BUS603: Managing People

3.3.1. Confirmatory Factor Analyses

We conducted a set of confirmatory factor analyses (CFA) in AMOS 20.0 to test the construct validity. First, we conduct baseline model that included all key four constructs (see Table 1) to calculate model fit indexes. The chi-square, CFI, RMSEA, and TLI were used to evaluate the model fit. The results show that the baseline model fit the data well: χ²(48) = 143.022, CFI = 0.977, TLI = 0.968, RMSEA = 0.065. In addition, all of the factor loadings were significant, demonstrating convergent validity is acceptable.

Table 1. Comparison of measurement models.

The discriminant validity of the four constructs was tested by comparing the baseline model fit to another four alternative models fit. The results in Table 1 are shown that the distinctiveness of the four constructs in this study is acceptable.

Additionally, following the suggestions of Fornell and Larcker, we further tested the discriminant validity of the four variables by evaluating average variance-extracted estimates (AVE). As shown in Table 2, AVEs of the four construct, namely EOR, organizational trust, innovative climate, and innovative behavior were 0.706, 0.766, 0.740, and 0.802, respectively. These variables were all exceeded the benchmark of 0.5. We found the square roots of AVE for the four constructs were more than the correlation between any two constructs. These statistics, together with the CFAs results, provided evidence that the construct validity of the four variables were acceptable.

Table 2. Descriptive statistics and correlations.

3.3.2. Descriptive Statistics

Table 2 shows the means, standard deviations, and correlations of all key constructs. As shown in Table 2, EOR was positively related to organizational trust (r = 0.553, p < 0.001) and innovative behavior (r = 0.089, p < 0.001). Moreover, organizational trust was positively related to innovative behavior (r = 0.276, p < 0.001). These results provide initial evidence to support our Hypotheses 1–4.

3.3.3. Hypothesis Testing

We conducted hierarchical multiple regression analysis with STATA 12.0 to test Hypotheses 1–5, the results are shown in Table 3. To test Hypothesis 1, the four control variables, and the independent variable (EOR) were entered into regression equation, where innovative behavior was the dependent variable. The result shows that EOR is significant positively related to innovative behaviors (β = 0.081, p < 0.05, Model 4). Hence, Hypothesis 1 is supported.

Table 3. Results of hierarchical regression analysis for the hypothesized relationships.

As shown in Table 3, EOR is significantly positively related to organizational trust (β = 0.554, p < 0.001, Model 2), therefore, Hypothesis 2 is supported. In addition, organizational trust is significantly positively related to innovative behavior (β = 0.206, p < 0.05, Model 5), showing Hypothesis 3 is supported. Finally, when EOR and organizational trust are entered simultaneously, the coefficient of EOR to innovative behavior declined from 0.081 (p < 0.05) to −0.048 (p > 0.05), and organizational trust is still positively related to innovative behavior (β = 0.232, p < 0.001, Model 6), therefore, Hypothesis 4 was supported.

Hypothesis 5 proposed that innovative climate moderates the effect of organizational trust on innovative behavior. We centralized organizational trust and innovative climate, and then generate interactive terms through organizational trust multiplied by the innovative climate. The interactive term between organizational trust and the innovative climate is positively related to innovative behavior (β = 0.147, p < 0.001, model 8). Furthermore, we plotted the interactive effect following the method recommend by Stone and Hollenbeck's. Figure 2 shows that the positive relationship between organizational trust and innovative behavior is stronger when employees perceived the climate as more innovative. Hence, Hypothesis 5 is supported.

Figure 2. Moderating effect of the innovative climate on the relationship between organizational trust and innovative behavior.

Hypothesis 6 predicts that innovative climate moderates mediating effect of organizational trust on the relationship between EOR and innovative behavior. We tested this hypothesis though the general path analytic framework proposed by Edwards and Lambert. The results are summarized in Table 4, which shows that the difference in indirect effect of EOR on innovative behavior was 0.150, with 95% confidence intervals, computed through bootstrapping by drawing 1000 random samples, excluding zero. Hypothesis 6 is supported.

Table 4. Results of the moderated path analysis.