Escolar Documentos
Profissional Documentos
Cultura Documentos
B U C K K. W. PEI
Abstract
Based upon Hogarth and Einhorn's "belief-adjustment model", the present research tested the model's
recency effects prediction in a "performance auditing" context using state auditors. The results showed that
the auditors' "belief revisions" were subject to the recency effects predicted by the model. The results also
validated an important assumption of the mendel that serial position effects were absent in a short evidence
series. That is, serial position had no effect on the subject's subjective evaluation of evidence. Finally, analyses
of the subjects' intermediate belief changes suggested that they displayed a "confirmatory attitude" towards
evidence rather than a "disconfirmatory attitude" found in the prior studies conducted in financial auditing
contexts.
T h e p r o c e s s b y w h i c h n e w i n f o r m a t i o n is
integrated with current beliefs has long been an
important area of research in decision making.
O n e i m p o r t a n t f o c u s o f t h i s r e s e a r c h is t h e
study of order effects on belief revision.
However, the research reveals that the order of
information presentation
sometimes has no
effect, m a y i n d u c e p r i m a c y effects, o r m a y
c a u s e r e c e n c y e f f e c t s o n b e l i e f r e v i s i o n (e.g.
N i s b e t t & Ross, 1 9 8 0 ) . R e c e n t l y , H o g a r t h &
Einhorn (1990)
proposed
and validated a
descriptive model, called the "belief-adjustment
model", to account for the mixed results found
in this literature. The model posits that decision
makers tend to use a simple anchoring-andadjustment strategy for belief revision, and that
* This research is partially funded by a summer research grant to the first author by the FGIA program of Arizona State
University. The authors gratefully acknowledge comments by Robin M. Hogarth, Robert Ashton and three anonymous
referees on the earlier versions of this paper.
169
170
performance
audit is to evaluate
a social
programs efficiency and effectiveness in utilizing
public resources
(e.g. Herbert,
1979). In contrast, the objective of a financial audit is to attest
to the company managements
assertions in the
financial statements.
While the generic process
of conducting
both financial and performance
audits may be identical,
the decision
consequences
perceived
by the auditors may be
quite different. Auditors performing
a financial
audit are subject
to potential
legal penalties
when they fail to detect misrepresentations
in
the financial statements. On the other hand, the
possibility
of legal penalties
is absent
for
the auditors performing
a performance
audit.
Evidence
in the decision-making
literature
suggests that different perceptions
consequences
(including
their
indeed, frequently
1987;
1982).
Kahneman
alter judgments
&
Tversky,
of decision
magnitude),
(e.g. Hogarth
1979;
Payne,
Fischhoff
& Reyth-Marom,
1983;
Einhorn
appropriate
for negative evidence to be given
more weight than positive evidence, since the
primary risk that auditors face is that errors
in financial statements
may
(pp. 639-640).
This suggests
transaction
cycles,
collectability
of accounts
receivable).
Thus, the evidence
presented
in
those studies may be high in diagnosticity.
Comparatively,
performance
auditing of a social
program
necessarily
involves
examining
a
broader scope of information with the available
evidence
subject
to uncertainty
in making
proper causal inferences.
That is, the available
go undetected
that the discon-
is low in diagnosticity,
The preceding argument does not suggest that the decision environment
of state auditors is penaltyfree.
was to highlight legal penalties as a salient factor that clearly distinguish the decision environments
between
of audit. Further, these legal penalties frequently are immense.
See
Asare ( 1990)
people
Our intention
the two types
for an exception.
This phenomenon
is frequently
labeled as confirmatory
bias. Note, most of thse confirmatory
bias studies were
interested in examining peoples hypothesis testing and/or information search behaviors. Thus, the results from Ashton &
Ashton (1988)
and Tubbs et al. (1990)
should not be interpreted
as contradictory,
since their findings relate to the
weighing of evidence rather than the searching of evidence.
AUDITOR BELIEFREVISIONS
are m o r e affected b y p o s i t i v e e v i d e n c e than
n e g a t i v e e v i d e n c e (e.g. H o c h & Ha, 1986;
D e i g h t o n , 1984; Reed, 1986). In brief, o t h e r
than testing the m o d e l ' s o r d e r effects prediction,
this s t u d y also e x p l o r e s w h e t h e r p e r f o r m a n c e
a u d i t o r s d i s p l a y a c o n f i r m a t o r y ( r a t h e r than
a d i s c o n f i r m a t o r y ) a t t i t u d e t o w a r d s evid e n c e b e c a u s e o f t h e d i f f e r e n c e s in t h e task
environment.
As a further a t t e m p t to validate the beliefadjustment model, the p r e s e n t study also tests a
crucial assumption o f the belief-adjustment model:
t h e a b s e n c e o f serial p o s i t i o n effects ( d e s c r i b e d
later). To the best o f o u r knowledge, this
assumption has n o t b e e n tested b y Hogarth &
Einhorn ( 1 9 9 0 ) , n o r b y the p r i o r auditing studies.
T h e r e m a i n d e r o f this p a p e r p r o c e e d s
as follows. First, w e s u m m a r i z e t h e beliefa d j u s t m e n t m o d e l p r o p o s e d b y H o g a r t h and
Einhorn and describe the model's predictions
t e s t e d in t h e p r e s e n t study. Second, w e highlight t h e d i s t i n c t n a t u r e o f t h e p e r f o r m a n c e
a u d i t i n g tasks u s e d in t h e study. Third, w e
d e s c r i b e t h e r e s e a r c h m e t h o d . Fourth, w e
r e p o r t t h e results. Finally, w e e l a b o r a t e o n t h e
r e s e a r c h findings and suggest s o m e a v e n u e s for
future r e s e a r c h .
BELIEF-ADJUSTMENT MODEL
Hogarth & Einhorn's ( 1 9 9 0 ) belief-adjustment
m o d e l a s s u m e s that p e o p l e u p d a t e beliefs using
a s i m p l e a n c h o r i n g and a d j u s t m e n t strategy.
Hogarth and Einhorn argue that decision makers
are a d a p t i v e to t h e task e n v i r o n m e n t ; t h e r e f o r e
the p r e d i c t i o n o f a l t e r n a t i v e o r d e r effects o n
belief revision can be better understood by
explicitly considering the characteristics of the
d e c i s i o n tasks ( a l s o see Payne, 1982; Payne e t
al., in press). To illustrate this adaptive behavior,
they identify t h r e e key task variables: complexity,
l e n g t h o f e v i d e n c e series, and r e s p o n s e m o d e .
Based u p o n the belief-adjustment model, Hogarth
and Einhorn have illuminated various task conditions u n d e r w h i c h alternative o r d e r effects (i.e.
p r i m a c y r e c e n c y no effect) m a y prevail. Following
t h e i r d i s t i n c t i o n , t h e m o d e l ' s p r e d i c t i o n s are
briefly s u m m a r i z e d using t h r e e task categories:
171
= r e v i s e d b e l i e f after e v a l u a t i n g k
p i e c e s o f e v i d e n c e , 0 <~S~<~ 1,
= a n c h o r o r p r i o r o p i n i o n . T h e initial
b e l i e f is d e n o t e d So,
172
s(x~)
-- s u b j e c t i v e e v a l u a t i o n o f t h e k t h p i e c e
of evidence,
R
= the r e f e r e n c e p o i n t against w h i c h t h e
i m p a c t o f t h e k t h p i e c e o f e v i d e n c e is
evaluated. (Similar to t h e p r i o r studies,
this s t u d y a s s u m e s R = 0 b e c a u s e t h e
e x p e r i m e n t a l tasks u s e d i n v o l v e evaluation r a t h e r than e s t i m a t i o n ) .
ct and ~ are t h e sensitivity t o w a r d n e g a t i v e a n d
p o s i t i v e e v i d e n c e , r e s p e c t i v e l y ( 0 ~< ~,
~<1).
To illustrate w h y r e c e n c y effects occur, c o n s i d e r
belief revisions w i t h t w o p i e c e s o f e v i d e n c e
u n d e r t w o a l t e r n a t i v e o r d e r s ( + , - ) vs ( - , + ) .
G i v e n e q u a t i o n ( l b ) , p o s i t i v e e v i d e n c e in t h e
(+,-)
o r d e r has a s m a l l e r u p w a r d r e v i s i o n
effect than t h e p o s i t i v e e v i d e n c e in t h e ( - , + )
order. This is b e c a u s e t h e a n c h o r o f t h e f o r m e r
( 1 - S o ) is s m a l l e r than t h e a n c h o r o f t h e
l a t t e r (1-$1 ), ceteris paribus. Likewise, given
e q u a t i o n ( l a ) , n e g a t i v e e v i d e n c e has a s m a l l e r
d o w n w a r d r e v i s i o n effect in t h e ( - , + ) o r d e r
than t h e e v i d e n c e in t h e ( + , - ) o r d e r b e c a u s e
t h e a n c h o r o f t h e former, So, is s m a l l e r than t h e
a n c h o r o f t h e latter, St, ceteris paribus. T h e s e
c o n t r a s t effects p r o v i d e t h e m o d e l p r e d i c t i o n s
that t h e final p o s i t i o n o f b e l i e f r e v i s i o n r e a c h e d
u n d e r t h e ( + , - ) o r d e r will b e l o w e r than its
counterpart reached under the (-,+)
order
d u e to t h e d i f f e r e n c e s ( b e t w e e n t h e t w o
o r d e r s ) in t h e a n c h o r s at k = 1. This p r e d i c t i o n
is t e r m e d t h e recency effect.
H o g a r t h a n d E i n h o r n also d e m o n s t r a t e d t h e
a b o v e r e c e n c y analytically. T h e y d e f i n e d the
r e c e n c y effect as ( s e e H o g a r t h & E i n h o r n 1990,
a p p e n d i x C for d e r i v a t i o n ) :
D = s(-,+)-s(+,-),
where D = -
~s(x_)s(x+) ( a n d
D>
0).
(2)
has n o effect o n s u b j e c t i v e e v a l u a t i o n o f
e v i d e n c e . That is, t h e s t r e n g t h o f s ( x k ) w i l l n o t
d e c l i n e d u e to d e c r e a s i n g sensitivity t o w a r d
e v i d e n c e . This a s s u m p t i o n is also i m p l i e d in
t h e task t a x o n o m y m e n t i o n e d earlier. A k e y
d i f f e r e n c e b e t w e e n c a t e g o r i e s ( B ) a n d ( C ) rests
o n the l e n g t h o f e v i d e n c e ; a s h o r t ( l o n g )
e v i d e n c e series i n d u c e s r e c e n c y ( p r i m a c y ) .
Hogarth and Einhorn argue that for a short ( l o n g )
e v i d e n c e series, sensitivity t o w a r d e v i d e n c e
s h o u l d b e c o n s t a n t ( m a y d e c l i n e ) . Thus, for
s(xk), t h e e v a l u a t i o n is insensitive to t h e serial
p o s i t i o n in a s h o r t e v i d e n c e series, w h e r e a s t h e
e v a l u a t i o n m a y b e s u b j e c t to fatigue ( t h u s
i n d u c i n g p r i m a c y ) in a l o n g e v i d e n c e series. In
t h e p r e s e n t study, w e e x p l i c i t l y t e s t e d t h e
p r e c e d i n g crucial a s s u m p t i o n w h i l e t e s t i n g t h e
r e c e n c y effects p r e d i c t i o n o f t h e m o d e l .
AUDITOR BELIEFREVISIONS
to d r a w u n e q u i v o c a l causal inferences. In part,
this is b e c a u s e a social p r o g r a m ' s efficiency a n d
effectiveness are f r e q u e n t l y affected b y factors
u n r e l a t e d to t h e p r o g r a m ' s i n t e r v e n t i o n p r o c e s s
( G a m b i n o & R e a r d o n , 1981; Nachmias, 1979;
Rossi et al., 1979). 4
The preceding observations may be further
e m p h a s i z e d from a causal r e a s o n i n g p e r s p e c tive. A c c o r d i n g to E i n h o r n & H o g a r t h ( 1 9 8 6 ) ,
a d e c i s i o n m a k e r ' s causal j u d g m e n t will b e
s u b j e c t to h i g h e r u n c e r t a i n t y w h e n t h e causeand-effect r e l a t i o n s h i p is l o w in c o n t i g u i t y in
t i m e and space, a n d w h e n t h e n u m b e r o f
alternative causal explanations increases. Jointly,
t h e s e o b s e r v a t i o n s suggest that in c o n d u c t i n g
a p e r f o r m a n c e audit, t h e a u d i t o r s m a y b e
c o n f r o n t e d w i t h h i g h u n c e r t a i n t y in j u d g i n g a
social p r o g r a m ' s efficiency a n d effectiveness,
s i n c e t h e d i a g n o s t i c i t y o f t h e available e v i d e n c e
is fairly low.
W i t h r e s p e c t to b e l i e f revision, t h e p r e c e d i n g
o b s e r v a t i o n s p r e s e n t an i n t e r e s t i n g question.
Will t h e d e c i s i o n m a k e r ' s sensitivity t o w a r d s
e v i d e n c e b e affected b y e v i d e n c e diagnosticity?
W h i l e t h e p r e c e d i n g q u e s t i o n has n o t b e e n
e x p l i c i t l y a d d r e s s e d in t h e p r i o r beliefa d j u s t m e n t m o d e l studies, available e v i d e n c e
suggests that w h e n e v i d e n c e d i a g n o s t i c i t y is
low, p e o p l e t e n d to b e m o r e sensitive to
p o s i t i v e e v i d e n c e t h a n to n e g a t i v e e v i d e n c e (cf.
K l a y m a n & Ha, 1987). F o r e x a m p l e , H o c h & Ha
( 1 9 8 6 ) h y p o t h e s i z e d a n d f o u n d that in j u d g i n g
t h e quality o f a p r o d u c t , c o n s u m e r s ' j u d g m e n t s
w e r e affected b y a d v e r t i s i n g w h e n t h e e v i d e n c e
o f p r o d u c t quality w a s l o w in diagnosticity,
whereas their judgments were unaffected by
advertising when the evidence of product
quality was high in diagnosticity. Such findings,
as t h e y argued, i n d i c a t e that p e o p l e are
m o r e easily p r o n e to e n g a g i n g in c o n f i r m a t o r y
b e h a v i o r s w h e n t h e available e v i d e n c e is
a m b i g u o u s , s i n c e t h e y t e n d to i n t e r p r e t ambig-
173
u o u s e v i d e n c e as b e i n g m o r e b e l i e f - c o n s i s t e n t ,
and t e n d to give such e v i d e n c e h i g h e r weightings
than w a r r a n t e d . Similar findings w e r e also
d o c u m e n t e d in D e i g h t o n ( 1 9 8 4 ) . In a d d i t i o n ,
in a s t u d y that i n v o l v e d j u d g i n g t h e effectiveness o f a c r i m e p r e v e n t i o n p r o g r a m , R e e d
( 1 9 8 6 ) f o u n d that s u b j e c t s r e a c t e d differently
to t h e s a m e e v i d e n c e d e p e n d i n g u p o n w h e t h e r
t h e y w e r e t o l d t h e p r o r a m h a d b e e n successful
o r u n s u c c e s s f u l in t h e past. Specifically, using
h i e r a r c h i c a l r e g r e s s i o n analysis, s h e f o u n d that
the amount of judgment variation explained by
t h e s a m e e v i d e n c e was h i g h e r for t h e successful
s u b j e c t g r o u p ( i n c r e m e n t a l R square, 3 3 % )
than for t h e failure s u b j e c t g r o u p ( i n c r e m e n t a l
R square, 15% ). In part, this e v i d e n c e suggests
that s u b j e c t s m a y h a v e r e a c t e d in a m o r e
c o n g r u o u s ( p e r h a p s s e n s i t i v e ) m a n n e r to t h e
evidence when they were prompted positively
r a t h e r t h a n negatively.
To e x p l o r e t h e effects o f e v i d e n c e diagnosticity o n b e l i e f revision, w e c o n d u c t e d an
e x p e r i m e n t ( d e s c r i b e d in t h e n e x t s e c t i o n )
involving t w o p e r f o r m a n c e a u d i t i n g tasks
characterized by high uncertainty. Our a priori
e x p e c t a t i o n w a s that t h e p e r f o r m a n c e a u d i t o r s
would display a confirmatory attitude towards
e v i d e n c e , as o p p o s e d to t h e d i s c o n f i r m a t o r y
a t t i t u d e f o u n d in t h e p r i o r a u d i t i n g studies.
RESEARCH M E T H O D
Design
A 2 2 2 e x p e r i m e n t a l d e s i g n w a s u s e d to
test t h e b e l i e f - a d j u s t m e n t m o d e l p r e d i c t i o n s .
All s u b j e c t s e x a m i n e d t w o c a s e s c e n a r i o s - - an
efficiency a u d i t a n d an effectiveness audit.
W i t h i n e a c h audit, s u b j e c t s w e r e r a n d o m l y
assigned to o n e o f t h e four b e t w e e n - s u b j e c t s
conditions. Two orders of presentation were
used: t w o p i e c e s o f p o s i t i v e e v i d e n c e f o l l o w e d
For example, the social program examined in the present study involves providing services to move the recipients of
public assistance into a self-support status, thereby reducing state welfare roles. If the output measures indicate that the
program is relatively "over-staffed" for a given fiscal year, such evidence may be easily refuted as resulting from "temlx~rary
idle capacity". Likewise, when the program shows a lack of success (e.g. a low employment rate of its recipients), such
evidence could be attributed to the local state's economy.
174
So
S~
S2
$3
$4
$4-So
51.88
51.44
45.01
65.78
36.38
67.44
54.14
54.22
61.62
55.78
9.75
4.33
(--++)
(++--)
41.67
47.22
27.22
53.01
27.22
56.01
44.44
52.56
52.22
48.33
10.56
1.11
(--++)
(++--)
46.47
49.33
35.39
59.39
31.53
61.72
49.01
53.39
56.65
52.06
10.18
2.72
52.51
54.22
45.01
67.01
40.63
73.33
58.88
68.74
69.01
64.78
16.51
10.56
(--++)
(++--)
55.01
57.22
42.78
64.44
42.89
70.22
56.11
60.78
69.44
55.56
14.44
-1.67
(--++)
(++--)
53.83
55.72
43.82
65.72
41.82
71.78
57.41
64.72
69.24
60.17
15.41
4.44
Efficiency audit
Serial position: RETS
Order
1.
2.
(--++)
(++--)
3.
4.
Overall
Effectiveness audit
Serial position: PGWR
Order
i.
2.
(--++)
(++--)
3
4.
Overall
AUDITOR BELIEFREVISIONS
Subjects
T h i r t y - n i n e state p e r f o r m a n c e auditors, w h o
w e r e c u r r e n t l y taking a p e r f o r m a n c e a u d i t i n g
c o n t i n u i n g e d u c a t i o n c o u r s e at t h e state headquarters, p a r t i c i p a t e d in the study. Four subjects
failed to c o m p l e t e t h e e x p e r i m e n t , thus leaving
35 u s a b l e r e s p o n s e s . T h e s u b j e c t s w e r e e x p e r i e n c e d a u d i t o r s w h o h a d w o r k e d an average o f
five and a half y e a r s for t h e State A u d i t o r ' s
Office. T h e e x p e r i m e n t a l session t o o k a p p r o x i m a t e l y 60 minutes.
175
The CASH program was established to move recipients of public assistance into productive unsubsidized jobs, thereby
increasing their self-sufficiency and reducing the state weffare roles.
6 Half of the subjects were randomly assigned to perfnrm the efficiency audit, followed by the effectiveness audit, whereas
the remaining half of the subjects were assigned to perform the tasks in the reverse order. The order of cases performed had
no effect on the subjects' performance.
176
M a n i p u l a t i o n check
T h e following manipulation c h e c k s w e r e
c o n d u c t e d . First, w e e x a m i n e d the e n c o d i n g o f
e v i d e n c e with r e s p e c t to its direction. It was
f o u n d that, o f 280 responses ( 3 5 subjects 2
senarios x 4 cues), only o n e r e s p o n s e involved
c o d i n g errors. Also, by e x a m i n i n g the subjects'
responses o n belief revision u p o n receiving
each piece o f evidence, it was f o u n d that only
ten ( o u t o f 2 8 0 ) responses involved c o d i n g
errors (i.e. an u p w a r d / d o w n w a r d revision u p o n
receiving negative/positive evidence, respectively). Also, the errors w e r e not disproportionally distributed across different t r e a t m e n t
c o n d i t i o n s for the eight pieces o f evidence.
Based u p o n the p r e c e d i n g results, w e view the
manipulation of positive and negative e v i d e n c e
as successful and the c o d i n g errors as random.
Second, using the subjects' subjective weightings o f e v i d e n c e as the d e p e n d e n t variable, w e
e x a m i n e d the equality of the subjects' weightings of evidence. T w o separate MANOVA
p r o c e d u r e s w e r e p e r f o r m e d to e x a m i n e the
equality o f e v i d e n c e weightings a m o n g the
four types o f e v i d e n c e used in each task. The
results s h o w e d that the null hypothesis of equal
e v i d e n c e weightings c o u l d not be r e j e c t e d (for
the effectiveness audit, F = .747, p = .533; for
the efficiency audit, F = .658, p = .584). That is,
for b o t h tasks, subjects, on average, v i e w e d the
four types of e v i d e n c e as equally important.
Also p e r f o r m e d w e r e eight (2 tasks x 4 types
of evidence used in each task)separate individual
ANOVA procedures. The d e p e n d e n t variable
was, again, the subjects' subjective weightings
o f evidence. The i n d e p e n d e n t variables w e r e
serial position (earlier in the s e q u e n c e vs
later in the s e q u e n c e ) and e v i d e n c e direction
(positive vs negative). Note, the focus o f the
analysis was to d e t e r m i n e w h e t h e r the subjec-
7 The belief-adjustment model defines recency effects as D = --ti~$(x )s(x+). Therefl)re, re c e nc y effects should always
prevail, regardless of a lack of c o m p l e t e symmetry b e t w e e n s ( x ) and s(x+). Tubbs et al.s' ( 1 9 9 0 ) findings also
( e x p e r i m e n t s 3 and 4 ) indicate a lack of equality in subjective weightings b e t w e e n l~)sitive and negative evidence, which
does not affect the p r e d i c t e d recency effects. A later analysis (s e e footnote 9) also seems to suggest that this inequality of
subjective weig htings does not confi)und the recency effects.
177
Dependent variables
Registration (R)
Employment (E)
Service (S)
Training ( T )
Direction of evidence*t
Average cell means
Main effect
Positive
Negative
F
67.83
(22.31)
65.39
(20.57)
70.59
(23.32)
64.41
56.47
(17.12)
52.94
(25.73)
61.28
(14.46)
61.83
(21.54)
(17.41)
2.741
.108
2.462
. i 27
1.921
.176
0.118
.734
Dependendent variables
Placement (P)
Grant ( G )
Wage ( W )
Retention (R)
Direction of evidence*$
Cell means
Main effect
Positive
Negative
F
76.22
(17.74)
79.00
(17.96)
69.71
(19.99)
72.06
(16.63)
49.71
(14.25)
40.88
(14.15)
61.00
(21.87)
56.33
(20.67)
20.919
.000
41.783
.000
1.676
.205
4.803
.036
* The cell means for serial position from the ANOVA analyses were not presented because, in each of the four ANOVAs, this
manipulation was not significant to the dependent variable ( t .416 ~< ps <~ .868; ~: .421 <~ ps ~ .982).
Standard deviations are in parentheses.
Recency effects
R e c e n c y effects p r e d i c t e d b y t h e m o d e l w e r e
t e s t e d using a 2 x 2 x 2 ANOVA w i t h o r d e r
178
(++--)
4.
(++--)
Effectiveness audit
Serial position: PGWR
Order
l.
(--++)
2.
(++--)
4.
(++--)
ABS(S2-So)
ABS(S4-S2)
15.50
(13.461
16.0{I
(9.43)
25.25
(18.32)
11.52
(12.85)
14.44
(19.91)
8.78
(14.39)
25.00
(24.11)
7.67
(11.57)
11.88
(10.02)
19.11
11.25)
28.38
(16.11)
8.56
(11.21)
12.11
11.57)
13.00
10.35)
26.56
(18.48)
14.67
(13.45)
Effectiveness audit
F
p
Between-subjects effects
Order ( O )
Serial Position (SP)
O x SP
2.68
.32
.20
.112
.574
.656
1.59
.01
.01
.217
.933
.933
Within-subjects effects
[$2-So) vs [S~-$2] (DP)
O x DP
SP x DP
O SP DP
2.08
6.23
.15
.2{I
.160
.018
.699
.656
6.48
21.12
1.38
2.71
.016
.I1011
.250
.ll6
AUDITORBELIEFREVISIONS
confronted with m i x e d e v i d e n c e in two different
tasks ( e x p e r i m e n t s 2A a n d 2B). Similar results
w e r e also f o u n d b y T u b b et al. ( 1 9 9 0 ) u s i n g
t w o o t h e r audit tasks.
Panel A of Table 3 displays the cell m e a n s of
the relative b e l i e f c h a n g e s at t w o - c u e intervals
(i.e. [$2---So] a n d [$4--$2]) across the four treatm e n t c o n d i t i o n s in each task. T h e p a t t e r n of
cell m e a n s reveals that, in b o t h tasks a n d u n d e r
the ( - - + + )
c o n d i t i o n s , the m a g n i t u d e of
b e l i e f revisions from [$4--$2] is n o t a b l y higher
than that of belief revisions from [$2-So] order.
In contrast, n o s u c h a p a t t e r n prevails u n d e r the
(++--)
c o n d i t i o n s for either task.
Recall that the belief-adjustment model predicts
a contrast effect o n belief revision. That is, w h e n
the p r e c e d i n g a n c h o r is l o w ( h i g h ) , positive
e v i d e n c e will result in a greater ( s m a l l e r )
u p w a r d belief revision; w h e r e a s negative
e v i d e n c e will result in a smaller ( g r e a t e r )
d o w n w a r d b e l i e f revision. This implies that,
u n d e r the ( - - + + ) c o n d i t i o n , the relative size
of [S4-Sz] s h o u l d b e larger than the relative size
of [$2-So] b e c a u s e $2 ( t h e a n c h o r of [$4-$2])
is l o w e r t h a n So ( t h e a n c h o r of [$2-So]), ceteris
paribus. Likewise, u n d e r the ( + + - - )
conditions, the relative size of [$4-$2], s h o u l d b e
greater t h a n the relative size of [$2-So] because the a n c h o r of [$4-$2], $2, is higher
than the a n c h o r of [$2-So], So, ceteris paribus.
T h e p a t t e r n of cell m e a n s in p a n e l A, Table
3, n o n e t h e l e s s , reveals that the p r e c e d i n g
c o n t r a s t effect p r e d i c t i o n s prevail o n l y u n d e r
the ( - - + + )
c o n d i t i o n , b u t n o t u n d e r the
(++--)
c o n d i t i o n . To e x a m i n e this observation, the subjects' i n t e r m e d i a t e belief c h a n g e s
w e r e analysed.
179
8 Additionally, we ran a set of paired t-tests. For the efficiency/effectiveness audit, the results showed that, under the
( - - + + ) condition, the combined average cell mean of [$4 - $2] for the two serial positions was greater than that of the
[$2-So](for efficiency audit t = 3.09,p = .007; for effectiveness audit t = 4.66,p ~<.000). In contrast, under the ( + + - - )
condition and in both tasks, the results revealed no such differences (for efficiency audit t = .73,p = .477; for effectiveness
audit t = 1.47, p = .160).
BUCK K. W. PEI et aL
180
b o t h a l t e r n a t i v e serial p o s i t i o n s , s u g g e s t i n g
that the results cannot be caused by the
d i f f e r e n c e s in e v i d e n c e d u e to serial p o s i t i o n
manipulations. 9
O n e p l a u s i b l e e x p l a n a t i o n is a l a c k o f s y m metrical sensitivity towards evidence. This can
be illustrated by examining the analytical
implications of the belief-adjustment model. To
s e e h o w , t h e r e l a t i v e size o f b e l i e f r e v i s i o n s f o r
[S,I--S2] a n d [S2---So] u n d e r t h e ( + + - - )
condit i o n c a n b e w r i t t e n , r e s p e c t i v e l y , as f o l l o w s
( s e e A p p e n d i x for d e r i v a t i o n ) :
[$4-6"2] = S2 ( 2 ~ - ( / 2 )
a n d [$2-So] = ( 1 - S o ) ( 2 1 3 -
[32).
G i v e n t h e c o n t r a s t effect, o n e is e x p e c t e d t o
o b s e r v e [$4-6"2] 2> [$2-So]. T h i s n e c e s s i t a t e s
$2 2> [ 1 - St, ] a n d / o r ( 2 { z - ~t2) 2> ( 2 [3- [32 ). N o n e t h e l e s s , o u r findings s h o w that, for b o t h a u d i t
tasks, t h e c o n t r a s t e f f e c t is absent u n d e r this
c o n d i t i o n (i.e. [84--82] = [82--80]). F u r t h e r , t h e
a v e r a g e c e l l m e a n s f r o m T a b l e 1 s h o w s t h a t $2
2> [ l~S'o] ( f o r e f f i c i e n c y audit, $2 = 6 2 % a n d So
~- 4 9 % ; for e f f e c t i v e n e s s audit, $2 -- 71% a n d ~'o
-- 5 5 % ) . T h i s i m p l i e s t h a t t h e a b s e n c e o f t h e
c o n t r a s t e f f e c t m u s t b e d u e to (2~t - c t 2) <
( 2 ] 3 - ]32), o r s i m p l y 13 2> ~.
Likewise, under the ( - - + + ) condition (see
A p p e n d i x for d e r i v a t i o n ) :
[$4--,S'21 = ( 1 -S2X2~J - ]32)
a n d [S2-So] = So ( 2 ~ - ~2).
Again, g i v e n t h e c o n t r a s t effect, o n e is e x p e c t e d
to o b s e r v e [$4-$2] 2> 1S2-So]. T h i s n e c e s s i t a t e s
(1 - $2) 2> St, a n d / o r (2~3 - ]32) 2> (2~t -or2).
O u r findings s h o w t h a t t h e c o n t r a s t e f f e c t is
present, a n d t h a t t h e a v e r a g e c e l l m e a n s f r o m
T a b l e 1 s h o w ( 1 - - 8 2 ) 2> So ( f o r e f f i c i e n c y
audit, $2 = 32% a n d So -- 4 6 % ; f o r e f f e c t i v e n e s s
audit, $2 = 4 2 % a n d So = 54% ). T h i s i m p l i e s
that t h e p r e s e n c e o f t h e o b s e r v e d c o n t r a s t
e f f e c t c a n b e c a u s e d b y e i t h e r (1 - $ 2 ) > So, o r
(2]3 - ~2) > (2Gt --ct2), o r b o t h .
Jointly, the preceding data analyses and
analytical derivations imply that the subjects
in this s t u d y e x h i b i t a c o n f i r m a t o r y a t t i t u d e
t o w a r d s e v i d e n c e , at least in t h e ( + + - - )
c o n d i t i o n s . In h i n d s i g h t , w e t h i n k this p h e n o m e n o n s e e m s to r e f l e c t t h e p o s s i b i l i t y t h a t
in m o s t s o c i a l p e r f o r m a n c e audits, o b s e r v i n g
n e g a t i v e r e s u l t s is t y p i c a l r a t h e r t h a n atypical;
t h e r e f o r e , s t a t e a u d i t o r s m a y n o t " o v e r e a c t " to
s u c h results. C o n v e r s e l y , o b s e r v i n g p o s i t i v e
r e s u l t s is a t y p i c a l r a t h e r t h a n t y p i c a l ; t h e r e f o r e
s t a t e a u d i t o r s m a y i n c o r p o r a t e this " s u r p r i s e "
i n t o t h e b e l i e f r e v i s i o n . T o this e n d , a r e c e n t
U.S. d o c u m e n t o n Nonprofit Entities" Performance Measures (AAA, 1 9 8 9 ) s u g g e s t s t h a t it is
difficult to u n a m b i g u o u s l y m e a s u r e t h e i m p a c t s
o f social p r o g r a m s . A c c o r d i n g l y , t h e p e r f o r m a n c e
m e a s u r e s o f s o c i a l p r o g r a m s s h o u l d reinforce
'~To rule out the possibility that the preceding results may be confounded by the unequal subjective weightings between
positive and negative evidence (found in the effectiveness audit earlier by the manipulation check), two expost tests were
performed. For each audit task, a 2 (order) 2 (serial position) ANOVA procedure was performed, using the relative size
of [$2-So] as the dependent variable. The results showed that order was not significant (for efficiency audit F = .207, p =
.653; for effectiveness audit F = .577, p = .453), neither was serial position (for efficiency audit F = .516, p = ,478; for
effectiveness audit F = .337, p = .566) or the interaction term (for efficiency audit F = .274, p = .807; for effectiveness
audit F = .436, p = .728). The absence of significant order effects, in both tasks, suggested that the subjects' belief revisions
for the same pieces of evidence were not "affected by the direction of evidence. Accordingly, the unequal subjective
weightings of evidence flmnd earlier in the effectiveness audit did not appear to be confounding.
AUDITORBELIEFREVISIONS
DISCUSSION AND CONCLUDING REMARKS
Prior to discussing the results, a s u m m a r y of
the p r e s e n t study's findings is provided. Consistent with the belief-adjustment model's prediction, p e r f o r m a n c e auditors' belief revisions
w e r e subject to r e c e n c y effects, although the
p r e d i c t e d effects w e r e m o s t salient in the
effectiveness audit. The results also validated an
important assumption of the belief-adjustment
model. That is, for a relatively short e v i d e n c e
series, the subjects' belief revisions w e r e not
affected by the serial position of e v i d e n c e (i.e.
s(xk) is the same for all pieces o f e v i d e n c e in
the series). Further, in b o t h tasks, the subjects'
intermediate belief changes reflect that the
contrast effects prevail only under the ( - - + + )
condition, but not u n d e r the ( + + - - )
condition. This suggests that the subjects' sensitivity
towards e v i d e n c e may be higher for positive
e v i d e n c e than for negative evidence.
Finding the p r e s e n c e o f r e c e n c y effects is
consistent with the prior studies c o n d u c t e d in
the financial auditing contexts. The present
findings, nonetheless, e x t e n d the m o d e l ' s predictions to a different task e n v i r o n m e n t (i.e.
p e r f o r m a n c e auditing).
Finding the absence o f serial position effects
is important b e c a u s e prior literature suggests
that a decision maker's sensitivity may not be
constant due to an attention d e c r e m e n t effect
(cf. Hogarth & Einhorn, 1990). While the
belief-adjustment m o d e l assumes this p h e n o m e n o n is unlikely to o c c u r for a short e v i d e n c e
series, this assumption has not b e e n explicitly
tested in the literature. Consistent with the
model's assumption, no serial position effects
were found throughout the analyses. Specifically,
the subjects' subjective weightings of evidence
w e r e unaffected by the serial position manipulations in b o t h tasks. In addition, with respect to
belief revisions, the manipulations had no effect
on the magnitude o f the overall revision (i.e.
[$4-So]. This suggests that the model's predicted
r e c e n c y effects are d u e to the a n c h o r i n g and
contrast implications of the m o d e l rather than
serial position effects.
Findings on the subjects' intermediate belief
181
182
BUCK K. W. PEI et aL
explore whether
other dimensions of task
uncertainty and decision consequences (and of
d i f f e r e n t m a g n i t u d e s ) affect t h e m o d e l ' s r e c e n c y
effects prediction. In the present study, the
s c o p e o f t h e i n v e s t i g a t i o n is l i m i t e d t o o n l y
o n e a s p e c t o f e a c h v a r i a b l e (i.e. f o r t a s k
u n c e r t a i n t y , e v i d e n c e d i a g n o s t i c i t y ; for d e c i s i o n
consequences,
legal liabilities). Thus, other
a s p e c t s o f t a s k u n c e r t a i n t y (e.g. d e g r e e s o f
t a s k s t r u c t u r e , cf. A b d o l m o h a m m a d i
& Wright
1 9 8 7 ) a n d d e c i s i o n c o n s e q u e n c e s (e.g. differential r e w a r d / p e n a l t y s t r u c t u r e s ) a l s o c o u l d b e
examined. To this end, we wish to point out
that by failing to explicitly manipulate the
d e c i s i o n c o n s e q u e n c e s v a r i a b l e ( e . g . via a l t e r native penalty structures), this study's results
BIBLIOGRAPHY
Abdolmohammadi, M. & Wright, A., An Examination of the Effects of Experience and Task Complexity on
Audit Judgments, The Accounting Review (January 1987) pp. 1-14.
American Accounting Association, Committee Report on Nonprofit Entities" Performance Measures,
Measuring the Performance of Nonprofit Organizations: the State of Art. AAA, Government and
Nonprnfit Section, Sarasota, FL 34233 (1989).
Asare, S. K., The Auditors' Going Concern Opinion Decision: Interaction of Task Variable and the
Sequential Processing of Evidence, Working paper, Fisher School of Accounting, University of Florida
(April 1990).
Asare, S. K. & Messier, W., A Review of Audit Research Using "lqae Belief-Adjustment Model, Working
paper, Fisher School of Accounting, University of Florida (May 1990).
Ashton, A. H. & Ashtnn, R. H., Sequential Belief Revision in Auditing, The Accounting Review (October
1988) pp. 623--641.
Deighton, J., The Interaction of Advertising and Evidence, Journal o f Consumer Research (December
1984) pp. 763-770.
Einhorn, H. J. & Hogarth, R. M., Ambiguity and Uncertainty in Prnbabilistic Inference, Psychol~Rical
Review (October 1985a) pp. 433~461.
Einhorn, H. J. & Hogarth, R. M., Judging Probable Cause, Psychological Bulletin (1986) pp. 3-19.
Einhorn, H. J. & Hogarth, R. M., Adaptation and Inertia in Belief Updating: The Contrast-Inertia Model,
Wurking Paper, Center for l)ecision Research, University of Chicago, October (1987).
Fischhoff, B. & Beyth-Marom, R., Hypothesis Evaluation from a Bayesian Perspective, Psychological
Revieu, (July 1983) pp. 239-260.
Gambino, A. H. & Reardon, T. J., Financial Planning and Evaluation f o r the Non-Profit Organization
(New York: National Association of Accountants, 1981 ).
Herbert, L., Auditing the Performance o f Management (Belmont: Lifetime l.earning, 1979 ).
Hoch, S.J. & Ha, Y-W., Consumer Learning: Advertising and the Ambiguity of Product Experience,Journal
o f Consumer Research ( September 1986) pp. 221-233.
Hogarth, R. M , J u d g m e n t and Choice, 2nd Edn (New York: Wiley 1987).
Hogarth, R. M. & Einhorn, H. J., Order Effects in Belief Updating: The Belief-Adjustment Model, Working
paper, Ccnter fi)r Dccisinn Rcsearch, University of Chicago (May 1990).
Kahneman, D. & Tversky, A., Prospect Theory: An Analysis of Decision Under Risk, Econometrica (March
1979) pp. 2263--2291.
Klayman, J. & tla, Y-W., Confirmation l)isconfirmation, and Infi)rmation in Hypothesis Testing,
Psychological Revieu, (April 1987) pp. 211-228.
A U D I T O R BELIEF REVISIONS
183
APPENDIX
Under the ( + + - - )
condition, the relative size of belief
revisions for positive evidence, [$2-So], can be derived as
follows:
letSt
-(;2
So+~(I-So)S(xl
(A1)
(A2)
$2
So + [3(1 -- So) + [~
= so+(1
Thus,
sz-so
(1
-so)(2~-
,s',,) (213 -
So ~(;(~s'(X 1)
= $1 - ~$'ls(x2).
$2
(A6)
~)
(AS)
(S , ) -
~%)
u(,%
~So)
The relative size of belief revis ons fur positive evidence, [$4$2], can be derived ms fifllows:
132).
S~
let $5
$4
=
:
Sz - ~2,S(X:~)
S:~ -- o,~s(x~).
(A3)
(A4)
S4
(Sz
ed2)
c*($2
s~
aS2)
(A7)
(A8)
$2 + ~(l -- S2)s(x))
= ,$'~ + ~(1 -- S3)s(x4).
Thus, S4-S2
s~ + ( I - s ~ ) (2[3 -
(1 - $2)(213 - [32).
13~).