BASEMENT
hmii''
/
-r^^^^.
ALFRED
P.
WORKING PAPER
SLOAN SCHOOL OF MANAGEMENT
THE PREDICTIVE ACCURACY OF BEHAVIORAL
EXPECTATIONS: TWO EMPIRICAL STUDIES
by
Paul R. Warshaw
McGill University
and
Fred D. Davis
Sloan School of Management
MIT
Working Paper
//
1450-83
MASSACHUSETTS
INSTITUTE OF TECHNOLOGY
50 MEMORIAL DRIVE
CAMBRIDGE, MASSACHUSETTS 02139
THE PREDICTIVE ACCURACY OF BEHAVIORAL
EXPECTATIONS: TWO EMPIRICAL STUDIES
by
Paul R. Warshaw
McGill University
and
Fred D. Davis
Sloan School of Management
MIT
Working Paper
//
1450-83
General Self-understanding
Moderates the Accuracy of Self -reported
Behavioral Expectations
Paul R. Warshaw
McGill University
and
Fred D. Davis
Massachusetts Institute of Technology
Running Head: Self -understanding
Self -under standing
1
Abstract
Wilson
While Nisbett and
Ericsson
&
researchers
other
between
self-report has
been
self-understanding
self-presentation.
the ability and the willingness to accurately
somewhat
must
self-understanding.
self-understanding is
a
Namely,
from
self-understanding
domains
The
overlooked.
differentiated
be
Further,
behavioral
must
present
be
differentiated
manuscript
potentially
argues
important
self-understanding
behavioral expectations
findings are
is
warranted.
is
and
discussed
issues
those
that
of
relates
from
that
of
to
general
general
individual difference
variable which has been neglected in the literature.
between general
(e.g.,
1980) argue about the accuracy of self-reported data,
Simon,
the distinction
specific
and
(1977)
The
relationship
the accuracy of self-reported
and
provocative
preliminary
reported which suggest that further research on the topic
Self -under standing
2
Prediction of individual
social psychological
has
Ajzen
Fishbein,
(e.g.,
behavioral
concerned with
in
theorizing and research (for example, see Ajzen
Bonfield, 1975), organizational (e.g.,
and political
important
been
long
&
Related applied domains like the consumer (e.g., Ryan
Fishbein, 1980).
&
behavior
predictions
Katerberg
Horn,
Hinkle,
&
1980)
because
Hulin,
&
1979)
fields are also
accuracy
their
has
for the success of planning and policy-making
significant implications
activities.
individual-level behavioral forecasts have
Generally,
upon intention
measures
predict (e.g., Ajzen
argued,
first
Ajzen
prediction purposes
expectation
(or
behavioral
is
his/her
Recently, however, it has been
and
Hartwick
self-prediction)
his/her
expectation
is
an
habits, abilities and
individual's
performing
Warshaw
&
and
intention
situational
factors,
Hartwick
as
confounded
Hartwick,
Note
2).
moderate
&
as
Hartwick,
anticipated
in press).
have delineated conditions
expectation
is
likely
to
be
behavioral predictor, although these two
a
been
that
well
as
(in press)
often
with factors
behavior,
beliefs, attitudes, social norms, intentions,
under which an individual's behavioral
concepts have
self-reported
specified
a
changes in these determinants (Warshaw, Sheppard
superior to
Behavioral
(BE).
cognitive appraisal of volitional and non-volitional
behavioral determinants:
Warshaw, Sheppard
and
press)
(in
that the more pertinent construct for
(Note 1),
subjective probability of
based on
Sheppard
based
those whose behavior we wish to
by
Fishbein, 1980).
&
Warshaw,
by
subsequently by
reported
been
the
in
the
literature
(Sheppard,
The present research is more concerned
predictive
accuracy
of
behavioral
Self -understanding
3
expectations per
particular,
In
se.
investigation of the influence of
self-understanding on
the
individual's
an
predictive
the
of a preliminary
results
accuracy
of
level
of
general
his/her behavioral
expectation is reported.
Accuracy of Self-reports
there has been much controversy and confusion
In general,
literature regarding
Wilson (1977)
the
argue
accurately report
validity of self-reported data.
individuals
that
their
on
mental
are
processes,
conscious of the products of such processes.
the
Nisbett and
observe
to
although
they may be
Wilson
Nisbett and
or
have
on the grounds that their distinction between process and
been refuted
tautological
product is
unable
in
Nevertheless, other
(Smith
research
Miller,
&
White,
1978;
1980).
suggests that an individual's ability to
accurately self-report is a function of the type of phenomenon they are
being asked to report on.
For
one's memory
tend
performance
example,
self-reported
beliefs
about
not to be accurate (Herrman, 1982) and
process-oriented, think-aloud verbal protocols of thought processes are
accurate only under certain circumstances
However,
choice,
self -predict ion
intellectual
&
1980).
Simon,
achievement,
vocational
job performance and the like are at least as accurate as
assessment methods
Osberg,
of
(Ericsson
1981).
with
Hence,
organizes various
which
research
self-related
they
is
have
other
been compared (Schrauger
needed to develop
a
taxonomy
£i
that
phenomena with respect to individuals'
ability to accurately self-report on them.
Research should be
shifted
Self -under standing
4
to the issue of when,
not whether,
self-reported data are accurate.
relationship
Research on the attitude-behavior
steps toward
addressing
behavior can be accurately
self-reports
issue
the
predicted
often a
&
performing
toward
behavioral expectation since attitude
Attitude-type
understanding
our
to
major
taken
(as opposed to whether)
attitudes.
from
pertinent
especially
are
when
of
has
behavior
of
is
major determinant of behavioral expectation (Warshaw, Sheppard
Hartwick,
in press).
In
fact, prior to Fishbein and
on behavioral intention,
1975) work
concept
extensive review
was
brought
capability
The predictive
by Wicker's
question
into
attitude-behavior
of
(e.g.,
researchers viewed attitude as the
best self-predictor of future behavior.
the attitude
Ajzen's
research,
which
of
(1969)
found
the
relationship between verbal expression of attitude and subsequent overt
behavior to be weak as generally reported in the literature.
was representative
concept as
a
of
an
era
of
disenchantment
predictor of behavior (e.g.,
with
Deutscher,
This work
the attitude
Ehrlich,
1966;
1969).
Reversing this trend, recent work has
the utility
of
the
generated
optimism
attitude concept (e.g., Kelman, 1974).
toward an attitude object have been shown to be
toward
Attitudes
predictors
better
of
multiple-act behavioral criteria with respect to the object than of any
particular act, while more specific attitudes toward the act tend to be
better predictors
of
single act behavioral criteria than are attitudes
toward the object (Fishbein
Based upon
&
Ajzen, 1974;
Fishbein
&
Ajzen
1975).
an extensive review of 109 attitude-behavior studies, Ajzen
Self -under standing
5
and Fishbein
(1977) argue that attitude predicts behavior
measures
when the
much
better
attitude and behavior correspond in specificity
of
with respect to target, action, context and
Including
time.
a
term
reflecting the social influence of significant referents has been shown
power
predictive
to increase
well
as
(Ajzen
predictive accuracy of attitudes declines
attitude
measurement of
Kelley
1978;
&
factors impede
Warner
&
Mirer,
Schwartz,
1974;
behavior
the
DeFleur, 1969).
elapsed
when
time
when
and
1978),
(Davidson
is
between
Jaccard,
Romer, 1981;
predictor
useful
a
&
The
situational
Jaccard, 1978;
&
attitude
Thus,
Fishbein, 1973).
increases (Davidson
behavior
and
L
of
behavior under appropriate circumstances.
Self-presentation Versus Self-understanding
Research on conditions affecting
data has
increasingly
on the role of
focused
For our purposes, we distinguish two
difference variables:
related to self -understanding
the
subject's
willingness
accurately
consider
self-report.
the
research design,
different causes.
individual differences.
individual
of
self-presentation and those
accurately
relate
Although
distinction
self -understanding phenomena
self-reported
relate
Self -presentation phenomena
.
to
self -understanding phenomena
to
of
categories
broad
related
those
accuracy
the
to
the
blurred
between
to
be
crucial
self-report,
subject's
in
the
whereas
ability
literature,
self-presentation
from
the
clarify
to
we
and
standpoint of
insofar as the error related to each arises from
To
to
very
the distinction, we will first discuss
Self -under standing
6
established
self-presentation phenomena, which have been well
Distinguishing
literature.
self-understanding,
will
we
self-understanding is
domain-specific
between
argue
then
the
in
and
general
domain-specific
that
established in the literature, whereas
somewhat
general self-understanding is a relatively neglected concept.
Self- presentation
Self-presentation phenomena
(e.g., Goffman,
have
Schlenker,
1959;
Edwards (1957), Crowne
Marlowe
and
Building
1980).
developed
(1979)
scale
a
self-monitoring individuals
appropriately in
given
a
are
social
Low
concerned
more
that
the
self-monitors
Snyder
Tanke, 1976;
&
behaving
about
tend
As
based
behavior
guide
to
would
attitude-behavior relation
is
be
been
found
self-reports (Turner
modeling framework
components of
to
£
to
attitude.
be
attenuated for
k
Zanna, Olson & Fazio, 1980).
A
£
White,
Thus,
with
associated
Peterson, 1977).
separate
the
their
expected,
1982;
related concept, public self-consciousness (Fenigstein, Scheier
1975) has
upon
Snyder
those high in self -monitor ing (Ajzen, Timko
Swann, 1976;
High
than low self-monitoring
situation
behavior based more on their internal states.
research shows
More recently,
self-monitoring.
of
individuals, and tend to control and guide their
situational contingencies.
individual
an
difference scale of social desireability response bias.
Snyder
the work of
on
developed
(1964)
extensively
with
dealt
been
and
Buss,
invalid
more
Romer (1981) employs
true
multiple
giving
&
a
causal
self-presentational
research traditions provide
Self -under standing
7
evidence that individuals high in concern for self -presentation tend to
be less willing to accurately report
their attitudes.
Self- understanding
Domain-specific Self -under standing
(as opposed to the willingness)
the ability
We draw
self-understanding
to self-presentation,
In contrast
the
distinction
Domain-specific
self-understanding.
Markus'
the nature of domain-specific
are "cognitive
self-understanding
experience, that organize and
refers
to
to
specific
a
self-schemata theory investigates
(1977)
self-understanding,
about
generalizations
general
and
internal knowledge of behavioral determinants relative
behavioral domain.
guide
the
the
where
self,
self-schemata
derived
processing
of
from
past
self-related
information contained in the individual's social experiences" (p.
Markus
(1977)
provides
a
provides
basis for
evidence
possession
that
individual's
an
"confident
behavior on schema-related dimensions" (p.
Two concepts from
(Sherif
&
&
Hovland,
attitude
theory,
of
64).
self -schemata
self-prediction
of
63).
the
latitude
of
rejection
Hovland, 1961) and affective-cognitive consistency (Rosenberg
1960),
recently
have
construe to be related to
and Zanna
to
to accurately self-report.
domain-specific
between
relates
(1977)
use
been used to explore issues that we
domain-specific
the
attitude-behavior consistency
latitude
is
of
self -understanding
.
Fazio
rejection notion to show that
significantly moderated by the
amount
Self -under standing
8
of direct
degree
experience,
of certainty with which the attitude
held, and how well-defined the attitude is, all relative to
Norman
behavioral domain.
a
poorly-defined
domain) and achieve results
behavioral
specific
which are consistent with that operationalization
Chaiken and Baldwin
.
(1981)
further suggest that affective-cognitive consistency
as
useful
a
measure
schematicity
of
specific
a
and Chaiken and Baldwin (1981) use
(1975)
affective-cognitive consistency to measure well- versus
attitudes (within
is
the
in
context
may
of
serve
Markus'
self -schemata theory.
General Self -understanding
From these
domain-specific
various
research
self -understanding
self-report accuracy.
self-understanding, the
differences in general
self-reported behavioral
Given
the
question
traditions,
emerges
as
self-understanding
expectations?
moderator
a
of
of
of
domain-specific
arises:
would individual
importance
naturally
concept
the
moderate
accuracy
the
of
In contrast to domain-specific
self-understanding, general self-understanding refers to one's internal
knowledge of his/her behavioral determinants in general,
not
relative
to any specific behavioral domain.
The general self-understanding concept
from
related
concepts
in
the
should
literature.
self-awareness theory posits that attention
at
be
differentiated
Wicklund's
any
given
(1975)
moment
is
Self -understanding
9
directed either
toward the self or toward external events.
state is also refered to as the state of
aroused,
be
can
himself /herself in
voice.
The
example
for
having
by
mirror or listen to
a
accuracy
of
attention,
individual
an
recording
a
and
view
his/her
of
own
self-reports of judgements of past and future
events (Pryor, Gibbons, Wicklund,
feelings (Gibbons,
self-focused
The former
Fazio
Carver, Scheier
&
Hood,
&
Hormuth, 1979)
is
present
and
1977)
increased under
conditions of heightened self-focused attention.
Fenigstein, Scheier and Buss (1975) developed
and private
previously,
the state
in
and
Turner
of
public
of
self-focused
Peterson
the
accuracy
self-reports.
of
self -consciousness has
been
associated
self-reports (Scheier,
Buss
&
self -consciousness may be
an
Buss,
mechanism
by
be
accurate
Private
Turner, 1978).
1978;
important
to
high private
predictively
with
public
phenomena)
In addition,
in
attention.
found
(1977)
self-consciousness (an example of self-presentation
related to
scale
self-consciousness that assesses individual differences
the proportion of time spent
As indicated
a
general
which
self -understanding is developed.
Self-perception (Bem, 1972) is
change which
says
that
a
theory of attitude formation
individuals
infer
their attitudes and other
internal states from observation of their own overt
contexts
in
which
the
behavior
ambiguous or uninterpretable"
significant
role
self-understanding.
in
the
and
behavior
and
the
occurs when "internal cues are weak,
(p.
2).
process
For a long time,
by
Self-perception
which
may
individuals
self-perception was viewed
play
a
gain
as
a
Self -under standing
10
with
competed
thoery that
Festinger's
theory for explaining attitude
attempted
Cooper (1977)
to
change
cognitive
(1957)
phenomena.
resolve
dissonance
Fazio,
Zanna
and
this controversy by arguing that
dissonance and self-perception are actually complementary theories, and
by delineating
appropriate
their
domains
self-awareness, self-consciousness
application.
of
self-perception
and
from self-understanding, they are clearly related.
While
distinct
are
Future research
on
self-understanding should take these bodies of theory and research into
account
Although general self-understanding has
literature, the
construct
has
with respect to
for example,
expectation (BE)
measures.
overlooked
been
implications for self-report accuracy;
predictive
the
accuracy
Under domain-specific self -understanding,
domain
independent
to
self-understanding
different domain (Markus, 1977).
as
domain-specific
If
with
respect
then
would
expect
a
that
self-understanding
correlation across
variety
individuals
would
a
of
evidence
behavioral
the
highest
evidence
a bility
between
these
extremes.
well
exhibit
A priori,
we
general
in
BE/behavior
mean
low
in
general
the lowest mean BE/behavior score.
The value for individuals who are medium in general
should lie
a
behavioral
their
high
are
who
in
as
should
domains.
sample of behaviors, while those
self-understanding should
behavior
people
systematic differences in the predictive accuracy of
largely
is
people differ in general
self-understanding,
expectations across
behavioral
of
self-understanding with respect to behavior in one
of
the
in
Namely,
self-understanding
we expect that overall
to accurately predict one's pattern of behaviors
is
positively
Self -under standing
11
related to the individual's general self-understanding level.
The study reported below is a
hypothesized relationships
preliminary
between
examination
general self-understanding and the
accuracy of BE forecasts for eleven behaviors.
maximally
to
so as
dissociate
these
of
The study was
designed
the measurement of self-understanding
from the self-reported behavioral expectation measurement, as an effort
to rule
possibility
the
out
self-understanding measure
mind.
subjects
that
with
relevant
the
responded
to
behavioral
the
domains in
Although any true assessment of general self understanding would
require that
a
multi-item, validated scale be
developed
and
applied,
the present study uses only a crude self-report measure, reflecting our
primary purpose
which is not to substantively research the topic
here,
but rather to point out that the field
important individual
pilot results
which
difference
suggest
has
overlooked
a
potentially
variable and to show some provocative
the
value
of
researching
general
self- understanding.
Method
Subjects
Subjects were 107 undergraduate student volunteers
east coast
university.
from
a
large
The sample contained 57 males and 50 females.
Prospects were recruited from four small coed dormatories
by
aides who knocked randomly on doors of students' private rooms.
all undergraduates who were approached agreed to participate.
research
Nearly
Self- understanding
12
Procedure
Each subject was interviewed in his or her private dormatory
separate
on three
likelihood
procedures
Several
each sitting.
a
connect
instrument (first interview) with the
separate questionnaire at
employed
were
would
subjects
that
completing
occasions,
minimize
to
subsequent
self-prediction
questions
pre-testing some attitude questions for
Included therein
measure:
"I
myself at
was
the
university
research
0123456789
don't
10
marketing
To further dissociate the two parts of this study, one
class project.
month expired between the first
questionnaire
and
subjects
had
interviews.
second
indicate
second
The
9-point
on
probable/improbable scales their behavioral expectations of
performing
eleven separate behaviors during the upcoming five days (e.g.,
drinking alcohol, skipping class).
those
understand
Research aides who administered the second and third
instruments claimed to be students collecting data for their
each of
team.
following simple general self-understanding
know myself well
all."
interview
Those who
subjects that they were
told
a
and
separate
First,
groups of interviewers were used for each part of the study.
personality
the
self-understanding
the
behavioral elicitations (second and third interviews).
administered the
room
acts
most
frequently
These
mentioned
particular
during
prior
behaviors
were
focus
group
discussions among 25 non-sample member students regarding non-sex typed
behaviors that students commonly perform.
one week
after
the
second,
subjects
During the third
checked
performed each behavior during the designated
once, once,
or
not
at
all
(a
'don't
off
time
remember'
interview,
whether they had
period
more
than
category was also
Self- understanding
13
provided)
.
In addition
questionning
separating
to
from
behaviorally-or iented
the
encouraged
responses were further
subjects
illusion that
were
providing
symbol
which
honest
,
created
anonymous responses.
his/her
write
questionnaire.
their
on
self-understanding
quest ionning
procedures
by
interview, the subject was told not to
identifying
general
the
When
name
the
At each
or
other
completed,
personality instrument was placed by the subject in a sealed
the
envelope,
while the expectation/behavior questionnaires were each collected via
procedure.
ballot box
However,
leaving the subject's private
inscribed
subject's
the
questionnaire.
room
for
analysis
data
occasion,
immediately upon
interviewer
surreptitiously
his/her
just-completed
each
the
room,
number
Hence, each subject's
matched later
on
questionnaires
three
could
a
personality and
self-prediction
tasks;
they
were
their
questionnaires.
Hence,
the
dropped from the
No subject reported an awareness that their identity was
matched to
be
During post-experimental
purposes.
few subjects postulated some connection between
debriefing, only
study.
on
a
being
we felt reasonably assured
that untainted data were gathered.
Results and Discussion
Twenty-four subjects were excluded from the data analysis.
were deleted
because they perceived that the first and second parts of
the study were connected.
were not
home
Three
The remaining 21 were excluded because
they
for either the second or third of the three interviews.
Self- understanding
14
Since the not-at-homes contained
members
each
of
rough proportion to its representation in the sample at large
group in
eight high, eight medium and five low),
(i.e.,
self-understanding
denigrated by their absence.
In total,
the
results
were
not
83 subjects provided analyzable
data.
Based upon
responses
their
to
general
the
self-understanding
question, subjects were divided into high (n=30), medium (n=31) and low
groupings.
(n=22)
between behavioral
calculated
for
each
For
group,
expectation
each
summarized in Table
of
(BE)
eleven
the
findings
behavior
and
behaviors.
test
measures
were
Results
are
1.
Insert Table
The
Spearman rank order correlations
supported
1
about here
the
hypothesis
that
general
self-understanding affects one's ability to self-predict and that those
who
are
low
self -predictors.
.34
(low),
Fisher
general
z
.60
in
general
self-understanding
are
the
Namely, the mean £s across all eleven behaviors
(medium)
and
.53
(high).
transformations of the rs showed
self-understanding
[F(2) = 6.16,
a
worst
were
An analysis of variance on
significant main effect for
2"^
.01].
Further,
as
Self -under standing
15
hypothesized
versus
and low
medium
Hence,
significant.
both the low versus high (t(20)= 2.58, 2< .02)
priori,
a
(t^(20)=
group
low
the
3.88,
2<
.001)
differences
was less able, across all eleven
behaviors, to predict their performance than were both the
self-understanding
high general
subjects.
medium
high
the
inadequacies
in
our
group.
rough
validated, multi-item
scale
high
result
found
r
a
(n.s.)
Perhaps this unexpected result stems from
general
would
self-understanding
certainly
give
measure.
more
a
estimate of general self-understanding and perhaps reverse
versus
and
However, contrary to our
priori expectations, the medium group evidenced a higher mean
than did
were
here.
Since
nobody
self-understanding, those who rate themselves
the
reliable
medium
complete
has
extremely
A
high
the
on
direct self-report scale are perhaps overstating their case, especially
compared with
values.
lower
those individuals who, more realistically, give slightly
problems
Related
self-understanding are:
1)
with
asking
someone
their
people who have little self-understanding
may be precisely those individuals who are unlikely to believe this
true of
themselves
(one more misunderstanding on their part);
and
is
2)
self-presentation phenomena may introduce bias.
While not conclusive, these results strongly suggest
research on general self-understanding is warranted.
that
future
Self -under standing
16
Reference Notes
1.
Ajzen,
From
I.
behavior.
Chapter
Action-control:
Springer,
2.
Sheppard, B.
a
more
intentions
in
to
From
to
action:
appear in J.
cognition
A
Kuhl
to
&
theory
J.
behavior
of
planned
Beckmann (Eds.)f
.
New
York:
preparation.
H., Warshaw,
general
for publication,
P.
R.
t
Hartwick,
theory of reasoned action
1983.
.
J.
On the need
for
Manuscript submitted
Self -under standing
17
References
Ajzen,
I.
Fishbein,
£<
theoretical analysis
and
Psychological Bulletin
Ajzen,
I.
&
Fishbein, M.
predictors of
I.
&
,
research.
22,
as
Journal of Personality
behaviors.
1973,
variables
41-57.
Understanding attitudes and predicting
Fishbein, M.
Englewood Cliffs,
.
empirical
of
A
1977, 8±, 888-918.
,
specific
social behavior
review
relations:
Attitudes and normative
and Social Psychology
Ajzen,
Attitude-behavior
M.
N.
J.:
Prentice Hall,
1980.
Ajzen, I., Timko,
C.
White,
&
attitude-behavior relation.
Social Psychology
Bern,
D.
J.
Chaiken, S.
and the
Journal
1982, £2,
of
and
the
Personality
and
426-435.
Self -percept ion theory.
Advances in
York:
,
Self -monitor ing
J.
In
L.
Berkowitz
experimental social psychology (Vol.
(Ed.),
6).
New
Academic Press, 1972.
&
Baldwin, M.
effect
of
W.
Affective-cognitive
salient
self-perception of attitudes.
Social Psychology
,
1981, 41,
consistency
behavioral information on the
Journal of
1-12.
Personality
and
Self -under standing
18
Crowne, D.
P.
Marlowe, D.
&
Evaluative Dependence
Davidson, A.
R.
&
1979,
Deutscher,
Journal
J.
Wiley,
Personality
of
of
and
in
1964.
Variables that moderate
Results
relation:
Studies
the
longitudinal
a
Social
Psychology
,
1364-1376.
37,
Words
I.
New York:
.
Jaccard, J.
attitude-behavior
survey.
The Approval Motive;
deeds.
and
Problems
Social
,
1966,
13
,
235-254.
Edwards, A.
The social desireability variable in personality
L.
assessment and research
Ehrlich, H.
New York:
Attitudes, behavior and the
American Sociologist
Ericsson, K.
A.
&
H.
&
Zanna, M.
the strength
Journal
398-408.
of
,
1969,
Simon,
Psychological Review
Fazio, R.
.
of
,
H.
4,
A.
Dryden, 1957.
intervening
variables.
35-41.
Verbal
reports
as
data.
1980, 82, 215-251.
P.
the
Experimental
Attitudinal qualities relating to
attitude-behavior
Social
Psychology
relationship.
,
1978,
14
,
Self -under standing
19
Fazio, R.
Zanna,
H.,
M.
self -percept ion:
proper domain
application.
1977,
,
Fenigstein, A., Scheier, M.
private
Cooper,
&
Journal
theory's
each
of
and
Experimental
of
464-479.
13,
F.
Dissonance
J.
view
integrative
An
of
Social Psychology
P.
&
Buss,
self-consciousness:
A.
Assessment
Journal of Consulting and Clinical
Public
H.
theory.
and
Psychology
,
dissonance
.
and
1975,
43
,
522-527.
Festinger, L.
A
theory
cognitive
of
Stanford:
Stanford University Press, 1957.
Fishbein, M.
of
Ajzen,
&
I.
Attitudes toward objects as predictors
single and multiple behavioral criteria.
Review
,
Fishbein, M.
behavior
59-74.
1974, 81,
&
Ajzen,
An
Belief,
I.
introduction
,
Ajzen,
understanding voting
external variables.
Hinkle,
in
American
Ajzen
In I.
Understanding attitudes
Englewood Cliffs, N.
theory
to
&
I.
attitude,
Addison-Wesley
Reading, Massachusetts:
Fishbein, M.
Psychological
J.:
and
,
R.
intention
M.
predicting
research
and
.
1975.
Predicting
elections:
&
and
Effects of
Fishbein
social
Prentice-Hall, 1980.
and
(Eds.),
behavior
.
Self- understanding
20
The presentation of self in everyday
Goffman, E.
Doubleday
City, NY:
Herrman, D.
&
Garden
.
Company, 1959.
Know thy memory:
J.
life
The use of
questionnaires
Psychological Bulletin
assess and study memory.
to
1982, 92
,
,
434-452.
Horn,
P.
W.
Katerberg,
,
examination
turnover.
R.
three
of
Journal
of
&
Hulin,
approaches
Applied
C.
to
Comparative
L.
prediction
the
Psychology
1979,
,
of
64
,
280-290.
Kelley, S.
Mirer, P.
&
W.
The
simple
American Political Science Review
,
act
of
voting.
1974,
68,
572-591.
Attitudes are alive and well and
Kelman, H.
employed
gainfully
American Psychologist
sphere of action.
in the
The
1974, 29
,
,
310-324.
Self -schemata and processing
Markus, H.
self.
35,
Journal
of
information
about
Personality and Social Psychology
,
the
1977,
63-78.
Nisbett, R.
E.
L
Wilson, T.
D.
Telling more than we can know:
Verbal reports on mental processes.
1977, 84, 231-279.
Psychological
Review
,
Self -under standing
21
Norman,
Affective-cognitive
R.
conformity and behavior.
Psychology
Pryor, J.
,
1975,
X., Wicklund, R.
Journal of Personality
A.,
Fazio, R.H.
,
1981,
41,
Rosenberg, M.
Journal
of
J.
Personality and Social Psychology
&
Hovland, C.
Cognitive,
I.
of attitudes.
In M.
(Eds.), Attitude organization
al.
Haven, Conn.:
&
of
,
562-576.
behavioral components
J.
&
513-527.
1977, 45,
A person-situation causal analysis of self-reports
attitudes.
Ryan, M.
Journal of Personality and Social
Self-focused attention and self-report validity.
Hood, R.
et.
attitudes,
83-91.
32,
B., Gibbons, F.
Romer, D.
consistency,
Yale University Press,
Bonfield, E.
consumer behavior.
H.
affective
Rosenberg,
J.
and
and
change
.
New
1960.
The fishbein extended model and
Journal of Consumer Research
,
1975,
2,
D.
M.
118-136.
Scheier,
M.
F.,
Buss,
Self-consciousness,
aggression.
133-140.
A.
self
H.
report
&
of
Buss,
aggressiveness
Journal of Research in Personality
,
1978,
and
12,
Self -under standing
22
Schlenker, B.
Impression
R.
Schrauger, J.
S.
&
Osberg, T.
Schwartz,
Psychological Bulletin
instability
Temporal
S.
as
attitude-behavior relationship.
Social Psychology
Sherif, M.
1978,
Hovland, C.
&
and Contrast
R.
cognitive
36,
322-351.
90.'
moderator
a
Journal of Personality and
Judgement;
Assim ilation
Miller,
F.
processes:
A
&
,
D.
reply
1978, 85,
Self-monitoring processes.
Limits
to
on
Nisbett
perception
.
Academic Press, 1979.
of
Wilson.
and
355-362.
In L.
Berkowitz
Advances in experimental social psychology (Vol.
York:
the
of
715-724.
Social
I.
1981,
.
Effects in Communication and Attitude Change
Psychological Review
Snyder, M.
of
Yale University Press, 1961.
New Haven:
Smith, E.
,
accuracy
relative
The
M.
judgements by others in psychological
self-predictions and
assessment.
Belmont,
.
1980.
Wadsworth,
California:
Relations
Interpersonal
and
Social Identity,
Self-concept,
The
Management:
(Ed.),
12).
New
Self- understanding
23
Snyder, M.
Tanke, E.
&
are more
consistent
1976,
501-517.
Snyder, M.
44,
Swann, W.
&
politics of
than others.
actions
When
,
1976,
reflect
predictive validity
The
Journal of Personality
and
Journal of Research in
measures.
,
1034-1042.
34,
typical
of
attitudes:
and
self-consciousness,
Consistency,
G.
people
Some
Journal of Personality
impression management.
and Social Psychology
Turner, R.
Behavior and attitude:
D.
personality
maximal
Personality
the
,
1978,
12,
117-132.
Turner,
R.
G.
Peterson,
&
self-consciousness and
G.
&
concept:
Attitude
Defleur, M.L.
Social
between
American Sociological Review
Warshaw, P.
R.
,
Sheppard, B.
and self-prediction
appear in
Communication
of
Bagozzi
R.
.
and
constraint
intervening variables
,
H.
1977,
as
private
and
emotional expressivity.
Consulting and Clinical Psychology
Warner, L.
Public
M.
Journal of
45,
490-491.
an
interactional
distance
social
attitudes
as
action.
and
1969, 3£, 153-169.
,
&
goals
(Ed.),
Greenwich, CT:
The
Hartwick, J.
and
Chapter to
behavior.
Advances
JAI Press,
in
in
intention
Marketing
press.
Self -under standing
24
White, P.
Limitations on verbal reports of internal
refutation of Nisbett and Wilson and of
Review
Wicker, A.
,
1980,
Psychological
Bern.
relationship
The
verbal and overt responses to attitude objects.
Social Issues
Wicklund,
R.
A.
,
8).
Zanna, M.
New York:
P.,
1969,
in
self-awareness.
Experimental
An
M.
&
In
L.
Berkowitz
Psychology (Vol.
Social
Fazio, R.
individual
Journal of Personality and
432-440.
Journal of
Academic Press, 1975.
Olson, J.
consistency:
of
41-78.
25,
Objective
(Ed.), Advances
A
105-112.
87,
Attitudes versus actions:
W.
events:
H.
Attitude-behavior
perspective.
difference
Social
Psychology
,
1980,
38
,
Self -under standing
25
Footnotes
Correspondence concerning the article should be sent
Warshaw, Faculty
of
Management,
McGill
Street West, Montreal, Quebec, H3A 1G5.
University,
to
1001
Paul
R.
Sherbrook
Self -under standing
26
Table
1
Spearman Rank Order Correlations Between
Behavioral Expectations (BE) and Behaviors for
High, Medium and Low General Self-understanding Groups
Self-reported General Self-understanding
Criterion Behaviors
The Accuracy of Behavioral
Intention Versus Behavioral Expectation
for Predicting Behavioral Goals
Paul
R.
Warshaw
McGill University
and
Fred
D.
Davis
Massachusetts Institute of Technology
Running Head:
Predicting Behavioral Goals
Predicting Behavioral Goals
1
Abstract
Researchers have largely overlooked the distinction between an individual's behavioral intention and his/her behavioral expectation as
predictors of his/her own behavior.
Moreover, the distinction between
purely volitional behaviors and behavioral goals
,
the latter of which
may be impeded by such non-volitional factors as lack of ability, lack
of opportunity,
habit and environmental impediments, has also been
blurred in the literature.
based upon
a
Behavioral expectation is theorized to be
cognitive appraisal of one's behavioral intention plus
all other behavioral determinants that one is aware of.
Therefore, the
present research argues and gives evidence that while behavioral expectation and behavioral intention may have similar predictive accuracy
for volitional behaviors, behavioral expectation is adequate while
behavioral intention may be inadequate for prediction of the accomplishment of behavioral goals.
Predicting Behavioral Goals
2
Predicting individual behavior has long interested social psycholoThe conceptual underpinning for
gists and other applied researchers.
much of this work is Fishbein and Ajzen's intention model
Fishbein, 1980; Fishbein
&
(Ajzen
&
Although the field has
Ajzen, 1975).
generally accepted their proposal that the best predictor of an individual's future behavior is his/her present behavioral intention to perform
that behavior (BI)
,
Warshaw, Sheppard
Hartwick (in press) have
&
argued that behavioral intention and behavioral expectations (BE) have
been confounded in the literature.
BI is measured by asking individuals
to indicate whether or not they intend to perform a behavior
on some subjective probability scale)
,
(usually
while BE is measured by asking
individuals whether or not they will perform a behavior (also on some
subjective probability scale).
Thus, while BI is a statement of con-
scious intention, BE is a self -predict ion of one's own behavior.
Although BE has been used inadvertantly and inappropriately to
operationalize intention (e.g., Bentler
et al.
(in press)
&
Speckart, 1979), Warshaw
theorize that BE is predictively superior when
criterion behavior is habitual or mindless
(Langer, 1978)
,
b)
a)
the
the indi-
vidual is able to foresee changes in his/her current intention during
or prior to the time in question, or c)
the individual is able to
foresee impediments to performing the behavior, such as lack of ability,
lack of opportunity, habits or environmental impediments.
Thus BE is
based on an individual's cognitive appraisal of all volitional and non-
volitional behavioral determinants of which he/she is aware: beliefs,
attitudes, social norms, present intentions, habits, abilities and
situational factors, as well as anticipated changes in these determin-
Predicting Behavioral Goals
3
(Warshaw, et al.,
ants
in press).
As formulated by Fishbein and Ajzen
is theorized to be the immediate
volitional behavior
.
(1975)
,
behavioral intention
psychological determinant of purely
Thus it does not apply when the probability of
performing a behavior given that it is not consciously attempted is
greater than 0.0
performing
than 1.0
a
(the case of habit),
or when the probability of
behavior given that it was consciously attempted is less
(the case of behavioral impediments).
is routinely,
The intention model
although inappropriately, applied in contexts where habits
or behavioral impediments are operative
(Warshaw, et al., in press).
We shall refer to behaviors where lack of ability, lack of opportunity,
habits or environmental impediments prevent the performance of the
behavior given that it is attempted as behavioral goals
goals are a special case of goals
.
Thus behavioral
(Warshaw, et al., in press),
including non-observable outcomes as well.
the latter
Behavioral goals have not
been distinguished from purely volitional behaviors in the literature
that attempts to predict behavior from intentions.
For purely voli-
tional behaviors, BE should be based largely on an individuals intention
(BI),
and BE and BI should both be predictive of behavior
(B)
.
However,
for behavioral goals, BE should reflect considerations of behavioral
impediments that BI does not include.
for
purely
volitional
behaviors
expectation (BE) and behavior
and behavior
(B)
(B)
,
Thus it is hypothesized that
the correlations between behavioral
and between behavioral intention (BI)
will all be significant; whereas for behavioral goals
the correlations between behavioral expectation
(BE)
and behavior
,
(B)
will be significant while the correlations between behavioral intention
Predicting Behavioral Goals
4
(BI)
and behavior
(B)
will be non-significant.
Method
Subjects
Subjects were 39 male
and female undergraduate students who were
enrolled in two introductory marketing courses at a large east coast
university during the summer of 1982.
The students voluntarily
participated in the study during class time.
Procedure
Subjects completed two questionnaires.
The first, administered
on a Monday and Tuesday, pertained to student activities the following
This questionnaire asked BI and BE questions regarding four
weekend.
behaviors.
Two of these behaviors were purely volitional behaviors
while two were behavioral goals.
The method of selecting these four
behaviors involved having thirty-seven non-sample-member students
rate each of several acts as either a behavior or a goal for them
during
the
upcoming
week.
Behaviors were defined for them as
acts which "nothing much would prevent you from carrying out the act
during
the-
upcoming
week,
assuming you intended to perform it,"
while goals were defined as those acts for which "lack of ability,
lack of opportunity, habits or environmental impediments might prevent
you from carrying out an intent to perform the act during the upcoming
week.
"
Predicting Behavioral Goals
5
Subjects rated various acts as behaviors or goals for them on five
point semantic differential scales whose endpoints were "definitely a
behavior for me"
point 5).
(scale point 1)
(scale
Those two acts most strongly rated as behaviors (i.e., eating
ice cream
(x =
and going swimming
1.6)
strongly rated as goals
= 3.3)
and "definitely a goal for me"
(i.e.,
(x =
2.2))
and those two most
talking with a goodlooking stranger
and finishing all my unfinished school work
selected as the experimental behaviors.
(x =
(x
4.2)) were
Namely, eating ice cream and
going swimming were the purely volitional behaviors while talking with
a
goodlooking stranger and finishing all my unfinished school work
served as examples of behavioral goals.
The scales used to measure BI and BE were 11-point
(0
to 10)
semantic differential scales with endpoints "highly likely/highly unlikely"
(the ordering of these endpoints was reversed for half the
subjects)
.
Regarding BE, subjects were asked to "please indicate the
likelihood that you will actually perform each act listed below sometime this upcoming weekend."
this upcoming weekend
follows:
I
Questions were of the form:
will [act]."
"Sometime
For BI, the instruction was as
"Please indicate what your intentions are at this very
moment regarding your performing each act listed below sometime this
upcoming weekend.
We are not asking about what you think your inten-
tions are going to be this weekend; rather, please focus only upon your
present intentions."
I
Questions were of the form:
intend to perform [act]
"At this very moment,
sometime this upcoming weekend."
When subjects had completed the questionnaire, they were instructed
to inscribe their initials on its face.
During the following Monday and
Predicting Behavioral Goals
6
Tuesday, subjects answered the second questionnaire, self-reporting
their performance or nonperformance of each act on questions of the
form:
"This past weekend, did you
[act]?
Yes
No."
Once again, each
subject's initials were requested so that their two questionnaires could
be matched.
Results and Discussion
The results of this study strongly support our hypotheses.
shown in Table
1,
the correlations between BE and B for the two purely
volitional behaviors
goals
(r
However,
=
As
.307 and
(r
=
.362)
.441 and .415)
as well as for the behavioral
are all significant at the p
<
.05 level.
the correlations between BI and B were significant only for
the two purely volitional behaviors
(r =
each case) but not for the behavior goals
.476 and
(r
=
.340; p
<
.05 in
.091 and .171; n.s.
in
each case)
Insert Table
1
about here.
These results suggest that researchers should verify that the
assumption of conscious volitionality is met when using intention
measures for prediction, and that behavioral expectation may be
alternative in the cases where this assumption is not met.
a
useful
Predicting Behavioral Goals
7
In fact,
hiMclh
since several researchers have argued recently that much
behavior has nonintentional determinants (e.g., Langer, 1978;
Zajonc,
1980;
Zajonc
&
Markus, 1982),
behavioral expectation may
be superior to intent for predicting the performance of many "behaviors"
of interest to social scientists.
Predicting Behavioral Goals
8
Ajzen,
Behavior
Understanding Attitudes and Predicting Social
Fishbein.
M.
&
I.
Englewood Cliffs, New Jersey:
.
Bentler, P. M.
G.
&
Fishbein, M.
I.
&
Models of Attitude-behavior relations.
Speckart.
Psychological Review
1979,
,
E.
J.
In J.
H.
Addison-Wesley
Harvery, W. J. Ickes
R.
P.
,
.
1975.
Rethinking the role of thought in social interaction.
in Attribution Research
Warshaw,
452-464.
8j6,
Belief, Attitude, Intention and Behavior
Ajzen.
Reading, Massachusetts:
Langer,
Prentice Hall, 1980.
,
B.
H.
(Vol.
Sheppard
Kidd (eds)
Hartwick.
Prediction of Goals and Behavior.
in Marketing ComiTiunications
F.
,
New Directions
New York: Halsted, 1978.
2).
J.
&
R.
&
In R.
The Intention and Self-
Bagozzi
(ed). Advances
Greenwich, Connecticut: JAI Press,
.
in press.
Feeling and Thinking: Preferences Need No Inferences.
Zajonc, R.
American Psychologist
Zajonc,
R.
&
H.
Markus
.
,
1980,
25.'
151-175.
Affective and Cognitive Factors in Preferences.
Journal of Consumer Research
,
1982,
9,
123-131.
Predicting Behavioral Goals
9
Footnotes
Correspondence concerning the article should be sent to Paul
R.
Warshaw, Faculty of Management, McGill University, 1001 Sherbrook Street
West, Montreal, Quebec, H3A 1G5.
Predicting Behavioral Goals
10
Table
1
Spearman Correlation Coefficients Between
Cognitive Measures (i.e., Intentions, Self-Predictions) and
Performance of Two Types of Acts
Variables Correlated
Behaviors
Purely volitional behaviors
BI/B
BE/B
:
Go swimming
476**
.441**
Eat ice cream
34C*
.415**
Behavioral goals
:
Finish all ones unfinished
schoolwork
091
307'
171
362'
Talk with a goodlooking
stranger
N = 39 for each table entry.
Note:
BI = Behavioral Intention
BE = Behavioral Expectation
B
*
p
<
.05
**
£
<
.01
=
Behavior
^353
Uii6
MIT
3
I
IP.RARIES
TDfiD DOM
MT2 TbQ
Date Due
Lib-26-67
fN3W3S'''