41 / 414
臺大管理論叢
第
27
卷第
1
期
41
Hypothesis 4 stated that co-production would have a positive effect on special treatment
benefits. The effect of co-production on special treatment benefits was significantly positive
(
β
= 0.181,
p
< 0.05), and therefore, H4 was supported. Hypothesis 5 stated that
co-production would have a positive effect on social benefits. The effect of co-production on
social benefits was significantly positive (
β
= 0.353,
p
< 0.05), and therefore, H5 was
supported. Hypothesis 6 stated that co-production would have a positive effect on confidence
benefits. The effect of co-production on confidence benefits was significantly positive (
β
=
0.229,
p
< 0.05), and therefore, H6 was supported.
Hypothesis 7 stated that special treatment benefits would have a positive effect on share
of wallet. The effect of special treatment benefits on share of wallet was significantly
positive (
β
= 0.068,
p
< 0.05), and therefore, H7 was supported. Hypothesis 8 stated that
social benefits would have a positive effect on share of wallet. The effect of social benefits
on share of wallet was significantly positive (
β
= 0.030,
p
< 0.05), and therefore, H8 was
supported. Hypothesis 9 stated that confidence benefits would have a positive effect on share
of wallet. The effect of confidence benefits on share of wallet was significantly positive (
β
=
0.055,
p
< 0.05), and therefore, H9 was supported.
This study used a five-point Likert-type scale to measure special treatment benefits,
social benefits, and confidence benefit (From 1 to 5), and it also used percentage to measure
share of wallet (from 10% to 100%). The different levels of scales may result in the low
estimation of coefficients. In fact, t-values were normal (from 3.194 to 8.057). In the study of
Verhoef (2003), the effects of commitment on customer share (
β
= 0.03,
t
= 2.58), loyalty
program on customer share (
β
= 0.04,
t
= 2.22), and direct mailing on customer share (
β
=
0.01,
t
= 2.31) were significantly positive. The estimation of coefficients in the study of
Verhoef (2003) was similar to that in the current study. The author also estimated a
regression model to relate special treatment benefits, social benefits, and confidence benefits
to share of wallet. The parameters of special treatment benefits (
β
= 0.065,
t
= 8.42), social
benefits (
β
= 0.031,
t
= 3.02), and confidence benefits (
β
= 0.051,
t
= 4.81) were positively
significant. The estimations of coefficients and t-values in the regression model were similar
to those in the SEM model.