The cognitive factors include perceived service quality, perceived service quality of
technology-based interfaces, perceived service quality of human-based interfaces, and
perceived risk (supporting H6 and H7). For this channel ‘perceived communication efforts’
did not emerge as a significant predictor of usage frequency, therefore H8 was not supported.
Branch channel
Table 5: Ordinal logistic regression of the branch channel usage | |||||
^ |
^ |
^ |
^ | ||
VARIABLES |
OR |
SE(OR) |
β |
95% CI (OR) |
p-value |
Perceived service quality |
2з3 |
.544 |
.846 |
(1.47,3.68) |
.000 |
Tangibles |
.68 |
.103 |
-.386 |
(.51,.92) |
.011 |
Perceived communication efforts |
.932 |
.135 |
-.07 |
(.702,1.24) |
.628 |
p-value = | |||||
Proportional Odds assumption |
χ2= 8.35 |
.214 |
Measures of fit
LRT (df) a |
17.22 (3) |
p-value |
Log- | ||
.000 |
likelihood |
24.34 | |||
AIC |
2.178 |
AIC*n |
790.761 |
Number obs. |
363 |
BIC b |
-1325.54 |
BIC’ |
0.46 |
a The likelihood ratio test is performed against the log-likelihood of the intercept-only model. In brackets are
degrees of freedom.
b The BIC criterion is more adequate for comparing models than just per se. The value of the statistic is reported
here for information purposes. It was used for comparison between models.
Implicit to the model presented, is the hypothesis that channel usage is influenced by
two groups of variables: a cognitive group of predictors and an affective group of predictors.
The final model for branch channel usage only included cognitive predictors: perceived
service quality, tangibles, and perceived communication efforts, supporting H7 and H8. No
affective variable was present in the final solution (H1, H2, H3 and H4 were not supported).
Perceived service quality of the ‘tangibles’ dimension only emerged as a significant influence
for the branch channel’s usage. The branch is the delivery channel where the physical,
22