3.1 Measurement model assessment
To ensure the measurement model's reliability and validity, a meticulous evaluation was conducted in accordance with established research methodologies[37]. Initially, the researchers implemented an ordinary PLS-SEM algorithm to identify the factor loadings. The advisable value for these loadings is 0.70 or above, suggesting that they encapsulate more than 50% of the variance of the indicator[37]. Subsequently, the study assessed the internal consistency reliability by applying Cronbach's alpha and composite reliability (CR). As demonstrated in Table 1, both measures surpassed the suggested threshold of 0.70, thereby underscoring item reliability[38]. The final stage of the analysis was the evaluation of the measurement model's convergent and discriminant validities, achieved by examining the average variance extracted (AVE)[37]. As shown in Table 1, the AVE values for all constructs exceeded the minimum acceptable value of 0.50, implying a commendable level of convergent validity[39]. Therefore, it can be concluded that the measurement model exhibits satisfactory reliability and validity characteristics.
Table 1 Reliability and convergent validity
|
Construct
|
Item
|
Factor loading
|
Cronbach's alpha
|
Average variance extracted (AVE)
|
Composite reliability (CR)
|
System-related Teacher Anger
(AN)
|
AN1
|
0.591***
|
0.848
|
0.592
|
0.851
|
AN2
|
0.763***
|
AN3
|
0.850***
|
AN4
|
0.846***
|
L2 Teacher Work Engagement
(WE)
|
WE1
|
0.721***
|
0.880
|
0.652
|
0.882
|
WE2
|
0.849***
|
WE3
|
0.851***
|
WE4
|
0.801***
|
L2 Teacher Teaching Enjoyment
(TE)
|
TE1
|
0.783***
|
0.858
|
0.638
|
0.874
|
TE2
|
0.923***
|
TE3
|
0.817***
|
TE4
|
0.648***
|
L2 Teacher Job Satisfaction
(JS)
|
JS1
|
0.780***
|
0.724
|
0.646
|
0.845
|
JS2
|
0.792***
|
JS3
|
0.838***
|
L2 Teacher Emotion
Regulation
Behaviour
(ERB)
|
ERB1
|
0.708***
|
0.816
|
0.661
|
0.903
|
ERB2
|
0.877***
|
ERB3
|
0.968***
|
ERB4
|
0.925***
|
ERB5
|
0.493***
|
L2 Teacher
Well-being
(TWB)
|
TWB1
|
0.735***
|
0.914
|
0.519
|
0.915
|
TWB2
|
0.788***
|
TWB3
|
0.757***
|
TWB4
|
0.711***
|
TWB5
|
0.708***
|
TWB6
|
0.749***
|
TWB7
|
0.715***
|
TWB8
|
0.704***
|
TWB9
|
0.683***
|
TWB10
|
0.644***
|
*** p<0.001
|
3.2 Correlation of the constructs
A noteworthy finding is that system-related teacher anger (AN) shows a significant but modest correlation only with L2 teacher work engagement (WE), suggesting that teachers' perceptions of systemic injustices or mistreatments within the educational system may influence their level of engagement at work. This correlation, however, is not strong, as indicated by the correlation coefficient of 0.169*, implying that other factors might also play a role in influencing work engagement. Contrastingly, L2 teacher teaching enjoyment (TE) does not exhibit a significant correlation with any of the other variables studied. This finding indicates that the personal joy that L2 teachers derive from their teaching might operate independently of their work engagement, job satisfaction, emotion regulation, and overall well-being. This lack of significant correlations could point to the unique nature of teaching enjoyment as a factor that is potentially influenced by different variables not captured in this study.
The correlations between work engagement (WE) and job satisfaction (JS) with well-being (TWB) are relatively stronger, as evidenced by the coefficients of 0.443** and 0.575** respectively. These correlations suggest that both engagement in work and satisfaction with the job are important contributors to the overall well-being of L2 teachers. Particularly, the stronger correlation with job satisfaction might imply that teachers' perceptions of their job roles and the fulfillment they derive from them are critical determinants of their psychological health. Additionally, emotion regulation behaviour (ERB) shows significant correlations with both work engagement (0.398**) and job satisfaction (0.367**), highlighting the importance of teachers' ability to manage and navigate their emotions in relation to their engagement with and satisfaction from their work. This interplay between emotional regulation and job-related factors underscores the complex emotional dynamics inherent in the teaching profession.
3.3 Multiple regression analysis
SPSS was utilised by the research team to perform multiple regression analysis, a method geared towards modelling and analysing several variables concurrently. Within this analysis, a regression equation was established. To gauge the variance in the dependent variable attributable to the independent ones, the coefficient of determination, R2, was computed. The relationships between system-related teacher anger (AN), L2 teacher work engagement (WE), L2 teacher job satisfaction (JS), L2 teacher teaching enjoyment (TE), and L2 teacher emotion regulation behaviour (ERB) are detailed in Table 2.
Table 2 section 1: Variation analysis – ANOVA a
|
Model
|
Sum of Squares
|
df
|
Mean Square
|
F
|
Sig.
|
Regression
|
15.012
|
4.000
|
3.753
|
11.308
|
.000b
|
Residual
|
55.095
|
166.000
|
0.332
|
|
|
Total
|
70.107
|
170.000
|
|
|
|
a. Dependent Variable: ERB
|
b. Predictors: (Constant), TE, WE, AN, JS
|
Table 2 section 2: Regression coefficients
|
|
Unstandardised
Coeff.
|
Standardised Coeff.
|
t
|
Sig.
|
95% Confidence Interval for B
|
Collinearity
Statistics
|
The Results of Hypothesis Testing
|
Model
|
B
|
Std. Error
|
B
|
|
|
Boot
LLCI
|
Boot
ULCI
|
Tolerance
|
VIF
|
|
Constant
|
1.146
|
0.456
|
-
|
2.511
|
0.013
|
0.245
|
2.047
|
-
|
-
|
-
|
WE
|
0.330
|
0.096
|
0.270
|
3.450
|
0.001
|
0.141
|
0.519
|
0.772
|
1.295
|
Accepted
|
JS
|
0.264
|
0.081
|
0.254
|
3.269
|
0.001
|
0.105
|
0.423
|
0.782
|
1.278
|
Accepted
|
AN
|
0.078
|
0.058
|
0.097
|
1.352
|
0.178
|
-0.036
|
0.193
|
0.927
|
1.079
|
Rejected
|
TE
|
0.009
|
0.049
|
0.013
|
0.190
|
0.849
|
-0.088
|
0.106
|
0.962
|
1.039
|
Rejected
|
a. Dependent Variable: ERB
|
Table 2 section 3: Estimation of standard deviation
|
Model
|
R
|
R Square
|
Adjusted R Square
|
Std. Error of the Estimate
|
1
|
0.463a
|
0.214
|
0.195
|
0.576
|
a. Predictors: (Constant), TE, WE, AN, JS
|
b. Dependent Variable: ERB
|
Table 2, section 1 reveals a significance value of 0.000, which is below 0.05. This meets the significance criteria, indicating that, based on the F-test, the regression model is statistically valid.
According to Table 2, section 2, the collinearity statistics reveal no evidence of collinearity between the independent variables. A meticulous test is paramount for a comprehensive validation of the multiple regression model.
Table 2, section 2, indicates that collinearity statistics reveal no evidence of collinearity between the independent variables. This evaluation seeks to ascertain that the regression coefficients of all independent variables approximate zero, thereby confirming that the variance explained is not simply coincidental. The regression coefficients in the sample are identified as B1, B2, B3, and B4. The corresponding null and alternative hypotheses, which will provide deeper insight into the study, are set to be detailed in the following section:
H0=B1=B2=B3=B4=0
H1= not all B coefficients are equal to 0
In this research, the Student's t-test is utilized, accounting for n−(k+1) degrees of freedom. As indicated in Table 2, section 2, the t-values obtained for the four variables are as follows: 3.450 for WE, 3.269 for JS, 1.352 for AN, and 0.190 for TE. For the purposes of hypothesis testing, these t-values are compared against the critical t-value, which is set at a 0.05 significance level for a two-tailed test. With the four independent variables considered, the study involves 162 degrees of freedom, resulting in a critical t-value of ±1.976.
In the case of L2 teacher work engagement (WE), the observed t-value (3.450) is higher than the critical t-value (1.976). The test reveals a significance level of 0.001, lower than the predetermined significance threshold of 0.05. The null hypothesis for WE is rejected. This confirms that B1 is statistically distinct from zero. For L2 teacher job satisfaction (JS), the observed t-value (3.269) is higher than the critical t-value (1.976). Similarly, with JS's observed t-value (3.269) exceeding the critical value (1.976) and the significance level (0.001) falling below 0.05, the null hypothesis for JS is also rejected. This indicates that B2 is significantly different from zero.
However, given the observed t-value (1.352) for AN being less than the critical value (1.976) and the significance level (0.178) exceeding 0.05, the null hypothesis for AN is accepted. This suggests that B3 is not statistically different from zero. Similarly, the null hypothesis for TE is accepted since its observed t-value (0.190) is lower than the critical value (1.976), and the significance level (0.849) is above 0.05. This implies that B4 is statistically zero. In summary, neither system-related teacher anger nor L2 teacher teaching enjoyment significantly influences L2 teacher emotion regulation behaviour.
Accordingly, the regression equation is derived as ERB = 1.146 + 0.330 x1 + 0.264 x2, where x1 = L2 teacher work engagement (WE), x2 = L2 teacher job satisfaction (JS).
The coefficient of determination, R2, representing the percentage of total variance explained by the independent variable, is 21.4%, as shown in Table 2 section 3.
Table 3 elucidates the relationships among system-related teacher anger (AN), L2 teacher work engagement (WE), L2 teacher job satisfaction (JS), L2 teacher teaching enjoyment (TE), and L2 teacher well-being (TWB).
Table 3 section 1: Variation analysis – ANOVA a
|
Model
|
Sum of Squares
|
df
|
Mean Square
|
F
|
Sig.
|
Regression
|
22.627
|
4.000
|
5.657
|
25.404
|
.000b
|
Residual
|
36.963
|
166.000
|
0.223
|
|
|
Total
|
59.590
|
170.000
|
|
|
|
a. Dependent Variable: TWB
|
b. Predictors: (Constant), TE, WE, AN, JS
|
Table 3 section 2: Regression coefficients
|
|
Unstandardised
Coeff.
|
Standardised Coeff.
|
t
|
Sig.
|
95% Confidence Interval for B
|
Collinearity
Statistics
|
The Results of Hypothesis Testing
|
Model
|
B
|
Std. Error
|
B
|
|
|
Boot
LLCI
|
Boot
ULCI
|
Tolerance
|
VIF
|
|
Constant
|
1.141
|
0.374
|
|
3.052
|
0.003
|
0.403
|
1.879
|
-
|
-
|
-
|
WE
|
0.283
|
0.078
|
0.252
|
3.616
|
0.000
|
0.129
|
0.438
|
0.772
|
1.295
|
Accepted
|
JS
|
0.443
|
0.066
|
0.463
|
6.702
|
0.000
|
0.313
|
0.574
|
0.782
|
1.278
|
Accepted
|
AN
|
-0.038
|
0.047
|
-0.051
|
-0.798
|
0.426
|
-0.132
|
0.056
|
0.927
|
1.079
|
Rejected
|
TE
|
-0.003
|
0.040
|
-0.004
|
-0.070
|
0.944
|
-0.082
|
0.077
|
0.962
|
1.039
|
Rejected
|
a. Dependent Variable: TWB
|
Table 3 section 3: Estimation of standard deviation
|
Model
|
R
|
R Square
|
Adjusted R Square
|
Std. Error of the Estimate
|
1
|
0.616a
|
0.380
|
0.365
|
0.472
|
a. Predictors: (Constant), TE, WE, AN, JS
|
b. Dependent Variable: TWB
|
As depicted in Table 3 section 1, the p-value is 0.000, below 0.05, denoting the statistical significance of the regression model.
According to Table 3 section 2, the analysis of collinearity confirms that the independent variables are free from collinearity concerns. The t-value (3.616) of L2 teacher work engagement (WE) is observed to be higher than the critical t-value (1.976). The test reveals a significance level of 0.000, lower than the predetermined significance level of 0.05. Given these results, the null hypothesis for WE is rejected, suggesting that B1 is statistically distinct from zero. For L2 teacher job satisfaction (JS), the observed t-value (6.702) is higher than the critical t-value (1.976). Following the same criteria, the null hypothesis for JS is also rejected, indicating B2 is significantly different from zero."
However, in the case of system-related teacher anger (AN), the recorded t-value stands at -0.798, which is higher than the critical t-value (-1.976). And the significance level (0.426) exceeds 0.05. Given these values, the null hypothesis for AN is accepted, indicating that B3 is not statistically different from zero. The t-value for L2 teacher teaching enjoyment (TE) is -0.07, which exceeds the critical t-value of -1.976. With a significance level of 0.944 (above the 0.05 threshold), it is not statistically significant. The null hypothesis for TE is accepted, signifying that B4 is statistically zero. Consequently, neither system-related teacher anger nor L2 teacher teaching enjoyment is significant predictors for L2 teacher emotion regulation behaviour.
Based on the unstandardised coefficients, the regression equation can be formulated as TWB = 1.141+ 0.283 x1 + 0.443 x2, where x1 = L2 teacher work engagement (WE), x2 = L2 teacher job satisfaction (JS).
In this case, the coefficient of determination R2 is 38% (Table 3 section 3).
3.4 Mediating analysis
The researchers have utilised the PROCESS macro for SPSS to execute a mediation analysis in the current investigation[40]. Model 4, tailored for single mediator models, was chosen for this research. The L2 Teacher Emotion Regulation Behaviour (ERB) is recognised as the mediator bridging the independent and dependent variables. Two mediation paths demonstrated statistical significance (P < 0.05) based on prior analysis.
In the subsequent discourse, "Effect Ratio" represents the Path(N) quotient to the total effect documented. Additionally, the "P/C" column delineates whether the pathway embodies a complete ('C') or partial ('P') mediator. A complete mediator signifies that the indirect effect encapsulates the relationship between variables. A partial mediator, on the other hand, indicates the coexistence of both direct and indirect effects.
Table 4 section 1: Significant mediating effects of the research model A
|
Total effect of WE on TWB
|
Effect
|
SE
|
t
|
p
|
LLCI
|
ULCI
|
c_cs
|
0.499
|
0.078
|
6.433
|
0.000
|
0.346
|
0.653
|
0.444
|
Direct effect of WE on TWB
|
Effect
|
SE
|
t
|
p
|
LLCI
|
ULCI
|
c'_cs
|
0.342
|
0.079
|
4.318
|
0.000
|
0.186
|
0.498
|
0.304
|
Indirect effect(s) of WE on TWB:
|
|
Effect
|
BootSE
|
BootLLCI
|
BootULCI
|
Effect Ratio
|
P/C
|
WE→ERB→TWB
|
0.157
|
0.045
|
0.077
|
0.253
|
31%
|
P
|
Table 4 section 2: Significant mediating effects of the research model B
|
Total effect of JS on TWB
|
Effect
|
SE
|
t
|
p
|
LLCI
|
ULCI
|
c_cs
|
0.425
|
0.057
|
7.423
|
0.000
|
0.312
|
0.538
|
0.496
|
Direct effect of JS on TWB
|
Effect
|
SE
|
t
|
p
|
LLCI
|
ULCI
|
c'_cs
|
0.329
|
0.056
|
5.889
|
0.000
|
0.219
|
0.440
|
0.385
|
Indirect effect(s) of JS on TWB:
|
JS→ERB→TWB
|
0.095
|
0.030
|
0.040
|
0.157
|
22%
|
P
|
Table 4 section 1 demonstrates a comprehensive overview of the total, direct, and indirect effects of variable L2 Teacher Work Engagement (WE) on L2 Teacher Well-being (TWB) within the proposed model. The total effect of WE on TWB is reported as 0.449 (p < 0.001), indicating a statistically significant relationship. Furthermore, the direct effect of WE on TWB, which isolates the impact of WE on TWB while disregarding the mediating variables, is also found to be statistically significant at 0.342 (p < 0.001). The total indirect effect is estimated as 0.157, which is statistically significant, as evidenced by the bootstrapped confidence interval not encompassing zero. This finding suggests that ERB significantly influence the relationship between WE and TWB. The effect ratios associated with this pathway indicate that it accounts for 31% of the total effect. In the "P/C" column, this pathway is categorised as partial mediation.
According to the information presented in Table 4, section 2, the total effect size of L2 teacher job satisfaction (JS) on L2 teacher well-being (TWB) is reported as 0.425, with a significance level of p < 0.001. Additionally, the direct effect, representing the impact of JS on TWB without mediation from other variables, is also found to be significant at 0.329 (p < 0.001). These findings indicate that L2 teacher job satisfaction (JS) significantly predicts L2 teacher well-being (TWB) even when the potential mediator is not considered. However, the indirect effects provide a more nuanced understanding of the relationship between JS and TWB. Indirect effects refer to the influence of JS on TWB through ERB. In this particular case, the total indirect effect of JS on TWB is estimated as 0.095, and its significance is confirmed by the bootstrapped lower limit confidence interval (BootLLCI) and upper limit confidence interval (BootULCI) not encompassing zero. The effect ratios associated with this pathway indicate that it accounts for 22% of the total effect. In the "P/C" column, this pathway is categorised as partial mediation, signifying that the effect of JS on TWB operates significantly through this particular pathway.