87 / 372
臺大管理論叢
第
26
卷第
2
期
87
as either accurate estimations or drop-off errors were considered non-drop-off errors. In the
example of the $1.99 nail clippers, quantity estimations greater than 47 or less than 23 were
categorized as non-drop-off errors; that is, errors that point to an incorrect overestimation or
underestimation due to various possible factors, including arithmetic mistakes, inattention,
carelessness, and so on (Bizer and Schindler, 2005).
3.2.3 Results
An ANOVA was performed for quantitative estimations. The interactions of evaluation
conditions and nine-ending effect were significant in the scenarios of nail clipper and baseball
cap showed in Table 1 (Nail clipper:
Difference
SE
= 1.75 vs.
Difference
SQE
= 0.7 vs.
Difference
JE
= 0.59,
F
SE-SQE-JE
(2,231) = 2.44,
p
SE-SQE-JE
< 0.1; Baseball cap:
Difference
SE
= 3.55 vs.
Difference
SQE
= 0.48 vs.
Difference
JE
= 0.36,
F
SE-SQE-JE
(2,231) = 2.52,
p
SE-SQE-JE
< 0.1), but
insignificant in the case of the battery (
Difference
SE
= 0.39 vs.
Difference
SQE
= 0.12 vs.
Difference
JE
= 0.1,
F
SE-SQE-JE
(2,231) = 0.53,
p
SE-SQE-JE
= 0.95). That is, the differences in mean
price perceived quantitative estimations of nine-ending and zero-ending prices in the SQE
condition was smaller than that in the SE condition but greater than that in the JE condition.
Besides, stimuli in SE and JE conditions were also significant for the nail clippers and
baseball cap (
Difference
SE
= 1.75 vs.
Difference
JE
= 0.59,
F
SE- JE
(1,163) = 10.35,
p
SE-JE
< 0.01;
Difference
SE
= 3.55 vs.
Difference
JE
= 0.36,
F
SE- JE
(1,163) = 8.64,
p
SE-JE
< 0.01), but not
significant for the battery (
Difference
SE
= 0.39 vs.
Difference
JE
= 0.1,
F
SE- JE
(1,163) = 1.40,
p
SE-JE
= 0.254). Most of the results, as expected, showed that the difference of quantitative
estimations between nine- and zero-ending digits was greater in JE mode than in SE mode.
These findings were re-confirmed and consistent with the findings in Study 1.
In order to examine the error analyses as proposed by Bizer and Schindler (2005), the
quantitative estimations were also classified according to their consistency with the nine-
ending pricing effect by calculating the difference in percentage between nine-ending and zero-
ending prices among each response type. However, the battery stimulus was excluded because
of its insignificance (see Table 2). The results demonstrated that the accuracy percentage in
zero-ending digit was almost the same in SE, SQE, and JE modes (Nail clipper:
Accurate
SE
=
83% vs.
Accurate
SQE
= 80% vs.
Accurate
JE
= 76%; Baseball cap:
Accurate
SE
= 77% vs.
Accurate
SQE
= 70% vs.
Accurate
JE
= 82%), but the accuracy percentage in nine-ending digit was
least in the SE condition, followed by the SQE and the JE condition in the products of the nail
clippers and baseball caps (Nail clipper:
Accurate
SE
= 44% vs.
Accurate
SQE
= 61% vs.
Accurate
JE
= 68%; Baseball cap:
Accurate
SE
= 41% vs.
Accurate
SQE
= 57% vs.
Accurate
JE
= 71%).
Consistent with H2 and H3, the findings, an decreasing difference with the order of SE,