161 /304
161
臺大管理論叢
第
28
卷第
1
期
the effect of popularity information is consistent. Finally, the student population in this
research is relatively high (48%). This leads to the result that the age distribution in this
research is significantly lower compared to other research. However, responses were
voluntary, thus, inevitably subject to self-selection biases. An additional K-S test was
performed to examine the collected responses between the students and non-students, and
the result showed no significance. Furthermore, students and young people are known as a
major population for online purchases (Dai, Forsythe, and Kwon, 2014). Thus, it is
legitimate to assume that there may be more student respondents to our online survey than
of other age groups. Much research even aims to examine the student sample when
conducting Internet-related studies (Kuo and Wu, 2012; Lin and Lekhawipat, 2014; Yeh
and Li, 2014). This leads us to believe that our sample in this experiment is appropriate
and reflects the characteristics of the target Internet users. An additional K-S test was
performed to examine the collected responses between the students and non-students and
the result shows no significant difference. Nevertheless, we still do not rule out the
possibility that subjects’ ages may sometimes exert influence on consumers’ online
purchase behavior. Therefore, generalization of our research, especially to older people,
may have limitations. Future research conducted with different age groups is strongly
recommended.
5.6 Future Research
There are many interesting future directions that could be derived from this research.
For example, our research has mentioned that social comparison may be a possible drive
to exert the social inference. Different kinds of social comparison may be triggered when
consumers compare with someone similar (e.g., other consumers) or someone better (e.g.,
experts and celebrities) for different motivations. Future research could focus even more
on this area and explore social comparison with different directions (e.g., someone similar,
better-off, and worse-off) and different motivations (e.g., self-evaluation and self-
enhancement). Another interesting direction would be to examine consumers’ emotions
when different kinds of inferences are induced. Particularly for the social inference, past
literature has indicated that consumers perceive stronger happiness and disappointment
when social inference (particularly when the effect of social comparison) is induced
(Ackerman, MacInnis, and Folkes, 2000; Hoch and Loewenstein, 1991; Wu and Lee,
2008a, 2008b). Consistent with this prediction, the results of our manipulation check show
that participants with the social inference feel marginally more disappointed when the