151 /304
151
臺大管理論叢
第
28
卷第
1
期
that the differences between the percentage of choosing broad-appeal and narrow-appeal
products were obvious and matched with the characteristics of broad-appeal and narrow-
appeal products, the products chosen by the norming test were thus used in the real
experiment. Interestingly, the average percentage of choosing the three products was 79%
for the broad-appeal products vs. 21% for the narrow-appeal products, which were close
to the 80/20 distribution reported in the literature (Brynjolfsson, Hu, and Simester, 2011;
Pareto, 1964).
In the real experiment, the sales volume of each product alternative was revealed to
the participants. As we have mentioned, the sales volume of the broad-appeal product was
higher than that of the narrow-appeal product in the condition of congruence, whereas the
sales volume of the narrow-appeal product was higher than that of the broad-appeal
product in the condition of incongruence. Prior research shows that consumers feel the
difference when the sales volume of two products is 2-fold apart (Tucker and Zhang,
2011). In order to make the difference of the sales volume sufficiently distinct, the
difference in each pair of products was approximately 6-fold, with 64 for the high sales
volume and 10 for the low sales volume. Therefore, in the congruent condition of cookies,
the sales volume of the broad-appeal alternative was 64, and 10 for the narrow-appeal,
while the sales volume was reversed in the incongruent condition.
In the experiment, two different scenarios were used to prime the inferences derived
from popularity information. Both scenarios explained that the sales volume was obtained
via a survey that was designed to explore how popular various products were among
different groups of people. In the scenario for quality evaluation, it was stated that the
sales volumes were contributed by food connoisseurs who were highly sensitive to food
quality. On the other hand, in the scenario for social inference, the sales volumes were
contributed by celebrities, who were favorably popular. The reason why these two groups
of people were used in the scenarios was due to their distinctive characteristics. Food
connoisseurs are known to have high standards regarding the quality of food and are
considered experts who are perceived to possess knowledge about the product (Ohanian,
1990). We therefore use food connoisseurs to induce the quality inference. On the other
hand, celebrities are typically recognized as a popular comparison target who can create a
significant effect of social comparison on consumers and drive consumers to feel entitled
to own what they have (Amos, Holmes, and Strutton, 2008; Silvera and Austad, 2004; Wu
and Lee, 2008a, 2008b). Prior research shows that young people especially tend to
consider celebrity idols as their idealized self-images and want to revise their physical