Hypothesis Testing

Hierarchical regression analysis was adopted to test all hypotheses. Control variable (i.e. gender, tenure, education, and department) were entered in Step 1 of all regressions.

Then, independent variables were entered to test Hypotheses 1–3 in step 2, and in step 3, trust was inputted to test hypotheses 4. The detailed methods and results are presented below.

4.2.1 Relationship among social network properties and affective organizational commitment

For testing the hypothesis that the network properties of teachers have a significant effect on their affective organizational commitment, a two-stage hierarchical multiple regression analysis was performed. This research looks into two properties of network, which are outdegree centrality and strength of tie, two separated hypotheses were formed.

The first one is that the network outdegree centrality of teachers have a significant effect on their affective organizational commitment (Hypothesis 1a). The other is that the tie strength of teachers has a significant effect on their affective organizational commitment (Hypothesis 1b). The outputs of these analyses were combined into one table, which model 1 is for the first network property and model 2 illustrated the second one. In the first step, control variables (i.e. Gender, Tenure, Education, Department) were entered for both models. In the second step, outdegree centrality was inputted in model1, and strength of tie was entered in model 2. The specific results of these two models are presented in table 4.2 below.

Table 4.2 Results of regression analysis for social network properties and affective organizational commitment

Independent variables Dependent variables (Affective Organizational Commitment)

Model 0 Model 1 Model 2

In model 1, it is apparent from the result table that outdegree centrality is not significantly related to affective organizational commitment (B = 1.558, p > 0.05). Thus, the hypothesis 1a was not supported. However, in model 2, there was a positive and significant relationship between strength of tie and affective organizational commitment (B = 0.011, p < .05); therefore, the hypothesis 1b was supported. Overall, only strength of tie was positively and significantly related to affective organizational commitment.

Accordingly, hypothesis 1 was partially supported.

4.2.2 Relationship among social network properties and trust

To test the hypothesis that the network properties of teachers positively influence the level of their relational trust in colleagues (Hypothesis 2), a two-stage hierarchical

multiple regression analysis was also performed. As mentioned above, this research looks into two properties of network, which are outdegree centrality and strength of tie, two separated hypotheses were formed. The first one is that the network outdegree centrality of teachers positively influence the level of their relational trust in colleagues (Hypothesis 2a). The other is that the tie strength of teachers positively influence the level of their

relational trust in colleagues (Hypothesis 2b). The outputs of these analyses were combined into one table, which model 1 is for the first network property and model 2 illustrated the second one. In the first step, control variables (i.e. Gender, Tenure, Education, Department) were entered for both models. In the second step, outdegree centrality was inputted in model1, and strength of tie was entered in model 2. The specific results of these two models are presented in table 4.3 below.

Table 4.3 Results of regression analysis for social network properties and trust Independent variables Dependent variables (Trust)

Model 0 Model 1 Model 2

In both models, education had a positive and significant relationship with trust (B

= .17, p < .05). In model 1, it is apparent that outdegree centrality is not significantly related to trust (B = 1.527, p > 0.05). Thus, the hypothesis 2a was not supported.

However, in model 2, there was a positive and significant relationship between strength of tie and trust (B = 0.012, p < .05); therefore, the hypothesis 2b was supported. Overall, only strength of tie was positively and significantly related to trust. Accordingly, hypothesis 2 was partially supported.

4.2.3 Relationship among trust and affective organizational commitment

A two-stage hierarchical multiple regression analysis was also used to test hypothesis 3, which is teachers’ relational trust positively influences the level of their affective organizational commitment. In the first step, affective organizational

commitment was entered as dependent variable and Gender, Tenure, Education, Department were inputted into independent variables box as control variables. In the second step, trust was entered. The specific results of these two models are presented in table 4.4 below.

Table 4.4 Results of regression analysis for trust and affective organizational commitment Independent

variables

Dependent variables (Affective organizational commitment)

Model 0 Model 1

It is clear that there was a positive and significant relationship between trust and affective organizational commitment (B = .654, p < .001) as reported in previous studies.

The regression coefficient is .654, which means that with every change in trust for one unit, the change in affective organizational commitment increases .654 unit while holding other predictors in the model constant. The coefficient of determination, R², is .267, which means trust explain 26.7% of affective organizational commitment. To conclude,

hypothesis 3 was supported.

4.2.4 Mediating role of trust for relationship between social network properties and affective organizational commitment

The results of mediated regression analysis used to test Hypothesis 4 is shown in Table 4.5. Hypotheses 4 predicted that relational trust positively mediates the relationship between teachers’ network properties, which are outdegree centrality and strength of tie, and their affective organizational commitment. Because the first requirement for

mediation was not met in outdegree case (hypothesis 4a), the mediation analysis was only used to test hypothesis 4b. It is relational trust positively mediates the relationship

between teachers’ tie strength and their affective organizational commitment.

Table 4.5 Results of regression analysis for trust and affective organizational commitment Independent variables Dependent variables (Affective Organizational Commitment)

Model 1 (ID  DV)

The first regression equation (Model 1) revealed that strength of tie was positively and significantly associated with affective organizational commitment (B = .011, p < .05), meeting the first requirement for mediation. In Model 2, trust was regressed on affective organizational commitment and a positive and significant relationship between them was

found (B = .654, p < .001). Thus, the second requirement for mediation was met. In Model 3, strength of tie and trust was inputted in step 2 and step 3, respectively. It is found that strength of tie turned from significant (B = .011, p < .05) in model 1 into insignificant (B

= .003, p > .05) in model 3 when trust was entered as mediator. Meanwhile, the significant relationship between the trust and affective organizational commitment was still

significant. Therefore, the third requirement was met. It is concluded that trust mediated the relationship between strength of tie and affective organizational commitment,

therefore, the hypothesis 4b was supported.

4.2.5 Summary of hypothesis testing results

Based on the results of hypothesis testing above, the table 4.6 provides a summary of the findings for each of the four hypotheses.

“Table 4.6 Summary of hypothesis testing”

Hypothesis Statement Result

H1

“The network properties of teachers are positively related to their affective organizational commitment.”

“The network outdegree centrality of teachers is positively related to their affective organizational commitment.”

“The tie strength of teachers is positively related to their affective organizational commitment.”

“The network properties of teachers are positively related to the level of their trust in colleagues.”

“The network outdegree centrality of teachers is positively related to the level of their trust in colleagues.”

“The tie strength of teachers is positively related to the level of their trust in colleagues.”

“Teachers’ trust is positively related to the level of their affective organizational commitment.”

“Trust positively mediates the relationship between teachers’ network properties and their affective organizational commitment.”

“Trust positively mediates the relationship between

Partially

H4b

teachers’ network outdegree centrality and their affective organizational commitment.”

“Trust positively mediates the relationship between teachers’ tie strength and their affective organizational commitment.”

Not supported

Supported