This study explored how teachers’ peer support climate (PSC) and supervisory support climate (SSC) were related to teacher self-efficacy (TSE), teacher job satisfaction (TJS), teacher emotional exhaustion (TEE), and motivation to quit the teaching profession (MQTP) among teachers in the Philippines. Participants were 457 teachers in the Central Visayas Region. Structural equation modeling (SEM) indicated that MQTP varies as to self-efficacy, emotional exhaustion, and job satisfaction. Responses among all constructs do not vary among novice and experienced teachers except on TJS. The findings of the research advocate the proposed model. The model can guide future researchers in developing countries like the Philippines to explain teachers’ attrition caused by social support, efficacy factors, burnout, and job satisfaction.
While the task of educating the citizens of a country is considered the primary role of teachers, the question of whether these professionals are satisfied with their work environment is usually forgotten. The working conditions of teachers have been the least attended in terms of policy trends in many countries because of assumptions that teaching-learning can be carried out by devoted teachers even if factors like school conditions, social support climates, the kind of learners, and policy directions are not so appealing [1, 2]. It is general knowledge that many teachers leave the profession for nonretirement reasons. Stressful working conditions, teacher salary, and heavy workloads are among the most common reasons for teacher attrition. Stressors in work environment could lead to emotional exhaustion, affecting teachers’ motivation towards the profession [3, 4]. The most important element on work environment is the relationship with the school heads and peers. School heads have crucial role on job satisfaction and self-efficacy of teachers. For example, in a study of 300 teachers in Indonesia , Hartinah et al. revealed that principal leadership and work environment had direct effects on improving teacher’s performance.
Social support in the workplace is a concept derived from attribution, coping, equity, loneliness, and social comparison theories so that career plans, behavioral patterns, and career decisions may be fully understood. Social scientists began to consider the role of social support climate in explaining the multiple motivational influences of employees . For example, supervisors’ views of their subordinates’ career motivation associates with the support and empowerment the subordinates consider they receive from the supervisor . In the context of teaching, Skaalvik and Skaalvik  revealed a positive impact of job satisfaction with social support climate following through belonging as a mediating variable. Research consistently shows that social support from peers and supervisors is associated with some behavioral responses, for instance, motivation to do challenging tasks [9, 10], help-seeking behavior [11, 12], and persistence to work [13, 14].
The length of experience also needs to be elucidated in explaining latent factors such as job satisfaction, self-efficacy, and burnout among teachers. According to Gavish and Friedman , teaching is one of the most exhausting and stressful professions, especially among novice teachers. The first five years of teaching can be the most challenging part of the teaching experience, knowing that the most potent were the sources of teachers’ self-efﬁcacy beliefs and mastery experiences . The teaching profession is a challenge among beginners since the responsibility is more pronounced than their experienced colleagues . Despite efforts to make the teaching profession more rewarding, many countries still experienced shortages of qualified teachers or having problems with novice teacher attrition (e.g., [18, 19]).
With all of the mentioned antecedents, less is known about the sources of teacher’s motivation to stay in the profession with social support climate, specifically from peers and supervisors/principals. The notion that these relationships vary among novice and experienced teachers is less explored from a developing country’s perspective like the Philippines.
The purpose of this study was to explore how the social support from peers and supervisors among novice and experienced teachers in Central Philippines is related to emotional exhaustion, self-efficacy, job satisfaction, and motivation to leave the teaching profession.
2. Research Model and Hypothesis Development
To develop a model to explain the associations of peer and supervisory support climates to teacher’s self-efficacy, job satisfaction, emotional exhaustion, and motivation to leave the teaching profession, we review a variety of models comprising different sets of predictive factors. Among these are the theoretical models of Hackett et al., Klassen & Chiu, Maurer et al., and Skaalvik & Skaalvik [8, 10, 14, 20, 21]. The notion of including the experience classified as a novice and experienced teachers is on the hypothesized model presented in the work of Klassen and Chiu,  while our choice of making the peer and supervisory support climates as the main explanatory variables is on the model of Skaalvik and Skaalvik . Figure 1 presents the proposed model of the study.
2.1. PSC and SSC
The PSC refers to support for learning and development by peers, while SSC refers to supervisors’ work environment support. These support climates come with the availability of development and learning resources, work environment, and policies . The most common sources of happiness in the workplace were the so-called verbal persuasion, which can be social support from colleagues and supervisors and psychological arousal such as excitement when facing a challenge . Social support includes how the principals and co-teachers could provide emotional and instrumental support . A variety of literature revealed PSC to have a direct effect on TSE. For instance, supportive colleagues revealed acceptable fit measures with collective teacher efficacy and teacher self-efficacy . Skaalvik and Skaalvik  highlighted problems among peers and five other potential stressors showed to have negative effects on TSE.
On the contrary, SSC is a good predictor of TSE [19, 24, 25], TEE , and TJS . To summarize these findings, we learn that leadership in schools must be “stretched over people” so that supervisory support will have notable impacts on teachers’ self-efficacy and job satisfaction and reduce emotional exhaustion. From a developing country’s perspective, principal leadership and other social support in the work environment were found to directly affect teachers’ performance : H1 : PSC will significantly and positively influence TSE. H2 : SSC will significantly and negatively influence TEE. H3 : SSC will significantly and positively influence TJS.
In this study, TSE refers to teachers’ beliefs about what they are capable of doing or how sure they are that they can perform specific actions . The social cognitive theory of Bandura in contemporary educational research serves as the necessary foundation of TSE. Bandura’s theory of self-efﬁcacy advocated a belief in one’s ability to perform a speciﬁc task. The theory of Bandura states that people will be motivated to act if they are conﬁdent that they can perform that action successfully and believe that the action would have a favorable result. A growing body of literature demonstrates a causal link from TSE to TJS [27, 28]. Researchers reported that while people tend to enjoy the mastery of the workplace activities, they are more complacent with their job  and that the principals improved self-efficacy correlates to job satisfaction among its subordinates . A growing number of related literature about teacher self-efficacy underpin Bandura’s  “the exercise of control” theorizing relationships to TJS and MQTP [7, 10, 16, 28]. From the discussions above, we propose the following hypotheses: H4 : TSE will significantly and positively influence TJS. H5 : TSE will significantly and negatively influence MQTP.
We define TJS as the commitment and fulfillment in teaching , which is derived from teachers’ affective reactions to their teaching role  and reflected as teaching satisfaction . Teachers’ satisfaction in the workplace depends on some aspects, such as academic freedom, less workload, and enough time for class preparation. Due to this facet, TJS is one of the best antecedents of MQTP. For example, Toropova et al.  argue that teachers who are contented will demonstrate a higher job commitment and are likely to stay in the teaching profession. General findings revealed a negative association of TJS to MQTP [10, 19, 28, 31]: H6 : TJS will significantly and negatively influence MQTP.
Skaalvik and Skaalvik  and Sarıçam and Sakız  characterized TEE as a depletion of energy, debilitation, long-standing fatigue, and the feeling of being worn out. In some literature, teacher burnout is defined as a malfunctioned reaction to severe psychological and relational stressors, specifically on showing the depletion of emotional sources [33, 34]. Many studies investigated the relationships between TEE and other constructs. For example, Skaalvik and Skaalvik  reported a negative correlation TJS mediated by a performance goal structure. Performance goal structure is characterized by an emphasis on achievement and test scores and by a conceptualization of success as doing better than others. On the contrary, TEE directly affects MQTP [10, 19, 28, 31]. Stressors identified in explaining this phenomenon include weaker self-efficacy beliefs and receiving less supervisory support , job dissatisfaction , and stressful working environment . With these results, we propose the following hypotheses: H7 : TEE will significantly and negatively influence TJS. H8 : TEE will significantly and positively influence MQTP.
2.5. Novice and Experienced Teachers
On the basis of  contention that efﬁcacy beliefs are most evident at the early stage of the career, we group the respondents into two subsamples: (1) teaching experience less than five years as “novice teachers” [37, 38] and (2) teachers with five or more as “experienced teachers”. Chang’s  findings revealed that teachers leave the teaching profession within the ﬁrst ﬁve years, arguing that the habitual patterns in teachers’ judgments are due to teachers’ repeated experiences of uncomfortable emotions, which may eventually lead to burnout. Early career teachers’ feelings of insufficient ability to deal with students’ problematic encounters negatively correlate with self-efficacy beliefs and a sense of professional agency in the classroom .
MQTP refers to reasons for teacher attrition. Several studies in different cultures have revealed that teachers’ departure from their teaching jobs has become a global problem [28, 32–34, 37]. Literature reported a high rate of teacher attrition in several countries, including Australia, China, England, and Norway . According to Chang , the high level of teacher attrition traced back to stress and burnout. Additionally, Skaalvik and Skaalvik  found that emotional exhaustion positively predicts MQTP while is negatively affected by job satisfaction.
A total of 457 teachers from the Central Visayas Region in the Philippines participated in the study. We gather data using online survey forms. In the data quality audit, we excluded 11 responses due to duplication, missing data, and failure to hold the sincerity test. The total number of respondents included in the analysis was 446. Table 1 reveals the demography of the final participants.
|Category||Total, N = 446||Novice ( yrs), N = 240||Experienced (5 yrs), N = 206|
|Type of school|
|Highest educational attainment|
|Master’s degree holder||104||23.3||30||12.5||74||35.9|
|Baccalaureate degree holder||322||72.2||210||87.5||112||54.4|
The questionnaire has two parts. The first part was designed to gather the teacher-participants’ demographic information, and the second part was the indicators of the constructs included in the study. The psychometric qualities of the scale were confirmed in previous studies [14, 16, 19–21, 23, 28, 30, 39–42].
Derived from Bacharach et al. , the perceived PSC of the teachers is measured by the following items: “how much do co-workers go out of their way to help you?” “How much could you rely on your co-workers to provide money or other things if you were in need?” “When things get tough at work, how much can you count on your co-workers to listen, show understanding or show that they care?” “When things get tough at work, how much can you rely on your co-workers for advice or information?” Responses were given on a five-point Likert scale ranging from “not at all” (1) to “very much” (5). Cronbach’s alpha for the scale was 0.787.
Derived from Bacharach and Bamberger , we measure the teachers’ perceived SSC using the following items: “how often can your unit heads be counted on to listen, show understanding or show they care when things get tough at work?” “How often can you rely on your unit heads for advice or information when things get tough at work?” “How often do your unit heads go out of their way to do things to make your work-life easier?” “How often could you rely on your unit heads to assist you with practical matters/minor emergencies off-duty?” Responses were given on a five-point Likert scale ranging from “never” (1) to “always” (5). Cronbach’s alpha for the scale was 0.926.
Derived from Tschannen-Moran and Hoy , we measure TSE using the teachers’ sense of efficacy scale (TSES): “how much can you provide an alternative explanation for examples when students are confused?” “How much can you implement alternative teaching strategies in your classroom?” “How much can you do to control disruptive behavior in the classroom?” “How much can you do to calm a student who is disruptive or noisy?” “How much can you do to motivate students who show low interest in schoolwork?” “How much can you do to get students to believe they can do well in schoolwork?” Responses were given on a five-point Likert scale ranging from “not at all” (1) to “very much” (5). Cronbach’s alpha for the scale was 0.750.
The following items measure the TEE [41, 42]: “I feel burned out from my work,” “I feel fatigued when I get up in the morning and have to face another day on the job,” “I feel frustrated by my job,” “I feel emotionally drained from my work,” and “I feel I am working too hard on my job.” Responses were given on a five-point Likert scale ranging from “never” (1) to “always” (5). Cronbach’s alpha for the scale was 0.826.
The following items measure the TJS : “teaching gives me a great deal of satisfaction,” “teaching enables me to make my greatest contribution to society,” “if I could plan my career again, I would choose to teach,” “I find my contact with students, for the most part, highly satisfying and rewarding,” “I feel that I am an important part of this school system,” “I enjoy working with my students,” and “I am well satisfied with my present teaching position.” Responses were given on a five-point Likert scale ranging from “strongly disagree” (1) to “strongly agree” (5). Cronbach’s alpha for the scale was 0.854.
The following items measure the MQTP [19–21, 28]: “I wish I had a different job to being a teacher,” “If I could choose over again, I would not be a teacher,” “I often think of leaving the teaching profession,” “I intend to quit the teaching profession,” “I intend to search for another job,” and “I wish I could take another career in the future.” Responses were given on a five-point Likert scale ranging from “strongly disagree” (1) to “strongly agree” (5). Cronbach’s alpha for the scale was 0.944.
3.3. Data Analysis
Before the main procedures of data analysis, we computed internal consistency in each group of indicators of the constructs. The results revealed Cronbach’s alpha values ranging from 0.750 to 0.944, as reflected in Instruments. It is followed by model specification among the constructs and indicators by exploratory factor analysis (EFA). We conducted the EFA with a separate data set. The number of respondents in EFA was determined based on a minimum of 20 cases per variable . Since there are six latent variables in the proposed model, the sample size carried out for EFA was 120. We conducted a confirmatory factor analysis (CFA) to ensure if all model fit measures are acceptable. Lastly, we tested the proposed model utilizing structural equation modeling (SEM) analysis. The determination of internal consistency by Cronbach’s alpha and the EFA was done in IBM SPSS statistics 24, while CFA and SEM analyses were done using the AMOS 26.
4.1. Preliminary Analysis
The first preliminary analysis was finding the internal reliability indices of each construct using Cronbach’s alpha of the original survey items. These indices ranging from 0.745 to 0.942 were reflected in Instruments. All indices showed good to very good evaluation . The next part was the visual inspection of multicollinearity and discriminant validity using the correlation matrix in Table 2.
|Standard deviation ()||0.94||0.69||0.39||0.51||0.72||0.93||1.39||7.34|
Intercorrelations among the constructs ranged from −0.66 to 0.52. The results show good discriminant validity since the study variables’ correlation indices are all less than 0.90 [43, 45]. All correlation coefficients are significant at either 0.05 () or 0.01 () alpha levels except between TEE and TSE. The strongest positive correlation was found between TEE and MQTP (0.52), while the strongest negative correlation was found between TJS and MQTP (−0.66). We also found moderate-positive correlations ranging from 0.32 to 0.46, with one moderate-negative correlation at −0.31. All other coefficients were having low correlations ranging from −0.29 to 0.29.
4.2. EFA Results
Due to some interrelated data to be extracted in the proposed model and the absence of a priori model, we define the structure behind the relationships of constructs included in the study using EFA . The analyses were done using dimensions reduction factor analysis in IBM SPSS 24. The EFA results, together with reliability indices by Cronbach’s alpha, is presented in Table 3. Fit measures and factor loadings were evaluated based on the following: the Kaiser–Meyer–Olkin (KMO) value must be >0.80 (highly satisfactory), Bartlett’s test of sphericity must be significant at the 0.05 alpha level, factors with eigenvalue <1.0 should be deleted, communality value should not drop for less than 0.30, and the factor loading for each item must be >0.40 .
The EFA procedures removed the items TJS3, TSE1, TSE5, PSC2, and TEE5 due to cross-loading, low factor loading, and low communality indices. After removing items, the results showed that the values of the factor loading range from 0.472 to 0.956, which indicates that the factors included in the study are considerably important. The KMO value for the model was highly satisfactory at 0.867, with Bartlett’s test of sphericity of 2178.351 (significant at ). Using varimax rotation, we found six factors with eigenvalues from 1.01 to 9.60. Cronbach’s alpha showed high measures ranging from 0.67 to 0.94.
4.3. Testing the Model by CFA
In order to validate the structures found in EFA, a total of 446 responses were loaded to CFA. The fit indices used to determine the model strength were the chi-square test , the root mean square error approximation (RMSEA), and the standardized root mean square residual (SRMR). The relative fit measures were the comparative fit index (CFI) and the Tucker–Lewis index (TLI). We implement the following cut-off scores to achieve a good model; RMSEA must be 0.060, SRMR must be 0.080, TLI must be 0.900, and CFI must be 0.900 . Table 4 reflects the standardized loadings, composite reliability (CR), average variance extracted (AVE), and Cronbach’s alpha of the final model.
|Motivation to quit the teaching profession||MQTP6||0.853||0.943||0.734||0.941|
|Teacher’s job satisfaction||TJS6||0.700||0.818||0.530||0.854|
|Supervisory support climate||SSC4||0.792||0.920||0.742||0.912|
|Teacher’s emotional exhaustion||TEE4||0.809||0.858||0.603||0.854|
|Peer support climate||PSC4||0.837||0.840||0.638||0.755|
Visual inspection of the initial CFA model results revealed that all fit measures met all the required threshold values. We found some issues on the modification indices, specifically correlated errors on the same factors. For a good fitting model, there are two ways to remove correlated errors during CFA. The first way is to remove the lower factor loading item between the two or to constrain the effects of correlated errors through covariation between them . Since all pairs of error terms belong to the same factors, they were constrained to covariate each other. All of the factor loadings in Table 4 were of acceptable range from 0.696 to 0.927. These numbers confirm the correct identification of factors and established multicollinearity in variables used in path analysis . There are no issues with the composite reliability (CR) as all indices are greater than 0.7 . The overall measurement model revealed very satisfactory fit measures of the RMSEA (0.037), SRMR (0.0315), TLI (0.978), and CFI (0.981).
4.4. Exploring the Relationship between the Latent Variables for SEM
We conducted the correlational analysis through the Pearson correlation coefficient to support the path analysis of the SEM. The study followed the value guidelines : 0.00–0.09, “negligible correlation;” 0.10–0.39, “weak correlation;” 0.40–0.69, “moderate correlation;” 0.70–0.89, “strong correlation;” and 0.90–1.00, “very strong correlation.”
Table 5 reflects the correlation matrix among the constructs included in the CFA. The correlation between MQTP and TJS is significant and moderate (), MQTP and SSC is significant and weak (, p < 0.01), TJS and SSC is significant and weak (), MQTP and TEE is significant and moderate (), TJS and TEE is significant and moderate (), SSC and TEE is signiﬁcant and weak (), MQTP and PSC is significant and weak (), TJS and PSC is significant and moderate (), SSC and PSC is signiﬁcant and moderate (), TEE and PSC is signiﬁcant and weak (), MQTP and TSE is not significant and negligible (), TJS and TSE is significant and weak (), SSC and TSE is not signiﬁcant and negligible (), TEE and TSE is not significant and negligible (), and PSC and TSE is signiﬁcant and weak (). It is noteworthy that all negligible correlations are not significant. There are three weak-positive, five weak-negative, two moderate-positive, and two moderate-negative correlations. As expected, the correlation between constructs was all higher than the zero-order correlation in the preliminary analysis.
We tested the relations among the variables using SEM and reported the standardized regression weights in Table 6. The reporting excludes the paths that were not significant. All of the fit measures of the ﬁnal model are acceptable ( TLI = 0.976, and CFI = 0.976). The RMSEA = 0.038 indicates an excellent fit between the hypothesized model and the observed data .
Table 6 revealed that all hypothesized relationships except H5 in the proposed model denote significant predictors in the specified path model. It is noteworthy that H5 stated as “TSE will significantly and negatively influence MQTP” failed to hold significant results and is removed from the path analysis, as reflected in Figure 2. In the final model, PSC is an endogenous variable that directly explains TSE. SSC directly explains TJS and TEE, which also have mediated indirect effects on MQTP. TSE has a direct effect on TJS and reflects a mediated indirect effect on MQTP. TJS and TEE have direct effects on MQTP. It should be noted that no further covariates were created as the overall measurement model was all found satisfactory during the CFA.
4.6. Significant Differences
The study also investigated whether experience type (novice and experienced teachers) differs regarding all constructs (TSE, TJS, TEE, and MQTP). The t-test results reported no statistical difference between novice and experienced teachers concerning all constructs except TJS. Complete information and comparison of the values across the demographic can be seen in Table 7.
Using the SEM, the proposed model that involves PSC, SSC, TSE, TJS, TEE, and MQTP is informed to be valid and acceptable . Through modiﬁcations of model specifications and removing items that are not working well, the final model achieved an excellent fitting relationship between predictors and motivation to quit the teaching profession. The model will generate discussion among developing countries like the Philippines. All fit measures satisfy the common threshold values used by SEM researchers [43, 44, 47].
The study includes a desirable sample size to test the hypotheses that have been assumed in Asian culture, as reported by Prasojo et al. . SEM results described that peer social support directly explains part and overall variation in the teachers’ self-efficacy. This finding affirmed our hypotheses, as reported by Skaalvik and Skaalvik . Also, the social support of supervisors and principals has been described to have a negative effect on teachers’ emotional exhaustion or burnout. The works of Chang , Gavish and Friedman , and Prasojo et al.  support these findings. Therefore, a social comparison should always be conducted in research about teachers’ burnout in the future with different samples of cultural diversity.
The TJS and TEE informed to have effects on the desire to leave the teaching profession. The more satisfied the teachers are, the more likely they stay in the profession, while emotionally exhausted teachers tend to quit from their jobs. These findings affirm those reported in the works of Hackett et al. , Skaalvik and Skaalvik , and Sun and Xia . As important members of the profession, teacher teaching experience, which was reported to be significantly different between novice and experienced, may aﬀect job satisfaction and thereby the tendency to move between schools or quit the profession altogether . These findings need further study to figure out what constitutes the variables of novice teachers’ job satisfaction, especially in developing countries. Job satisfaction varies inversely as emotional exhaustion or burnout at work. The more exhausted the teachers are, the less they are satisfied at work. The ﬁndings of the research facilitate our hypotheses that have been assumed based on the study reported by Klassen and Chiu  and Skaalvik and Skaalvik . Generally, TEE provided a basic offer for teachers to perform judgment on burnout. TSE, TJS, and TE could work together as factors affecting MQTP.
The findings also informed that, among all constructs included in the proposed model, only TJS significantly differs in the mean comparisons. The study suggests that teachers’ perceptions of PSC, SSC, TSE, TEE, and MQTP are just the same among novice and experienced teachers. Thus, the model’s reported direct and indirect effects possessed the same variation among novice and experienced teachers. The current study results argue what Gavish and Friedman  stated: novice teachers have higher efficacy than experienced teachers.
Novice teachers have been reported not to vary with experienced teachers in all constructs except TJS. PSC and SSC have been reported to explain TSE, TJS, and TEE. MQTP has been reported to change in accordance with self-efficacy, emotional exhaustion, and job satisfaction. The findings of the research advocate the proposed model. The model can guide future researchers in developing countries like the Philippines in explaining teacher attrition caused by social support, efficacy factors, burnout, and job satisfaction. Educational stakeholders should ensure suitable social support climates to be successful institutions in promoting learning activities. Working with peers is useful while developing lesson activities; enhanced teaching skills of novice teachers by direct observations with experts in the field and receiving verbal appreciations are factors to improve self-efficacy, job satisfaction, and eliminate burnout .
Data are available on request through the e-mail address of Cebu Technological University Educational Research and Resource Center (CTU-ERRC) ([email protected]).
Conflicts of Interest
The authors declare that they have no conflicts of interest.
This work was funded under the General Appropriations Act (GAA) 2019 of the Cebu Technological University, Cebu City, Philippines.