39 / 414
臺大管理論叢
第
27
卷第
1
期
39
Thus, Harman’s one-factor test was conducted to determine the extent of this variance. The
unrotated factor analysis for pooled data indicated that the first factor accounted for merely
29.57% of the variance. Based on Harman’s one-factor test, CMV was not considered
significant (Podsakoff et al., 2003). A confirmatory factor analysis was estimated in which all
items included in the structural model were restricted to load on a single factor (Podsakoff et
al., 2003). The fit indices indicated a poor model fit. Overall, the results of the two tests
indicated that CMV is not an issue in the current study. Table 1 shows the means, standard
deviations, and correlations matrix for the constructs. Appendix A summarizes the results of
the item description, factor loadings, AVE, and reliability tests.
Table 1 Means, Standard Deviations and Correlations
M D 1
2
3 4 5 6 7 8
1. Asset Specificity
3.26 0.95 1
2. Quality of Customer Interaction 3.24 0.87 0.60* 1
3. Decision-making Uncertainty 2.72 0.94 -0.03 -0.02 1
4. Co-production
2.91 0.75 0.38* 0.35* 0.20* 1
5. Special Treatment Benefits
3.35 1.09 0.26* 0.12* 0.01 0.11* 1
6. Social Benefits
3.25 0.92 0.20* 0.16* -0.01 0.31* 0.11* 1
7. Confidence Benefits
3.41 0.88 0.15* 0.11* -0.07 0.21* 0.12* 0.46* 1
8. Share of Wallet
0.48 0.19 0.36* 0.34* 0.06 0.36* 0.40* 0.29* 0.34* 1
Note: *
p
< 0.05.