152 /304
產品受歡迎資訊與網路購物：以消費者解讀為干擾變數
152
appearance, abilities, values, and attitudes in order to imitate that of their idols (Chan and
Prendergast, 2008). Young consumers reported that they were more likely to use products
endorsed by famous celebrity (Lafferty and Goldsmith, 1999). Thus, we use celebrity to
induce participants’ social inference. The scenarios are demonstrated in Appendix A and
the sample page of the experiment on the Website is illustrated in Appendix D.
In order to examine whether the expected inference (quality evaluation vs. social
comparison) was induced by the scenarios, three questions of manipulation check were
constructed. Participants were asked to answer these questions based on their decision
making process in the experiment. Two questions were used to examine whether the social
inference was successfully induced. The first question was: What level of impact did other
consumers’ purchase decision have on your purchase decisions? The second question
asked participants to rate their disappointment level if the product they intended to
purchase was out of stock. Past research has shown that if consumers fail to purchase the
same product as their reference group, they experience even stronger disappointment than
those who do not undergo the process of social comparison (Loewenstein, 1988; Luo,
2005). As for the quality inference, participants were asked to rate what level of impact
product quality had on their purchase decisions. The manipulation check questions are
listed in Appendix B.
3.3 Pilot Test
A pilot test was conducted in a computer lab located at a university in Taoyuan,
Taiwan. The purpose of conducting this preliminary study is to test logistics and gather
information prior to the formal study, in order to improve the study’s overall quality and
clarity. Specifically, the clarity of the experiment is referred to whether instructions are
comprehensible and easy to follow, wordings used in the experiment are appropriate, etc.
The samples consisted of 20 volunteer staff and students from the university, who were
randomly assigned to one of the four conditions, with 5 participants in each condition.
After completing the experiment, participants were asked to perform a face-to-face
interview to answer a series of open-ended questions, including 1) assess whether the
wordings used in the scenarios are clear and appropriate, 2) the length of the experiment is
reasonable, 3) the steps in the experiment are easy to follow, and 4) the difference of the
sales volume in each pair of products is sufficiently distinct. The pilot results consistently
matched the expectations of the researchers and confirmed the appropriateness of the
experimental design and material.