Fit statistics resulting from testing this model are shown in
Table II. Most fit statistics improve slightly with this change.
Further, the resulting nested models can be compared with a x2
difference test. It suggests a significant improvement in fit
through the addition of these paths (Dx2 ¼ 18.2, 2 df, p , 0.01).
The path estimate from positive affect to consumer satisfaction is
0.17 (p , 0:01) suggesting that in addition to the indirect effect
implied above, positive affect positively and directly relates to
satisfaction. In contrast, the path estimate from negative affect to
customer satisfaction is insignificant at 20.07 (p . 0:01). The
R2 for positive affect improves from 0.75 to 0.77 with the
additional path while the R2 for WOM remains at 0.73 after
rounding (to the 2nd decimal). Therefore, the results here and
above suggest that positive affect has greater diagnosticity than
does negative affect. Thus, there is support for a dual effects
theory, meaning that affect influences service consumer reactions
both directly and indirectly through perceived service value.