Unpacking the role of school climate factors in fostering teacher enjoyment and self-efficacy in higher education

PLOS One | https://doi.org/10.1371/journal.pone.0329001 August 13, 2025 6 / 17 Instruments. To ensure the validity of this study, the first author, holding a master’s degree in Translation and Interpreting, translated and then back-translated all measures employed in this study. Perceived school climate scale (R-SLEQ): We used the Revised School-Level Environment Questionnaire (R-SLEQ), developed by Johnson et al., to measure teachers’ views on their school climate [75]. This instrument consists of 21 items across five dimensions: (a) colleague collaboration, (b) teacher-student relations, (c) school resources, (d) decision- making, and (e) instructional innovation. Participants rated items using a 5-point scale, with 1 denoting “strongly disagree” and 5 denoting “strongly agree”. Among the 21 questions, 8 were negatively phrased and reverse scored, such as “Good teamwork is not emphasized enough at my school.” (1 = strongly agree, 5 = strongly disagree). The R-SLEQ exhibited great reliability, with a reported coefficient of 0.90 that ranged between 0.77 and 0.86 for the five sub-scales [75]. The scale’s robust construct validity was further validated by Hosford and O’Sullivan [91]. In this study, the reliability index of the R-SLEQ, evaluated through Cronbach’s alpha, varied between 0.74 and 0.86. Teacher self-efficacy scale (TSES): To measure teacher self-efficacy, we employed the short form of Teachers’ Sense of Efficacy Scale (TSES) developed by Tschannen-Moran and Hoy [39]. This scale evaluates teachers’ beliefs in their

ability to (a) engage students, (b) apply instructional strategies, and (c) manage classroom. The instrument consists of 12 items, rated on a 9-point scale from 1 (nothing) to 9 (a great deal). The TSES demonstrated strong reliability, with reported coefficients for instructional strategies, classroom management, and student engagement of 0.86, 0.86, and 0.81, respec- tively, accounting for 65% of the variance in teachers’ responses [39]. Klassen et al. further validated the validity of the TSES measure across five different nations [62]. In this study, the internal consistency of the TSES was confirmed with a Cronbach’s alpha of 0.87. Teaching enjoyment scale (TES): To measure teacher enjoyment, the Teacher Emotions Scale (TES) developed by Frenzel and his fellow researchers was used [12]. This scale includes three main emotions: happiness, anxiety, and anger, with a total of 12 items. Our research focused on the sub-scale measuring teachers’ positive emotion—enjoyment in the classroom. The subscale contained four items: “I generally enjoy teaching.”; “I generally have so much fun teaching that I gladly prepare and teach my lessons.”; “I often have reasons to be happy while I teach.”; and “I generally teach with enthusiasm.” Participants rated these items on a 4-point scale from “strongly disagree” to “strongly agree.” In Frenzel et al.’s study, Cronbach’s alphas for the entire TES and for the enjoyment subscale were 0.90 and 0.77, respectively [12]. In our study, the reliability of the TES was confirmed with a Cronbach’s alpha of 0.92.

Measurement validation. To ensure the construct validity, we conducted confirmatory factor analyses (CFA) on each scale individually to test the proposed factor structure of the measurement tools. First, the structure of the R-SLEQ with five constructs (collaboration, student relations, school resources, instructional innovation, and decision- making) was probed for low estimates in standardized estimation. The standardized estimates for individual items ranged from 0.569 to 0.868. Second, the TSES was evaluated, with all twelve items set to load on one factor. The results of the factor loading varied from 0.316 to 0.896. Therefore, six items (CM1, SE1, SE2, SE4, IS2, IS4) were removed due to lower factor loadings (lower than 0.4). Following adjustments, the final construct demonstrated excellent fit indices for the data. Third, for the TES, all four items were loaded onto one factor, with factor loadings ranging from 0.819 to 0.889. Following adjustments to the individual models, we tested the final CFA model incorporating all measurement scales. The results indicated an excellent fit to the data: CMIN/df = 1.755, RMSEA = 0.035, CFI = 0.968, TLI = 0.964. We then assessed the model’s convergent and discriminant validity for each construct. Analysis in Table 1 revealed that all variables exhibited CR values above 0.7, demonstrating an acceptable reliability. Additionally, the mean variance for each factor exceeded 0.5, supporting the model’s convergent validity. The correlations between all pairs

of factors were significant at p < .01, signifying large effect sizes. Moreover, the square root of AVE values (bolded in the table) was above the inter-correlations of the factors, confirming discriminant validity, as defined by Fornell and Larcker [92].