CHAPTER 4 RESULTS
4.2 HYPOTHESIS TESTS
4.2.1 Regression results of salesperson competence on customer trust
We ran hierarchical regression analyses to test the hypotheses. The regression results of social and professional competence on affective and cognitive trust are shown in Table 4-2. In Model 1 and 3, six control variables, including gender, age, education, tenure, self-esteem, and prior performance, were entered into the regression. Tenure (unstandardized beta = .03, p< .01) and prior performance (unstandardized beta = –.11, p< .05) explained a significant amount of variability in affective trust, but prior performance was not significant in cognitive trust. In Model 2 and 4, the two independent variables, social and professional competence, were added to the regression. As shown in Table 4-2, the two independentvariables explained a significant amount of variability in affective trust beyond that of the control variables (Δ𝑅𝑅2
= .38, p< .001). The unstandardized regression weight was significant for social competence (unstandardized beta = .42, p< .001) higher than for professional competence (unstandardized beta = .22, p < .01) in affective trust, but professional competence (unstandardized beta = .69, p< .001) was significant in cognitive trust (Δ 𝑅𝑅2 = .48,p< .001). Therefore, there is a significant relationship between social competence and affective trust, but not cognitive trust (unstandardized beta = .08, p> .05). There is also a significatn relationship between professional competence both affective and cognitive trust.Hypothesis 3a and 3b was supported.
29
Table 4-2. Regression results of salesperson competence on customer trust
Variables
Affective trust Cognitive trust
Model 1 Model 2 Model 3 Model 4
4.2.2 Direct effect of salesperson competence and mediation of customer trust on objective performance
Table 4-3 reports results of direct effect of salesperson competence and mediation of customer trust on objective performance. Model 5 includes control variables only, and we introduce the two independent variables (IV) (i.e., social and professional competence) in Model 6. As Model 6 shows, there is a significant positive relationship (Δ𝑅𝑅2 = .26, p< .001) between social competence (unstandardized beta = .60, p< .001) and professional competence (unstandardized beta = .44, p< .01), and objective performance, suggesting that when owning more social and professional competenceenhance objective performance, supporting Hypothesis 1a and 2a.
30
In Model 7, the two mediators(i.e., affective and cognitive trust)were added to the regression. The twovariables explained a significant amount of variability in objective performance beyond that of the control variables (Δ𝑅𝑅2 = .33, p< .001). The unstandardized regression weight was significant for affective trust (unstandardized beta = .67, p< .001) and cognitive trust (unstandardized beta = .30, p< .05). Therefore, there is a significant relationship between the mediators and the dependent variable.
In Model 8, social competenceand the two mediators were added to the regression. The twovariables explained a significant amount of variability in objective performance beyond that of the control variables (Δ𝑅𝑅2 = .33, p< .001). The unstandardized regression weight was significant for social competence (unstandardized beta = .50, p< .001) and affective trust (unstandardized beta = .47, p< .001), but not for cognitive trust (unstandardized beta = .19, p> .05). Therefore, Hypothesis 4a was supported.
In Model 9, professional competence, and the two mediatorswere added to the regression.
The two variables explained a significant amount of variability in objective performance beyond that of the control variables (Δ𝑅𝑅2 = .31, p< .001). The unstandardized regression weight was significant for professional competence (unstandardized beta = .31, p< .05) and cognitive trust (unstandardized beta = .62, p< .001), but not for affective trust (unstandardized beta = .28, p> .05). Therefore, Hypothesis 5a was supported.
In Model 10, the two IVsand the two mediators were added to the regression. The two IVs and affective trust explained a significant amount of variability in objective performance beyond that of the control variables (Δ𝑅𝑅2 = .31, p< .001). The unstandardized weight was significant for social competence (unstandardized beta= .40, p< .01) and professional competence (unstandardized beta = .31, p< .05) and affective trust (unstandardized beta = .48, p< .001) andcognitive trust (unstandardized beta =.28, p< .05).Therefore, the result supported a partial mediation because social and professional competences were still significant.
Table 4-3 Direct effect of salesperson competence and mediation of customer trust on objective performance
Objective performance
Model 5 Model 6 Model 7 Model 8 Model 9 Model 10
31
4.2.3 Direct effect of salesperson competence and mediation of customer trust on subjective performance
Table 4-4 reports results of direct effect of salesperson competence and mediation of customer trust on subjective performance. Model 11 includes control variables only, and we introduce social and professional competence in Model 11. As Model 12 shows, there is a significant positive relationship (Δ R2 = .17, p < .001) between social competence (unstandardized beta = .26, p< .01) and professional competence (unstandardized beta = .27, p< .01), andsubjective performance, suggesting that when owning more social and professional competence enhance objective performance, supporting Hypothesis 1b and 2b.
32
In Model 13, the two mediators were added to the regression. The two variables explained a significant amount of variability in subjective performance beyond that of the control variables (Δ 𝑅𝑅2 = .20, p< .001). The unstandardized weight was significant for affective trust (unstandardized beta = .35, p< .001) and cognitive trust (unstandardized beta
= .22, p< .05). Therefore, there is a significant relationship between the mediators and dependent variables.
In Model 14, social competence and the two mediators, were added to the regression.
The two variables explained a significant amount of variability in subjective performance beyond that of the control variables (Δ𝑅𝑅2 = .21, p< .001). The unstandardized weight was significant for affective trust (unstandardized beta = .27, p< .01) and cognitive trust (unstandardized beta= .19, p< .05), but not for social competence (unstandardized beta= .18, p> .05). Therefore, Hypothesis 4b was supported.
In Model 15, professional competence and the two mediators were added to the regression. The two variables explained a significant amount of variability in subjective performance beyond that of the control variables (Δ𝑅𝑅2 = .21, p< .001). The unstandardized weight was significant for cognitive trust (unstandardized beta= .32, p< .001), but not for affective trust (unstandardized beta= .10, p> .05) and professional competence (unstandardized beta= .20, p> .05). Therefore, Hypothesis 5b was supported.
In Model 16, the two IVs and the two mediators were added to the regression. The four variables explained a significant amount of variability in subjective performance beyond that of the control variables (Δ𝑅𝑅2 = .22, p< .001). The unstandardizedweight was significant for affective trust (unstandardized beta= .27, p< .01), but not for social competence (unstandardized beta= .14, p> .05), and professional competence (unstandardized beta= .14, p> .05) and cognitive trust(unstandardized beta= .11, p> .05). Therefore, the result supported afull mediation because social and professional competences wereno significant.
Table 4-4. Direct effect of salesperson competence and mediation of customer trust on subjective performance
Subjective performance
Variables Model 11 Model 12 Model 13 Model 14 Model 15 Model 16
B SE B SE B SE B SE B SE B SE
33
4.2.3 Sobel and bootstrapping test
Sobel test result in Table 4-5 demonstrates the mediating role of customer’s trust (i.e., affective and cognitive trust) on the relationship between salesperson competence (i.e., social and professional competence) positive affectivity and salesperson performance (i.e., objective and subjective performance). As given in Table 4-5, the results confirmed that all the indirect effects were statistically significant. The relationships between social competence and objective performance (Sobel test statistic: 5.54, p< 0.01), and subjective performance (Sobel test statistic: 4.85, p< .05) were mediated via affective trust. The results also shown the relationships between professional competence and objective performance (Sobel test statistic:
34
5.27, p< 0.001), and subjective performance (Sobel test statistic: 4.95,p< .05) were mediated via cognitive trust.
In addition, the bootstrap test (Preacher & Hayes, 2004, 2008)examined the robustness of mediation test by generating 1,000 bootstrap samples. If the confidenceinterval (CI) does not include 0, the indirect effectis significant and mediation is established. As shown in Table 4-5, the bootstrap results of indirect effect via two mediators (i.e., affective and cognitive trust) were significant. The 95% confidence interval excluded 0.
Table 4-5 Results of Sobel and bootstrapping test
Path Sobel test
Bootstrap results for indirect effects Indirect effects of IV
on DV through Indirect effect testing Test
statistic Std. Error Bias SE Lower Upper Social competence → affective
trust → objective performance 5.54*** .080 -.001 .078 .1684 .4830 Social competence → affective
trust → subjective performance 4.85*** .050 -.001 .067 .0529 .3259 Professional competence →
cognitive trust → objecti ve performance
5.27*** .090 -.002 .015 .0369 .3785 Professional competence →
cognitive trust → subjective performance
4.85*** .060 -.002 .013 .0676 .4186
*p﹤.05; **p﹤.01; ***p﹤.001.
Bootstrap same size =1000.