
This theoretical framework informs the variables included in the model investigated in 
this study. Said differently, the present study addresses the following research question: How 
do feedback from one’s administrators (verbal persuasion), observation of one’s peers (vica‐
 Physiological and affec‐
tive states 
.. 
e.g., stress, fatigue, an‐
xiety, enjoyment, or 
happiness 
 Social/Verbal persuasions 
. 
e.g., praise, appraisal, or 
evaluative feedback 
Vicarious experience 
. 
Observations of the self, 
others, or models 
Enactive mastery experience 
.. 
Performance accomplish‐
ments/successes 
Nguyen Phuoc Hong Chau, Aaron Samuel Zimmerman Vol. 128, No. 6B, 2019
74 
rious experience), satisfaction with one’s own performance (enactive mastery experience), and 
enjoyment of one’s own work (physiological and affective states) contribute to the development 
of teachers’ instructional self‐efficacy? 
3. Literature review 
In the existing literature, there is a large body of both quantitative and qualitative re‐
search examining the influence of Bandura’s four hypothesized sources on teacher self‐efficacy. 
The majority of the extant studies indicate that all four sources play roles in teacher self‐efficacy 
development (e.g., Hendricks, 2016; Pfitzner‐Eden, 2016; Phan & Locke, 2015; Poulou, 2007). 
However, those study results are not consistent with respect to the strength of each source. For 
example, Morris and Usher (2011), Pfitzner‐Eden (2016), Poulou (2007), and Tschannen‐Moran 
and McMaster (2009) found that mastery experience wields the greatest effect on teacher self‐
efficacy. In contrast, the studies by Johnson (2010) and Mills (2011) demonstrated that vicarious 
experience is the most influential source of teaching self‐efficacy. Meanwhile, other researchers 
have observed that verbal persuasion is the predominant source that can raise or diminish 
teacher self‐efficacy (Aydin & Hoy, 2005; Milner, 2002; Milner & Hoy, 2003; Phan & Locke, 
2015). Completely different from the above‐mentioned findings, the study by Palmer (2006) 
indicated that most of Bandura’s sources of self‐efficacy, including enactive mastery experience, 
vicarious experience, and verbal persuasion are not significant factors contributing towards 
teacher self‐efficacy. The main source of teacher self‐efficacy, found in Palmer’s study, is cogni‐
tive pedagogical mastery – that is, “successes in understanding how to teach” (p. 349). Palmer 
considered this factor (cognitive pedagogical mastery) a potential additional source of self‐
efficacy in relation to those hypothesized by Bandura. 
The literature suggests the inconsistencies regarding the impact level of Bandura’s 
sources on teacher efficacy beliefs. In addition, there are still few studies investigating this issue 
in international or cross‐nations contexts. In response to this need, the present study was con‐
ducted utilizing a large‐scale, international data set with the participation of teachers from 34 
countries to reexamine this issue. 
4. Methodology 
4.1. Statistical method 
To examine how feedback from one’s administrators, observation of one’s peers, satisfac‐
tion with one’s own performance, and enjoyment of one’s own work contribute to the devel‐
opment of teachers’ instructional self‐efficacy, multiple linear regression was employed. Urdan 
(2017) writes that multiple linear regression “allows researchers to examine the nature and 
Jos.hueuni.edu.vn Vol. 128, No. 6B, 2019
75 
strength of the relations between variables, the relative predictive power of several independent 
variables on a dependent variable” (p. 183). In addition, as explained by Field (2013), this form 
of statistical modeling is “a method in which all predictors are forced into the model simulta‐
neously” (p. 322). The predictors used in the present study (i.e., feedback from one’s adminis‐
trators, observation of one’s peers, satisfaction with one’s own performance, and enjoyment of 
one’s own work) were selected based entirely on the self‐efficacy theoretical framework by 
Bandura (1997). 
4.2. Data sources and variables 
The sample of this study, which was drawn from the OECD Teaching and Learning In‐
ternational Survey (TALIS) 2013 data set, consisted of 14,583 teachers from 34 countries (OECD, 
2014). The survey questions included in the model as independent variables were presented in 
Table 1. These questions were answered on a Likert scale, with Items 46‐TT2G46I, 31‐TT2G31C, 
and 46‐TT2G46E using a 4‐point scale from strongly disagree to strongly agree, and Item 33‐
TT2G33B asking about the frequency of peer observation using a 6‐point scale from never to once 
a week or more. Note that the TALIS 2013 employed a 4‐point scale from strongly disagree to 
strongly agree in lieu of a 5‐point scale. However, this is not a concern because, as admitted by 
Dillman, Smyth, and Christian (2014): 
[W]hile many people agonize over the decision of whether or not to offer a midpoint 
[such as “neutral” or “neither disagree nor agree"], the literature suggests whether one of‐
fers a midpoint has little effect on the resulting data quality and conclusions drawn from 
the data. (p. 154) 
Table 1. Independent variables included in the model 
Independent Variables Questions/Items Scales 
Performance satisfaction 
46. How strongly do you agree or disag‐
ree with the following statement? 
TT2G46I – I am satisfied with my perfor-
mance in this school. 
1 – Strongly disagree 
2 – Disagree 
3 – Agree 
4 – Strongly agree 
Peer observation 33. On average, how often do you do 
the following in this school? TT2G33B – 
Observe other teachers’ classes and provide 
feedback. 
1 – Never 
2 – Once a year or less 
3 – 2‐4 times a year 
4 – 5‐10 times a year 
5 – 1‐3 times a month 
6 – Once a week or more 
Administrative feedback 31. How strongly do you agree or disag‐
ree with the following statement about 
this school? TT2G31C – Teacher appraisal 
1 – Strongly disagree 
2 – Disagree 
3 – Agree 
Nguyen Phuoc Hong Chau, Aaron Samuel Zimmerman Vol. 128, No. 6B, 2019
76 
and feedback are largely done to fulfil ad-
ministrative requirements. 
4 – Strongly agree 
Work enjoyment 46. How strongly do you agree or disag‐
ree with the following statement? 
TT2G46E – I enjoy working at this school. 
1 – Strongly disagree 
2 – Disagree 
3 – Agree 
4 – Strongly agree 
The dependent variable in the model was assessed through the following survey ques‐
tion, measured on a 4‐point Likert scale: 
 Table 2. Dependent variable included in the model 
Dependent Variable Question/Item Scale 
Teachers’ instructional self‐
efficacy 
34. In your teaching, to what extent can 
you do the following? TT2G34L – Imple-
ment alternative instructional strategies in 
my classroom. 
1 – Not at all 
2 – To some extent 
3 – Quite a bit 
4 – A lot 
Concerning the validity of self‐efficacy measures, Bandura (2006) emphasizes, “Efficacy 
items should accurately reflect the construct. Self‐efficacy is concerned with perceived capabili‐
ty. The items should be phrased in terms of can do rather than will do. Can is a judgment of ca‐
pability; will is a statement of intention” (p. 308, italics original). Furthermore, Bandura (2006) 
maintains, “Perceived self‐efficacy is a judgment of capability to execute given types of perfor‐
mances” (p. 309). That is, as elucidated by Klassen and Usher (2010), self‐efficacy meas‐
ures/scales need to adhere to the tenet of specificity; they should be constructed in the way that 
can measure one’s judgements of their capabilities “to carry out specific tasks or in a specific 
domain” (p. 19). The item/question used in this study (which measures teachers’ self‐efficacy for 
implementing alternative instructional strategies) is congruent with Bandura’s (1986) definition 
of self‐efficacy (i.e., one’s beliefs about their competence to produce given attainments rather 
than their actual capability or intention) as well as with Bandura’s (2006) principle of specificity. 
This item can, therefore, be seen as a valid measure of self‐efficacy within the scope of the 
present study, for it “measure[s] what [it] purport[s] to measure” (Bandura, 2006, p. 318). 
4.3. Data analysis 
The data related to the five above‐mentioned variables were first screened to identify and 
evaluate the seriousness of missing cases. This first step produced the results that the variable 
job satisfaction had 3.7% cases missing, peer observation 3.7%, administrative feedback 7.3%, 
work enjoyment 3.8%, and teacher self‐efficacy for instructional strategies 3.7%. Because the 
sample size is large (14,583 participants), and all the variables each had less than 10% cases 
missing, Listwise was utilized to exclude those missing cases from the whole analysis (Mertler & 
Jos.hueuni.edu.vn Vol. 128, No. 6B, 2019
77 
Reinhart, 2017). Thereafter, the data were screened again to evaluate the normal distribution of 
the variables; the evaluation indicates that the normality of distributions of these five variables 
is tenable. Finally, multiple linear regression analysis was conducted employing SPSS to deter‐
mine how feedback from one’s administrators, observation of one’s peers, satisfaction with 
one’s own performance, and enjoyment of one’s own work contribute towards nurturing teach‐
ers’ instructional self‐efficacy. 
5. Results 
A multiple linear regression analysis was conducted to examine the manner in which 
feedback from administrators, time spent observing colleagues’ classes, job satisfaction, and 
work enjoyment predicted teacher self‐efficacy for instructional strategies. As previously men‐
tioned, all these four predictors were concurrently entered into the model. The regression re‐
sults demonstrated that tolerance was high (0.859 for job satisfaction, 0.980 for peer observation, 
0.947 for administrative feedback, and 0.832 for work enjoyment), suggesting that multi‐
collinearity was not a concern (Mertler & Reinhart, 2017). The results of the estimation of the 
coefficients for the linear regression model revealed that the overall model significantly pre‐
dicted the level of teacher self‐efficacy for instruction R2 = 0.062, R2adj. = 0.062, F(4,13154) = 
218.617, p < 0.001. However, as the results demonstrated, this model accounted for only 6.2 per‐
cent of the variance in teacher self‐efficacy for instructional strategies, leaving up to 93.8 percent 
of the variance in teachers’ instructional self‐efficacy unexplained. The results also indicated 
that out of the four independent variables included in the model, job satisfaction (β = 0.203, p < 
0.001) was the strongest predictor and positively related with teacher instructional self‐efficacy, 
followed by time spent observing other teachers’ classes (β = 0.089, p < 0.001), and work enjoy‐
ment (β = 0.046, p < 0.001). These two predictors were also positively associated with teacher 
instructional self‐efficacy. Interestingly, the coefficient estimate for feedback from administra‐
tors (β = –0.010, p = 0.226) was not statistically significant. A post hoc analysis was then also con‐
ducted using G*Power to calculate the power for this multiple regression design. The power 
achieved was 1.0, indicating a high level of power which reduces the chance of making a type II 
error (Mertler & Reinhart, 2017). A summary of regression coefficients is presented in Table 3. 
Table 3. Coefficients for model variables 
 B β t p Bivariate r Partial r 
Job satisfaction 0.281 0.203 22.279 <0.001 0.227 0.191 
Time spent observing colleagues’ 
classes 
0.042 0.089 10.465 <0.001 0.108 0.091 
Feedback from administrators –0.010 –0.010 –10.210 0.226 –0.052 –0.011 
Work enjoyment 0.054 0.046 4.974 <0.001 0.131 0.043 
Nguyen Phuoc Hong Chau, Aaron Samuel Zimmerman Vol. 128, No. 6B, 2019
78 
6. Discussion and conclusion 
The results indicate that first, teachers who are more satisfied with their teaching perfor‐
mance are most likely to have a higher level of instructional self‐efficacy than those who are not. 
Second, teachers who spend more time observing the teaching of their peers tend to be more 
confident in their ability for instruction as opposed to those spending less time in so doing. 
Third, teachers with more work enjoyment report having a higher level of self‐efficacy for in‐
struction compared with those with less work enjoyment. It is worth noting that of the three 
variables significantly contributing to informing teacher instructional self‐efficacy, teaching 
performance satisfaction is the most influential predictor; whereas, feedback from administra‐
tors does not appear to have an influence. These results are, on the one hand, consistent with 
Bandura’s (1997) theory as well as previous empirical research (e.g., Morris & Usher, 2011; 
Pfitzner‐Eden, 2016; Poulou, 2007; Tschannen‐Moran & Hoy, 2007) suggesting that enactive