95 / 372
臺大管理論叢
第
26
卷第
2
期
95
S-K both comprised one half of the S-K-R-T group; S-K-R-T-P-U comprised only one third
of the S-K-R-T-P-U group and R-T-S-K-P-U and R-T-P-U-S-K comprised the other two-
thirds. The results showed no order effect among these groups.
3.4.2 Dependent Variable
Study 4 had participants answer questions regarding pricing perception by responding
to the same statements as Study 1. Besides, following the study of Thomas and Morwitz
(2005), Study 4 measured the response time that is often used by numerical cognition
researchers to determine the cognitive processes used by participants (e.g., Dehaene, 1997;
Dehaene et al., 1990; Hinrichs, Yurko, and Hu, 1981). Chaiken (1980) also proposes that the
response time one spends on finishing a task can be regarded as a good way of measuring
cognitive effort. In brief, the shorter the response time one needs to evaluate the price
magnitude, the more likely that person is in a heuristic process. By contrast, the longer the
response time, the more likely the nine- and zero-ending digits are being evaluated in a
systematic process.
3.4.3 Results
In Study 4, in order to re-confirm the boundary conditions of nine-ending effect, we used
an ANOVA to examine the results. As shown in Table 4, the interaction among two evaluation
conditions, high and low task complexity, and nine- and zero-ending prices was significant
(
F
(3, 76) = 14.330,
p
< 0.01). That is, the difference of perceived price magnitude between SE
and JE was a decreasing curve when task complexity was increasing (see Figure 1).
Furthermore, in order to re-confirm the inference that the increasing degree of task complexity
would play an important role on nine-ending effect, first, the difference between SE and JE in
S-K was examined and showed the significance (
Difference
SE
= 1.03 vs.
Difference
JE
= 0.23,
F
S-K
(1, 38) = 32.916,
p
< 0.01). This finding was consistent with Study 1 that nine-ending
effect occurred under the tasks that are less complex. Secondly, the difference in S-K-R-T was
also significant (
Difference
SE
= 0.91 vs.
Difference
JE
= 0.24,
F
S-K-R-T
(1, 38) = 15.344,
p
< 0.01).
This finding also proved the hypothesis that nine-ending effect was supported when the
degree of task complexity did not exceed an individual’s cognitive loading.
Combined with both of the results, the existence of nine-ending effect was re-confirmed
to be the same as that in Study 1 under less complex conditions. However, as expected, the
difference in S-K-R-T-P-U, a condition with high task complexity that may exceed one’s
cognitive burden, was insignificant (
Difference
SE
= 0.90 vs.
Difference
JE
= 0.77,
F
S-K-R-T-P-U
(1, 38)
= 0.468,
p
= 0.498). These results not only re-confirmed H5 that the more task complexity, the
less nine-ending effect between the comparison of the difference of the perceived price