Accessibility of self-images following self-presentations


Material Information

Accessibility of self-images following self-presentations
Physical Description:
Trudeau, James Vincent
Publication Date:

Record Information

Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 24125413
oclc - 19681276
System ID:

Table of Contents
    Title Page
        Page i
        Page ii
    Table of Contents
        Page iii
        Page iv
        Page v
        Page vi
    Chapter 1. Introduction
        Page 1
        Page 2
        Page 3
        Page 4
        Page 5
        Page 6
        Page 7
        Page 8
        Page 9
        Page 10
        Page 11
        Page 12
        Page 13
        Page 14
        Page 15
        Page 16
        Page 17
        Page 18
        Page 19
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
        Page 29
        Page 30
        Page 31
        Page 32
        Page 33
        Page 34
        Page 35
        Page 36
        Page 37
        Page 38
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
        Page 45
        Page 46
        Page 47
        Page 48
        Page 49
        Page 50
        Page 51
        Page 52
        Page 53
        Page 54
        Page 55
        Page 56
        Page 57
    Chapter 2. Method
        Page 58
        Page 59
        Page 60
        Page 61
        Page 62
        Page 63
        Page 64
        Page 65
        Page 66
        Page 67
        Page 68
        Page 69
        Page 70
        Page 71
    Chapter 3. Results
        Page 72
        Page 73
        Page 74
        Page 75
        Page 76
        Page 77
        Page 78
        Page 79
        Page 80
        Page 81
        Page 82
        Page 83
        Page 84
        Page 85
        Page 86
        Page 87
        Page 88
        Page 89
        Page 90
        Page 91
        Page 92
        Page 93
        Page 94
        Page 95
        Page 96
    Chapter 4. Discussion
        Page 97
        Page 98
        Page 99
        Page 100
        Page 101
        Page 102
        Page 103
        Page 104
        Page 105
        Page 106
        Page 107
        Page 108
        Page 109
        Page 110
        Page 111
    Appendix A. Interview protocol
        Page 112
        Page 113
    Appendix B. Questionnaire for computer-aided personality assessment
        Page 114
    Appendix C. Questionnaire for simulated interview project
        Page 115
        Page 116
        Page 117
        Page 118
        Page 119
        Page 120
        Page 121
        Page 122
    Appendix D. Supplemental figures
        Page 123
        Page 124
        Page 125
        Page 126
        Page 127
    List of references
        Page 128
        Page 129
        Page 130
        Page 131
        Page 132
        Page 133
    Biographical sketch
        Page 134
        Page 135
        Page 136
Full Text









I would like to thank the members of my commitee for

graciously accommodating me in their busy schedules. I

would especially like to thank my chairman, Dr. Barry

Schlenker, for his continued support and guidance. Finally,

I gratefully thank those people who have provided the

emotional and logistical support that allowed me to persist

in this endeavor, most especially my family and my wife,



ACKNOWLEDGEMENTS ..................

ABSTRACT . . . . . . . . . . .


I. INTRODUCTION ........ ...................

Prologue ....... ...................
William James and the Consciousness of Self
Self-Identification Theory ... .........
Internalization Studies ..........
The Stability and Malleability of the Self
The Self and Social Cognition .......
Reaction Time for Responses to Scale
Positions . . . . . . .
Predictions . . . . . . . .

II. METHOD . . . . . .

Overview ............
Subjects and Procedure .

III. RESULTS .. . . . .

Manipulation Checks .
Dependent Measures . .
First Computer Task
Second Computer Task
Questionnaire ...



. . . . . . 58

. . . . . . 58
. . . . . . 59

. . . . . . 72

. . . . . . 72
. . . . . . 75
. . . . . . 75
. . . . . .. 89
. .... . . . . 93

. . . . . . 97

A. INTERVIEW PROTOCOL ...............

ASSESSMENT . . . . . . . .





* V








D. SUPPLEMENTAL FIGURES .............. 123

LIST OF REFERENCES ....... .................. 128

BIOGRAPHICAL SKETCH ....... .................. 134

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



James Vincent Trudeau

August, 1988

Chairman: Dr. Barry R. Schlenker
Major Department: Psychology

In using decision latency to examine the accessibility of

self-images, this study expanded upon previous studies by

incorporating decisions about the self-descriptiveness of

degrees of the target trait in addition to such decisions

for the trait at large. Further, these measures were

collected before and after an interview in which the subject

made a self-presentation reflecting a degree of the target

trait that was previously labeled self-descriptive or non-

self-descriptive. As expected, trait degrees that comprised

the borders of the latitude of acceptance showed

significantly longer response latencies than degrees that

were more clearly self-descriptive or non-self-descriptive.

This differential latency effect resembles the inverted-U

relationship between the prototypicality of an item and the

time needed to decide whether it belongs to a category, as

shown by Rosch (Journal of Experimental Psychology: General,

1975, 104, 192-233) and extended to decisions concerning the

self by Kuiper (Personality and Social Psychology Bulletin,

1981, 7, 438-443).

The interview presentation significantly attenuated the

differential latency effect. The attenuation was not due to

practice, for the effect persisted for all groups on two

traits that were not the focus of the interview and on the

target trait for a no-presentation control group. While all

types of presentations attenuated the differential latency

effect, the latency for the overall trait decision was

reduced only following presentations of the most positive

acceptable degree or of an extremely negative degree.

Neither the response toward the overall trait nor the

pattern of responses for the trait-degrees was affected by

the interview presentation. These findings support the

usefulness of examining the accessibility of self-images

along with the content of those images in studying the

stability and malleability of the self (Markus & Kunda,

Journal of Personality and Social Psychology, 1986, 35,

63-78). It is further suggested that in this line of study

it is prudent to address decisions involving degrees of a

trait rather than only the overall trait.



This paper addresses one of the basic issues of human

social life: How are our self-appraisals affected by our

behavior vis-a-vis others? Related questions concern the

implications for our understanding of the self, particularly

regarding its stability and malleability, and the sense of

identity it confers. These topics have been touched upon in

a variety of ways in the history of social psychology, but

have only recently come under full scrutiny.

Although the roots of this paper are in classic social

psychology, much will be borrowed from the relatively new

field of social cognition. In that light, the following

passage from T. B. Rogers's chapter "A Model of the Self as

an Aspect of the Human Information Processing System" is

offered as a prologue.

It is commonly held that a major feature that
distinguishes the human from other animals is our
capacity for self-awareness. The puzzlement and
wonder attending this capacity have been fodder
for religion, myth and art--and only recently, in
an historical time frame, become the subject of
intensive scientific inquiry. Our capacity for
self-awareness is most strongly reflected in our
day-to-day experience of continuity of identity.
* This continuity of identity has become a
critical construct of most self-theories. From
James (1890) and his conception of the "I" through

to Brewster Smith's recent analysis of selfhood as
encompassing "the feeling of identity over a
lifetime" we find this theme interwoven throughout
our attempts to understand ourselves. As the
research begins to unfold in the cognitive domain,
and social and personality psychologists begin to
interact with cognitive psychologists it
becomes increasingly appropriate to ask if we can
analyze these "feelings of identity" from a
cognitive perspective. Can we, in the tradition
of our colleagues studying the processes of
pattern recognition or semantic memory, begin to
unravel some of the mysteries attending selfhood?
Perhaps an understanding of the processes involved
in the interpretation of social and personal
information will provide insight into this age-old
problem of identity. (1981, p.193)

The cognitive approach has proven helpful in understanding

the self, as can be seen in the work of Rogers and others.

However, until very recently, cognitive approaches to

identity have for the most part ignored this notion of

selfhood across time, focusing instead on self-related

phenomena at a given time.

The present paper sought to build upon the framework

provided by social cognition researchers such as Rogers and

extend the approach to address changes over time, to examine

the paradox of the malleable yet stable self. First, in a

brief historical retrospective the treatment of the feeling

of identity over time in the writing of William James will

be discussed. This account will lead to a brief review of

the theoretical approach upon which the proposed study is

based: Schlenker's self-identification theory. Next, the

relevant current literature from the study of changes in

self-appraisals following self-presentational behaviors and

from the social cognitive study of the self will be

discussed. Finally, a study will be discussed that applied

the techniques borrowed from social cognition to the

assessment of changes in self-appraisals following a self-


William James and the Consciousness of Self

William James's chapter "The Consciousness of Self" in

his classic work Principles of Psychology is a common

setting-off point in discussions of modern thought on the

topic of the self. Many of the insights James provided form

the basis of much present-day theory and research on the

topic, which is currently enjoying a resurgence. This

section, while discussing these insights, will also address

the possibility that James did not fully pursue some of his

more influential ideas; later we shall see some implications

of this possibility in the work of some of James's

intellectual prodigy.

In his "Historical Perspective" preceding a collection of

chapters on The Self and Social Life (edited by Schlenker,

1985), Scheibe described the resurgence of interest in the

self, writing:

One must return to the very beginnings of American
psychology to encounter another period of strong
interest in self and identity. The pragmatist-
functionalist school, of which William James was
the foremost representative, gave central
importance to the self. In this as in much else,
James must be regarded as a pivotal figure in the
history of American psychology. (Scheibe, 1985,

Scheibe went on to trace the demise and resurrection of

scientific interest in the self, concluding that, "with the

emergence of modern cognitive psychology and with the

growing recognition of the artificiality of the separation

of psychological and philosophical questions, James is very

much in the ascendancy" (p.36). The following section will

address some areas of James's influence, in particular those

resurfacing in the recent cognitive orientation to the self.

Of James's many contributions, three inter-related points

are especially relevant to the present paper. First is

James's distinction between the pure ego (self-as-knower,

"I") and the empirical self (self-as-known, "me"); second is

his famous declaration of the multiplicity of social selves;

third is his exposition of the sense of identity provided by

the pure ego. The impact of the first idea cannot be

overstated, for subsequent acceptance of the empirical self

as a viable topic of scientific study has been crucial in

the re-emergence of self research. The second and third

points will be afforded more than the typical cursory

mention, given their importance to this paper. It will be

proposed that the multiplicity of the social selves is never

fully reconciled with the sense of identity over time; later

we shall see how social psychologists are still grappling

with these two points.

The second point of interest appears in what Scheibe

describes as the most quoted passage in the chapter, in

which James states that an individual

has as many different social selves as there are
distinct groups of persons about whose opinion he
cares. He generally shows a different side of
himself to each of these different groups.
From this there results what practically is a
division of the man into several selves; and this
may be a discordant splitting . or it may be a
perfectly harmonious division of labor as where
one tender to his children is stern to the
soldiers or prisoners under his command. (1890,

Here James is speaking quite specifically about the social

(public) aspects of the self, as opposed to the resultant

(private) self-feelings. The focus is on the various roles

we play and the reactions of others, rather than on the

effects our role-playing has upon ourselves. He does not

scrutinize the possible impact on the man of this "division

into several selves."

Later, though, James does address the "Rivalry and

Conflict of the Different Selves":

I am often confronted by the necessity of standing
by one of my empirical selves and relinquishing
the rest. Not that I would not, if I could, be
both handsome and fat and well-dressed, and a
great athlete, and make a million a year, be a
wit, a bon-vivant, and a lady-killer, as well as a
philosopher; a philanthropist, statesman, warrior,
and African explorer, as well as a "tone poet" and
saint. But the thing is simply impossible. The
millionaire's work would run counter to the
saint's; the bon-vivant and the philanthropist
would trip each other up; the philosopher and the
lady-killer could not well keep house in the same
tenement of clay. (pp. 309-310)

Here still, though, the emphasis is on the effects of laying

claim to certain of one's possible selves and foregoing the

others. It is in this context that James suggests the

equation "self-esteem = success/pretensions."

He also


posits a hierarchical arrangement of the types of empirical

selves (material, social and spiritual selves) and their

proper relation to each other. But the impact of their

rivalries and conflicts on the individual's self-concept is

not addressed.

In discussing the pure ego, the third point of interest,

James begins to address the issue of identity over time. The

most famous line in this regard occurs in the summary: "It

is a Thought at each moment different from that of the last

moment, but appropriative of the latter, together with all

that the latter called its own" (1890, p. 401). This image

of the passing thought moving through time is entwined with

the idea that thoughts which were once present are now past,

and so are now different in important ways from the present

thought. Of most interest to this paper is the consequence

that previous self-images and behaviors can only live

through the present thought. Within the chapter itself

James presents an impressive argument in support of this

contention. In brief, he draws as an analogy for the stream

of thought and temporally distant selves the herdsman and

his herd: As the herdsman rounds up only those beasts with

his mark on them, so does the thought accept distant selves

that bear its mark, which is a warm, intimate feeling. He

points out that this metaphor evades the Associationist

problem of having the thoughts adhere to themselves. As the

cattle constitute a herd only potentially, with the

potential realized through the herdsman, so do the distant

selves constitute an identity only when brought together in

the stream of thought. The thrust of this argument is that

"Passing Thought is the only Thinker which psychology

requires," which is a central focus of James's work. This

argument is not, however, directed toward the discomfort

clashing selves might instill in the self.

James is quick to disabuse the reader of the notion that

he is to place undue emphasis on the unity provided by the

passing thought:

The past and present selves compared are the same
just so far as they are the same, and no farther.
A uniform feeling of "warmth," of bodily existence
pervades them all; and this is what gives
them a generic unity, and makes them the same in
kind. But this generic unity coexists with
generic differences just as real as the unity.
And if from the one point of view they are one
self, from others they are as truly not one but
many selves. And similarly of the attribute of
continuity; it gives its own kind of unity to the
self--that of mere connectedness, or unbrokenness,
a perfectly definite phenomenal thing--but it
gives not a jot or tittle more. And this
unbrokenness in the stream of selves, like the
unbrokenness in an exhibition of "dissolving
views," in no wise implies any farther unity or
contradicts any amount of plurality in other
respects. (p.335)

Here we see James confronting the paradox of multiplicity

and unity, but it is unclear that he has offered a firm

handhold on the problem.

Of this passage, Scheibe has written,

It is possible to regard James as contradicting
himself when he argues first for a plurality of
selves and then argues with equal conviction for
the continuity and unity of self. . But the
real point about James's view of self is that he
does insist on having it both ways. (1985, p.39)

In support of the idea of "having it both ways" Scheibe

elaborates on the "dissolving views" metaphor, which he sees

as referring to predecessors of the motion picture. He

feels that James's conception of the unity and plurality of

selves is aptly illustrated in the cinematic technique of

montage, where the camera moves around the subject or

changes perspective.

The visual sequence . is normally experienced
as having unity, despite the jerkings about, the
discontinuities, the lack of realistic
presentation. The self is similar. James makes
the claim that despite our many social roles and
the masks they require, despite our combining
dreaming with planning and hoping, and loafing,
and loving, and feeling secure and alone at the
same time--despite all of these discontinuities in
the empirical constituents of of our selves, we
retain an overall conviction of unity and
continuity. Just as shifting one's attention from
the montage to the making of the montage somewhat
spoils the illusion, so the focusing of attention
on that which creates the montage "me" tends to
threaten that unity which once seemed natural.
(1985, p.40)

This analogy can apply to the literature of objective self-

awareness (Duval and Wicklund, 1972), strategic self-

presentation (Jones & Pittman, 1982), and dramaturgy

(Goffman, 1959). Each addresses the possible feelings of

discontinuity and insincerity that may arise when we become

aware of our efforts at creating certain impressions. The

analogy addresses the feeling of unity that usually persists

despite the intermittent awareness of inconsistencies, but

it does not throw much light on the changes in self-image

following these episodes of "inconsistency."

The closest James comes to discussing these concerns is

near the end of the chapter, in a section entitled "Certain

vicissitudes in the me demand our notice."

In the first place, although its changes are
gradual, they become in time great. . The
identity which the I discovers as it surveys this
long procession {of changes}, can only be a
relative identity, that of a slow shifting in
which there is always some common ingredient
retained. [James cites as an allegory an account
of a pair of worsted stockings that are repaired
with silk until they are made entirely of silk.)
Thus the identity found by the I in its me is only
a loosely construed thing, an identity "on the
whole," just like that which any outside observer
might find in the same assemblage of facts. We
often say of a man "he is so changed one would not
know him;" and so does a man, less often, speak of
himself. These changes in the me, recognized by
the I, or by outside observers, may be grave or
slight. (p.371)

James then goes into a section on "Mutations of the Self,"

but he is referring to extreme cases such as multiple

personalities and the like. Other than this passage on the

gradual changing of the me, James does not appear to

reconcile the multiple social selves with the resultant

identity. And this passage seems to refer more to naturally

evolving changes than to changes wrought by the existence of

many social selves.

In large part, this "deficiency" is understandable given

the nature of certain of James's examples and images. At

the same time, some of these instances, viewed slightly

differently, could easily have led to discussions

immediately applicable to some of the topics of current

inquiry. While the bon-vivant and the philanthropist would


well "trip each other up," the multiplicity of the social

selves would allow their mutual existence in front of

different audiences. And more to the point that will be

developed in this paper, one need not see oneself as either

a bon-vivant or a philanthropist; one may, even in front of

a single audience, be each of these to a certain degree.

This discussion is in no way intended as a rebuke to

James. He appears to have framed his exposition of the

consciousness of self in a manner that did not highlight the

concerns outlined here; he cannot be blamed for failing to

anticipate precisely the interests of those living 90 years

hence. This discourse has been indulged in primarily as a

preface to the forthcoming critique of current research,

which oftens harkens back to James. It was hoped that such

a look at some of the thoughts seminal to much of the

research to be discussed might serve to illuminate some of

the underlying assumptions and in turn suggest some

alternatives. This line of thought will be picked up again

later; for now let us turn to one modern approach to some of

the issues James discussed.

Self-Identification Theory

Schlenker (1980, 1985, 1986) has provided a comprehensive

yet detailed account of many of the same phenomena that

James addressed. Starting with the basic features of self-

identification, he moves from the determinants of behavior

through to the consequences of that behavior for the

identity, which in turn influences subsequent behavior.

Self-identification is defined as "the process, means or

result of showing oneself to be a particular type of person,

thereby specifying one's identity" (Schlenker, 1985, p.66).

Self-identification involves fixing and expressing
one's own identity, privately through reflection
about oneself and publicly through self-
disclosures, self-presentations, and other
activities that serve to project one's identity to
audiences. (1985, p.66)

To further define self-identification, and to clearly

differentiate the intended meaning from possible

misconstruals based on similar or competing approaches,

Schlenker elaborates:

Self-identification is, on any particular
occasion, an activity. It is not merely a
reflection of the self-concept, nor is it simply a
mindless reaction to situational pressures or a
cunning action with Machiavellian intent. Self-
identifications are contextually bound and
influenced by the person, situation and audience.
Yet the actor extracts from them generalizations,
that wittingly or unwittingly, comprise the self-
concept, and once these generalizations are
derived, they in turn influence subsequent self-
identifications. (1986, p.23)

In the last part one hears echoes of James: The self-

identifications are appropriated by the identity and are

constituents of that identity.

Schlenker supports James's argument for both the unity of

the self and the multiplicity of selves that are presented

to others. He says, in effect: We have but one self, a

theory that was created to explain and predict our actions.


And we have many self-identifications that have shaped and

are shaped by our encompassing theory of self. Schlenker

stresses that, like a scientific theory, the theory of self

must receive consensual validation by relevant others.

Self-identification theory also offers a pragmatic

analysis of why we come to hold particular beliefs.

Schlenker identifies two interactive elements: (a)

believability, or the extent to which a belief is a

reasonably accurate construal of the salient evidence; and

(b) personal beneficiality, or the extent to which the

belief serves the holder's goals or values. These two

elements are components of all our beliefs, though the

emphasis may be on one or the other in any given situation.

Schlenker explains:

When applied to self-identification, this analysis
suggests that within the range of potentially
believable self-identifications, that is, the set
of self-beliefs that can be justified and defended
based on salient evidence, people endorse those
that best serve their goals and values. (1986,

He labels these "desirable identity images" or "desirable


This believability X beneficiality formulation applies to

the selection of a self-identification and to the reactions

to a given self-identification (e.g., the degree to which it

is appropriated by the self). The selection of self-

identifications is an area that has been fairly well

researched; see, for example, Baumeister (1982); Baumeister

and Jones (1978); Jones and Pittman (1982); Schlenker

(1980); Schlenker and Leary (1982); Schlenker, Miller and

Leary (1983). The present paper addresses the reactions to

self-identifications, an area that has not been as widely


Regarding the selection of self-identifications and the

reactions to self-identifications, Schlenker (1986) proposes

two possible modes or processes. In the passive mode, self-

identifications occur fairly automatically, without a great

deal of thought or planning, and are made up of familiar,

habitual patterns of behavior organized in well-rehearsed

scripts. Unless problems arise, the behavior is carried out

according to the script. Of the subsequent influence on the

self, Schlenker says:

To the extent that these self-identifications
influence private self-appraisals, they are likely
to do so through a more passive process without
accompanying reflection, such as by the activity
making a particular self-image salient, rather
than by a more active process of contemplation and
rationalization. (1986, p.40)

At other times, though, much thought and energy go into

creating and carrying off a desired self-identification.

This more active mode is engaged if the values, goals and

identity images that exist in the situation are highly

important, or if the individual anticipates or perceives

impediments to the construction of a desired identity.

Active assessment produces more intensified
processing of information pertinent to the
problem, including information about one's
identity. Further, it produces attempts to
reconcile this information with one's desired
identity images as best as possible. (1986, p.41)


When active processing is engaged, people are more sensitive

to relevant information than previously, including self-

information (e.g., self-schemata and self-beliefs). The

increased accessibility of the self-information is thought

to provide a rich data base for resolving the perceived

identity problem.

An important consequence of this active mode of

processing is that the range and type of potential responses

to a given self-identification are greatly increased. As we

shall see shortly, a self-identification that is perceived

as threatening a desired identity image may elicit the

change in self-appraisal predicted by certain other

theories, but it may also cause a "boomerang effect" where

the change in self-appraisal is opposite the direction

expected, or it may lead to the use of accounts (excuses and

justifications) with little or no change in self-appraisal.

With this overview of self-identification in hand, we

turn now to the few studies examining the effects of self-

presentational behaviors on changes in self-appraisals and

on other related responses. Then, we shall look at some

studies applying the cognitive perspective endorsed earlier

by Rogers.

Internalization Studies

In an early study, Gergen (1965) addressed changes in

self-presentation over the course of an interview caused by

reinforcement and instruction, and also looked at the

generalization of these presentations to post-interview

measures. He predicted that subjects instructed to be self-

enhancing ("ingratiation" in Gergen's terminology), knowing

their self-presentations were inflated, would not show as

much carry-over of positivity as those presenting themselves

accurately. In fact, the ingratiators showed just as much

carry-over. To quote Gergen:

In terms of positive increase from the neutral
testing to the generalization period, subjects in
the ingratiation condition were quite similar to
those in the reinforced-accuracy condition. It
seemed that some changes in covert feelings of
self-regard had been produced by the intentional
modification of overt self-ratings. (p. 423)

Gergen found further that the generalization shown by

ingratiation subjects was accounted for almost entirely by

those who were aware of having used the reinforcement

provided by the interviewer as cues for modifying their

responses; unaware subjects showed virtually no


Gergen noted that somewhat similar results had been

discussed in the literature on role playing and attitude

change. For example, studies by Kelman (1953) and Janis and

King (1954) found that the more initiative taken in playing

a role, or the greater improvisation shown, the greater the

subsequent change in attitude. Gergen notes that these

findings have been explained in two ways. First, they have

been explained in terms of implicit verbal responses which

differentially accompany high and low initiative groups.

Those who demonstrated more initiative, it was reasoned,

thought of many more arguments supporting their endorsed

position. Second, Brehm (1960) suggested that subjects in

the high initiative groups were committing themselves to a

position that was dissonant with their initial attitude and

the attitude change was the result of dissonance reduction.

Regarding the first explanation, Gergen writes,

It might thus be said that the aware ingratiation
groups, in their more vigorous involvement,
brought to mind many more positive features of
their own personalities. Once out of the
interview they manifested a higher level of
positiveness in congruence with their immediately
preceding view of self. (p.243)

With regard to the second explanation, Gergen notes

Although such an explanation would be consistent
with the present data, there seems to be no good
reason for contending that the ingratiation
subjects were in a position which would arouse
dissonance. Subjects in these conditions were
offered no choice as to whether to be ingratiating
or not, and the experimenter made it clear that
falsification of self-ratings was appropriate
behavior during the interview situation. (p.243)

Several of the studies to be discussed will return to

explanations similar to these two: (1) the greater pre-

eminence of recently presented features and (2) dissonance

reduction or similar active attempts at coming to terms with

the situation.

Note that Gergen explains the changes in self-

presentation as due to actual changes in self-concept.

Upshaw and Yates (1968) offered an alternative explanation


of the preceding findings. They proposed that elevations

in self-esteem on the post-interview measures were due not

to feedback or self-presentation on a specific dimension,

but rather to a feeling of success on the part of the

subject for having successfully performed the self-

presentation task. To test their hypothesis, they crossed

computer-generated personally-evaluative feedback

(positive/negative) with presentation goal

(positive/negative impression), reasoning that if it was a

sense of achievement or heightened mood that had led

Gergen's subjects to increase their self-esteem ratings,

then subjects who were given the task of presenting

themselves negatively and received feedback that they had

succeeded (that is, made a negative impression) should also

raise their self-esteem. In line with their predictions,

they found that subjects who were instructed to present

themselves negatively and received negative feedback showed

subsequent self-esteem elevations, as did those who were

told to present themselves positively and received positive

feedback. This finding is hard to explain in terms of

Gergen's reflective reinforcement model, which would predict

a drop in self-esteem following a negative presentation

which had elicited negative feedback. However, one may

question how similar the situations were. If Gergen's was a

more truly social situation, the feedback may be responsible

for the change; if Upshaw and Yates's task was seen as a


game of fooling the computer, the feeling of success may be

more important. Of course, it remains possible that both

processes are at work in different situations (see

Schlenker, 1986). In any case, this study points out the

need for focusing on content-specific measures (e.g.,

specific traits) rather than global self-esteem, in order to

reduce the number of alternative explanations.

Fazio, Effrein and Falender (1981) reported an

interesting study investigating changes along a specific

dimension. In a technique borrowed from Snyder and Swann

(1978), they asked subjects a set of questions that were

designed to elicit, and did elicit, either introverted or

extraverted responses. Fazio et al. hypothesized that this

manipulation would increase the accessibility of self-images

congruent with the responses. In accord with this

hypothesis, subjects given the extraverted questions later

rated themselves as more extraverted and even behaved in a

more extraverted manner than those subjects given the

introverted questions. The authors discuss the "intriguing

possibility" that the modified self-concept may maintain

itself for a period of time in other situations, in a self-

fulfilling prophecy: "Believing himself or herself to be

extraverted, an individual may make warm overtures to

others, who then respond and treat the individual warmly,

prompting the individual to respond in that way and

maintaining the self-ascription" (pp.241-242). They relate


their findings to the Jones, Rhodewalt, Berglas and Skelton

(1981) article (which shall be discussed next), noting that

"The present data suggest that such a strategic self-

presentation can have implications for later behavior in a

different situation" (p.240).

In the study just referred to, Jones, Rhodewalt, Berglas

and Skelton (1981, Study 3) were interested in identifying

the conditions necessary for the carry-over of self-

presentation that was observed in two previous studies.

They found that subjects who were instructed to self-enhance

in a simulated interview showed subsequent elevations in

self-esteem only if their presentation was self-referencing

(as opposed to being yoked to a previous subject); the

degree of choice in playing the role these self-enhancing

subjects had did not influence the carry-over effect. On the

other hand, subjects who were instructed to be self-

deprecating showed the carry-over effect (lowered self-

appraisals) only if they had been given high choice; the

self-referencing/yoked variable had no effect.

The explanation Jones et al. offer for these results is

that the self-enhancing subjects' changes represent a self-

perception effect (or more precisely, a biased-scanning

variant), wherein the crucial element is the degree to which

the behavior is seen by the actor as self-reflective; the

self-deprecating subjects' changes represent a cognitive

dissonance effect, wherein the perception of choice is

crucial. The authors openly admit that this account is

speculative and post-hoc, but note that it is fully in

keeping with the Fazio, Zanna and Cooper (1977) exposition

of the differing domains of dissonance and self-perception

theories. They also comment on the implications for the

nature of the self-concept:

Since preference among the theories hinges in part
on assumptions concerning the degree of stability
and structure of the self-concept, it is tempting
to turn this around and suggest that the degree of
self-concept structure may vary with the
evaluative implications of the behavior being
accomodated. The self may thus be asymetrically
structured: at the favorable end there is a range
of values for each disposition that a person will
consider applicable to the self-concept.
Behaviors implying positive self-appraisal are
likely to fall somewhere within this applicable
range and raise few problems of inconsistency or
dissonance. At the unfavorable end, however,
there are few alternatives considered applicable,
and the boundaries of self and nonself are sharply
defined. (p.420)

In this conceptualization it is the positive nature of the

presentation which makes it easily accepted and thus

amenable to self-perception based internalization, or

conversely, the negative tone which forces one to come terms

with one's presentation and thus lower one's self-esteem.

Rhodewalt and Agustsdottir (1986) sought to improve upon

this conceptualization by addressing the possibility that

for some people negative behaviors will be seen as

applicable to the self and positive behaviors as

inapplicable. In this case the pattern of influence of the

variables discussed above should be reversed. Rhodewalt and

Agustsdottir tested this idea by adopting an individual

difference approach (using non-depressed and mildly

depressed subjects) in repeating the Jones et al.

experiment. They predicted that the non-depressed subjects

would replicate the Jones et al. findings but that for the

depressed subjects self-deprecating presentations would fall

in the latitude of acceptance and self-enhancing

presentations would fall in the latitude of rejection, and

the influence of the variables would show the appropriate

reversal. Their results supported their hypothesis: For

depressed subjects, self-deprecating presentations carried

over only if they were self-referencing, with choice having

no effect; self-enhancing presentations carried over only

under high choice, with the self-referencing/yoked variable

having no effect. The authors again propose that dissonance

processes are responsible for carry-over following behaviors

in the latitude of rejection, while carry-over following

behaviors in the latitude of acceptance is due to a variant

of self-perception theory stressing increased accessibility.

We shall return to this proposed "increased accessibility"

shortly; first let us examine some studies related to other

aspects of the above studies.

Note that both the Jones et al. (1981) and the Rhodewalt

and Agustsdottir (1986) papers refer to and implicitly rely

upon the Fazio, Zanna and Cooper (1977) notion of different

processes being at work in one's latitude of acceptance and


latitude of rejection. Neither Jones et al. nor Rhodewalt

and Agustsdottir make explicit use of these latitudes,

however. The former assumed that self-enhancing self-

presentations would fall in the typical college students

latitude of acceptance and self-deprecating self-

presentations would fall in the latitude of rejection.

Rhodewalt and Agustsdottir modified this idea somewhat,

noting that the latitude of acceptance could contain some

negative features and the latitude of rejection could

contain some excessively positive features; moreover, they

noted that for depressives, the latitude of acceptance might

contain more negative features. Nonetheless, they were

still operating on assumptions about the subjects' latitudes

of acceptance and rejection.

Trudeau and Schlenker (1986) adopted a more rigorous

ideographic approach, having each subject explicitly denote

the degrees of independence he or she could (latitude of

acceptance) or could not (latitude of rejection) accept as

self-descriptive. The subject then presented him- or

herself as being described by a level of independence that

was in either the latitude of acceptance or the latitude of

rejection, and was either more or less independent than the

description he or she felt was most self-descriptive.

Further, to test the hypothesis that self-presentations in

the latitude of rejection would generate a more active

response than those in the latitude of acceptance (Fazio et


al., 1977; Schlenker, 1986) half of the subjects were given

a misattribution cue. Also, in keeping with the idea that

this active processing might involve responses other than

changes in self-appraisal (Aronson, 1969; Schlenker, 1986),

subjects were asked about their responsibility for their

self-presentation and how representative and truthful it


In support of the notion of more active processing

following self-presentations in the latitude of rejection,

it was found that the responses were affected by the

presence or absence of the misattribution cue. However, as

anticipated, these differences emerged not in the changes in

self-appraisal but rather in the ratings of the self-

presentation. Subjects who presented in their latitude of

rejection (whether more or less independent than they viewed

themselves) and did not receive the misattribution cue

tended to rate the self-presentation as less representative

of themselves than did subjects in the other conditions.

Ratings of truthfulness and responsibility showed precisely

the same pattern, but the differences were not statistically

significant. Presumably, a subject who received the

misattribution cue and presented in their latitude of

rejection misattributed any tension he or she may have been

experiencing from this self-discrepant behavior, and did not

feel the need to justify or come to terms with the

presentation and its implications for his or her identity.

Subjects who presented in their latitude of acceptance

were generally not affected by the misattribution

manipulation. However, those who presented in the least

independent position they could accept and received the

misattribution cue lowered their independence ratings, the

only group to do so. The authors offered the tentative

explanation that perhaps this presentation, being negative,

was threatening even though it was in the latitude of

acceptance; the misattribution cue may have somehow hindered

counterarguing, and the effect of the presentation was a

lowered self-rating (rather than the use of accounts).

The active process of coming to terms with unacceptable

self-presentations, which led to greater use of excuses in

the Trudeau and Schlenker study, has appeared in other

guises in other studies. Of most interest is the "boomerang

effect" wherein the individual is so intent on counter-

acting an undesirable self-presentation that the resulting

self-appraisal goes beyond the original self-appraisal.

Spivak and Schlenker (1985) found that subjects who

presented themselves negatively on the trait social

sensitivity and received information that the presentation

was representative of themselves later rated themselves more

positively on the trait than either other subjects who made

an equally negative presentation but believed the

presentation not to be representative of self, or control

subjects who did not make a presentation.


Dlugolecki and Schlenker (1987) also found evidence of a

boomerang effect. Subjects presented themselves either

positively or negatively on the trait sociability, with all

presentations coming under conditions designed to foster

perceived representativeness. Following positive

presentations all subjects showed significant increases in

self-ratings of sociability. Following negative

presentations, males' self-ratings did not differ from those

of a no-presentation control group; females' self-ratings

were higher than those for both the control subjects and the

positive presentation subjects. It was suggested that

sociability was more important for females than males, which

led to stronger counterarguing in the females. Support for

this supposition can be seen in the fact that in the

negative presentation condition the females' presentations

were significantly less negative than the males', which the

authors interpret as showing resistance to generating an

unflattering self-presentation.

McKillop and Schlenker (1986) explicitly examined the

influence of the perceived importance of the dimension in

question. When the subjects were told that the trait

practicality was unimportant, self-presentations on the

trait led to passive processing, with subjects shifting

their self-ratings in line with the presentation, increasing

them after a positive presentation and decreasing them after

a negative presentation. When subjects were led to believe


the trait practicality was important, negative presentations

led to increased self-ratings on the trait and also to

denials of the representativeness of the presentations.

Here we see the boomerang effect and the use of accounting,

the results of active processing following undesirable

presentations on an allegedly important trait. Also, the

active memory search involved in the counterarguing process

increased the salience of examples of practical behavior, as

seen in the finding that for subjects presenting negatively,

those who were told the trait was important generated more

positive examples of practical behavior than those told the

trait was unimportant.

Dlugolecki and Schlenker (1985) also tested the idea that

self-presentations should make relevant self-information

more accessible, thereby influencing subsequent self-

appraisals and behavior. Half the subjects took part in an

interview for which they were instructed to create a

positive impression of their sociability, under conditions

maximizing the likelihood of seeing the behavior as

representative of self (i.e., they were given high choice in

selecting the role and were allowed to provide their own

responses; cf. Jones et al., 1981). Other subjects were told

about the interview and its emphasis on sociability, but

were told that they had been randomly chosen to provide

comparison data and would not be participating in the

interview. The subjects who participated in the interview


later described themselves as more sociable on an anonymous

measure and actually behaved more sociably in a waiting room

with a confederate (who did not know whether the subject had

participated in the interview), being more likely to

initiate conversation, talking more, and being rated more

sociable by the confederate.

The Stability and Malleability of the Self

In the wake of these studies investigating changes in

self-appraisal following self-presentations, some

researchers have sought to explicate the underlying

implications for the self, particularly in regard to its

stability and malleability. Recall that in the Jones et al.

and Rhodewalt and Agustsdottir papers the change in self-

appraisals is attributed to cognitive dissonance and biased

scanning or increased accessibility. Rhodewalt (1986) notes

that this formulation is helpful in reconciling the paradox

of the stability and malleability of the self. He proposes

that "the apparent contradiction may be resolved by

recognizing the difference between underlying stable

representations of the self and the experience of the self"

(p. 122). He sees the underlying stable representations as

constituting the latitude of acceptance on any given

dimension, and says these representations are "available."

The phenomenal experience of the self is that particular

self-conception that is accessed at the given point in time.


Which self-conception is accessed can be influenced by the

social environment, a finding seen in McGuire's work on the

spontaneous self-concept (e.g., McGuire and Padawer-Singer,

1976). Rhodewalt labels this accessed self-concept the
"phenomenal self," a term borrowed from Jones and Gerard

(1967), who defined it as "a person's awareness, arising out

of his interaction with the environment, of his own beliefs,

values, attitudes, the links between them, and the

implications for his behavior" (Jones & Gerard, 1967, p.


Thus, in the Jones et al. and the Rhodewalt and

Agustsdottir papers, self-presentations (presumably) in the

latitude of acceptance represent self-descriptions already

available, and the observed change in self-appraisal is due

merely to a change in the phenomenal self (i.e., which self-

conception is accessed). Relating this idea to the

stability/maleability issue, Rhodewalt writes:

The latitude of acceptance is the underlying core
of the phenomenal self. Current contextual cues,
behaviors and affect serve to increase the
relative accessibility or salience of one facet of
the self over others. To the extent that other
people and situations provide consistent cues, the
individual will display apparent consistency and
stability of self-relevant behaviors.
Nonetheless, within any individual's experience is
enough contextual variation and varied social
feedback to shift the person's focus among social
selves (e.g., academic to parent to athlete) and
within self-categories (e.g., the thrill of
victory and the agony of defeat). (p.129)

Note that in this account, for a given dimension (i.e., a

given social self), the available facets of the self (the

latitude of acceptance) should remain fairly stable, even

though the facet being accessed (the phenomenal self) at any

one moment can differ.

For new degrees of a dimension to become available as

self-descriptive, the latitude of acceptance must assimilate

descriptions which had not previously been acceptable.

Rhodewalt suggests that self-presentations of degrees that

had previously fallen in the latitude of rejection can

elicit the strongest change in self-appraisal, provided they

are not otherwise explained away (cf. Trudeau and

Schlenker, 1986). The original social judgement literature

(e.g., Sherif and Hovland, 1961) found that attitude

statements that were in the latitude of non-commital, just

short of the latitude of rejection, elicited the most

attitude change ("assimilation"), while statements in the

latitude of rejection elicited little change and were

frequently seen as more extreme than they were ("contrast"

effect). In the present context, if a person can be induced

to behave in a manner which he would normally not, and yet

not explain the behavior away, his self-appraisals on

relevant dimensions might change substantially. The

literature in social judgement, self-verification (Swann,

1985), cognitive dissonance (e.g., Wicklund and Brehm,

1976), and cognitive conservatism (Greenwald, 1980), among

others, bear testimony to the fragility and unlikelihood of

such a process. Thus, as Rhodewalt proposes, within the

range of acceptable self-descriptions there may be

significant variability as to which self-image is focused

upon, and the self may appear malleable; outside that range,

change in self-description is harder to come by, and the

self appears rather stable.

Markus and Kunda (1986) also confront the paradox of the

malleable yet stable nature of the self (James, 1890). They

propose a "working self-concept" that consists of 1) certain

self-conceptions that are so important to the individual

that they form a stable core, that are "chronically

accessible" (Higgins, King & Mavin, 1982); and 2) more

tentatively held self-conceptions that are occasionally

drawn from the universe of possible applicable self-

conceptions in response t9 the ongoing social environment.

With regard to the question of the malleability and

stability of the self, they write:

Malleability in the self-concept--variations in
the working self-concept--occurs as the context of
self-conceptions surrounding the core element
changes. This mutability or fluidity will be
fairly subtle; it will not, under most
circumstances, involve a major revision or
reorganization of significant self-relevant
thoughts and feelings. Indeed, many typical
assessments of the self-concept will not reflect
these variations in self-conceptions. Changes of
this nature require measures that reveal the
differential availability of self-conceptions and
measures that reveal changes in the meaning or
interpretation given to various self-descriptions.
(p. 859)

The measures that are being endorsed here have previously

been shown useful in investigating certain aspects of the

self. Some of these earlier studies laid the groundwork for

the Markus and Kunda paper, and will be presented prefatory

to its discussion.

The Self and Social Cognition

In a widely cited paper, Markus (1977) showed the

influence of self-schemata on the processing of information

about the self. Subjects who rated at least two of three

independent-synonymous traits as important and also rated

themselves near the end of the scale were assumed to have

self-schemata on the dimension, and were labelled

Independents or Dependents, depending on which end of the

scale they had placed themselves. Subjects who did not see

the traits as important and who fell near the middle of the

scale were labelled Aschematics. Markus found that having a

schemata influenced subjects' processing of relevant

information on a variety of tasks.

In one task, subjects were asked to respond "me" or "not

me" to a series of dependent, independent and neutral

adjectives. For the dependent adjectives, Dependents were

more likely to say "me" than were Independents; for the

independent adjectives, Independents were more likely to say

"me" than were the Dependents. Of more interest is the

finding that these subjects were generally faster at making
"me" judgements for words congruent with their self-schema

than for words incongruent with it. Aschematics showed no

difference in response latency between the dependent and

independent words.

To test the influence of schemata on resistance to

counter-schematic self-relevant information, in a later

session Markus gave subjects bogus test feedback:

Independents were told they were suggestible, Dependents

were told they were not suggestible, and half the

Aschematics were given each type of feedback. In performing

the above task again, none of the three groups changed

significantly from the previous session in their likelihood

of endorsing the two types of adjectives. On the reaction

time measures, the Independents and Dependents were again

faster at responding to schema-congruent words, with the

Aschematics again showing no difference, as above. However,

the Independents and Dependents showed significantly longer

average latencies for the second session compared to the

first; the Aschematics did not show this difference. Markus


It appears, then, that the longer processing times
for the independent/dependent words may well have
been due to the counterschematic information
provided by the suggestibility test. Subjects
with schemata appear to realize that they have
received information about themselves that does
not fit with their current self-conception on this
dimension. While this realization is not
sufficient to warrant a change in self-
characterization (and thus no change in adjective
endorsement is observed), it probably caused these
subjects to reflect slightly longer to check this
information against their schemata before making a
judgement. . Because the Aschematics did not
have an integrated picture of themselves on the
independence-dependence dimension, the
suggestibility information was not perceived as

relevant to the judgments being made. The self-
judgments and latencies for these judgments were,
therefore, not affected. (p. 75)

Note that counterschematic feedback caused an increase in

reaction time in the responses of Independents and

Dependents for both independent and dependent words

generally, with no interaction of the word-type and their

schema (or the type of feedback). We shall return to this

finding shortly.

Several other studies have examined the influence of the

self on reaction time for self-relevant decisions. In line

with their previous studies concerning the self as a stable

underlying structure (e.g., Rogers, Kuiper & Kirker, 1977),

Rogers, Kuiper and Rogers (1979) demonstrated a "symbolic

distance effect" for self-referent judgements. Subjects

first made self-ratings on 14 trait adjectives. For each

possible pair of adjectives, the difference in self-rating

was calculated (labeled a "step"). Then subjects responded

to all possible pairs of the adjectives, indicating which

adjective in each pair described them best. The steps and

the reaction time showed a near-perfect negative correlation

(r=-.97); that is, the greater the difference in self-rating

for a given pair, the quicker the choice between them. This

effect was so strong that the self-ratings accounted for 95

percent of the reaction time variance. Further, the

likelihood that a subject would choose the adjective which

had previously been given the lower self-rating (i.e., an


inconsistent response) increased linearly as the difference

in prior ratings decreased (that is, as the adjectives

became more equally self-descriptive). The authors

conclude that ". . trait discriminability is clearly

related to self-ratings, with larger step-size indicating

better discriminability among the traits in the paired

comparison task" (p.439).

Kuiper (1981) provided convergent evidence for the self

as a prototype. He found that adjectives that were

independently assessed on a self-rating scale as being

extremely like or extremely unlike the self had

significantly faster reaction times when subjects were to

decide if the adjective described them than adjectives which

were only moderately prototypical or self-descriptive. This

inverted-U function has been shown to support the presence

of a prototype (Rosch, 1973; Schnur, 1977). We shall return

to this prototypicality effect and the above symbolic

distance effect when discussing the present study.

McDonald and Kuiper (1985) have shown that content

congruent with a self-schema is processed more quickly than

schema-incongruent content, a finding reminiscent of that of

Markus (1977). Additionally, a concurrent memory task (cf.

Shiffrin and Schneider, 1977) did not interact with this

pattern of results, which the authors say "provides initial

evidence for self-schema processing as an automatic process,

rather than as a process that demands attentional capacity"


Fazio, Herr and Olney (1984) used reaction time to

investigate attitude accessibility following a self-

perception process. They had subjects either recall (Study

1) or anticipate performing (Study 2) behavior that was

either manded or unmanded, and then make evaluative

judgements on items relevant to the behavior. In both

studies, attitude accessibility, as measured by the latency

of the response to the attitudinal inquiries, was enhanced

by the consideration or anticipation of unmanded behavior

but not by manded behavior. The authors conclude that

Apparently, just as Bem's self-perception theory
assumes, freely performed behavior is viewed as
highly reflective of one's attitude toward the
object in question. Consequently, the actor in
such a case can strongly associate the evaluation
implied by the behavior with the attitude object,
producing an attitude that is highly accessible
from memory. (p. 284)

In light of the Markus (1977) study finding that counter-

schematic feedback caused the Independents and Dependents to

show longer latency for both dependent and independent

adjectives, it is worth noting that in the Fazio et al.

study the analysis of the reaction time was conducted on

items referring to the general attitude in question, not to

one end of the attitudinal dimension (i.e., only the

favorable or unfavorable end).

This contrast in possible approaches to analyzing changes

in reaction times becomes important when we return to the

abovementioned Markus and Kunda article that incorporated

reaction time in addressing the stability and malleability

of the self. In this study, subjects were made to feel

either exceptionally similar to ("similarity" subjects) or

different from ("uniqueness" subjects) three confederates.

Following Fromkin (1970), the authors reasoned that extreme

similarity or uniqueness are aversive states that should

lead subjects to reaffirm their less-extreme self-

conceptions by calling to mind self-veryifying self-

conceptions (Swann & Hill,1982; Tesser & Campbell, 1983).

They say:

In fact, the consequences of an embarrassing or a
challenging self-relevant event may actually be a
momentary rise in self-esteem or a brief period of
self-promotion caused by the positive self-
conceptions that are recruited to counteract the
initially negative thoughts about the self (see
also Greenberg & Pyszczynski, 1985; Steele & Liu,
1983). (Markus & Kunda, 1986, p. 859)

This quote calls to mind the "boomerang effects" discussed

earlier. In this study, though, the increased accessibility

of the self-conceptions that are called to mind to

counteract the extreme self-images is not expected to affect

self-ratings. The authors did not expect the manipulations

to cause broad revisions in the core self-conceptions, such

as could be seen in self-descriptions on an adjective

checklist. Rather, they expected the "called to mind" self-

verifying self-conceptions to be more accessible than other

self-conceptions, rendering subtle effects.

Rather than merely infer increased accessibility (cf.,

Rhodewalt, 1986) Markus and Kunda measured reaction time in

accepting or rejecting adjectives relating to similarity and


uniqueness. As they expected, the manipulations did not lead

subjects to differentially accept or reject the uniqueness

or similarity words. Also as expected, they found that, with

respect to uniqueness words, the similarity subjects were

faster to respond "me" than were the uniqueness subjects;

with respect to the similarity words, the uniqueness

subjects were faster to respond "me" than were the

similarity subjects. They conclude that the stability of

the self-concept is evident in the first (null) finding, and

the malleability of the self-concept is evident in the

second finding. This account will now be scrutinized, not so

much to find fault with the article as to anticipate the

present study.

First, the conclusions drawn from the lack of differences

between the treatment groups in the endorsement of

uniqueness and similarity adjectives are questionable. The

use of a null effect to support a prediction is always

troublesome, and is especially so when the measure used is

of less than optimal sensitivity. Although it is rank

speculation, one can imagine that the uniqueness and

similarity subjects may well have differed on self-ratings

of uniqueness and similarity; such a reaction to social

feedback is common in the literature (e.g., S. C. Jones,

1973). In contrast, it would likely require a very powerful

manipulation to lead a subject to respond differently to a

broad adjective such as "unique," when the response involves

only accepting or rejecting that adjective as self-

descriptive. One could argue that this decision is a

probabilistic event and that including many such traits

would allow for the effect. However, for any summative

effect to emerge there must be some difference at the

individual response level, so the problem remains the same.

To facilitate discussion of the Markus and Kunda study

and lay the groundwork for the present study, a working

model of the process of accepting or rejecting a trait or

phrase as self-descriptive is offered. It is proposed that

the decision to accept or reject a general trait description

as self-descriptive occurs in two stages: First, the

phenomenal-self is placed along a scale, relative to the

currently salient comparison group; second, if this

placement is close enough to the appropriate end of the

scale in question, the trait is accepted. The salient

comparison group and what is "close enough" may both be

affected by the social situation (e.g., Morse & Gergen,

1970), mood, instructions, and so forth.

The internalization of self-presentation studies

discussed earlier that showed changes in self-appraisal

showed these changes on self-rating scales. We can apply

this working model to a hypothetical situation in which

these subjects would be asked to accept or reject general

trait adjectives: If the changes in self-rating were extreme

enough to cross the "accept the trait" threshold, then the

subject would alter the endorsement of the trait (e.g.,

accepting a trait which had been rejected). However, the

self-rating could change significantly and yet not cross the

threshold, in which case the trait endorsement would remain

the same (e.g., the trait would still be rejected). Thus,

providing the subject a range of responses (e.g., along a

scale) would allow for change in self-appraisal that a

dichotomous decision--or a series of them--would obscure.

This point, while not original, is important to the study to

be discussed.

The model also addresses the second point, that

concerning the malleability of the self-concept. Recall, it

was reported that with respect to the uniqueness words, the

similarity subjects were quicker to say "me" than the

uniqueness subjects, while the uniqueness subjects were

quicker to accept the similarity adjectives than the

similarity subjects were. Before looking at how this finding

is inconsistent with some previous findings and with the

underlying rational behind the study, it should be noted

that a close reading of the paper reveals that the

statistics for the simple effects supporting this statement

are not reported, only those for the feedback by wordtype

interaction (p. 862). Thus, it is questionable whether the

described effects are significant.

Assuming that they are, the pattern differs from those of

Markus (1977) and Fazio et al. (1984), where the


manipulations were seen as influencing the accessibility of

all aspects of the given dimension, rather than just one

side. Markus found that counterschematic feedback slowed

down decisions for independent and dependent adjectives

generally, with no mention of any effect of the schema-type

(or, equivalently, of the feedback). Fazio et al. simply

analyzed response latency to the presented attitudinal

target, with no consideration of whether it was congruent or

incongruent with the subject's behavior. However, in the

Markus and Kunda paper it is proposed that the self-

conceptions elicited by the threatening feedback make one

end of the relevant scale more accessible, but not the other

end. In the present view, any self-conception that speaks to

one's degree of similarity also speaks to one's uniqueness.

The two might colloquially be called two sides of the same

coin; it would be more precise to say that each self-image

falls somewhere along a similar-unique continuum.

If an elicited self-conception is of a "similar" self-

aspect or behavior, it might lead to increased self-ratings

for similarity, but, as discussed above, it should not lead

one to change such a global self-appraisal as whether or not

one is "similar" (i.e., a yes/no response). At the same

time, if it is assumed that similar is "the opposite" of

unique, we could equivalently say that this self-conception

would lead to lower uniqueness ratings. (If this assumption

is questioned, one must ask why the uniqueness manipulation

is assumed to elicit this "similar" self-conception in the

first place.) If a given cluster of self-conceptions (e.g.,
"similar") can be placed on either a uniqueness or a

similarity scale--which they can--we must ask why they would

enhance reaction time for decisions involving one end of the

scale but not the other. To the extent that this effect

holds, it would appear to be because the cluster is near

that end of the scale ("similar"), making that end of the

scale more like the phenomenal self (cf. Kuiper, 1981).

Recall, however, that Kuiper also found that adjectives that

were extremely unlike the self also had fast reaction times.

If the uniqueness manipulation elicits "similarity" self-

conceptions, it would appear that the unique adjectives

should seem even more unlike the self; in this case, one

might expect the reaction times for rejecting the unique

adjectives to also decrease. Although its context was quite

different, the Fazio et al. (1984) study seemed to support

the notion of the entire scale becoming more accessible.

Similarly, recall that Markus (1977) found that

counterschematic feedback (analogous to the similarity and

uniqueness manipulations) increased reaction time for both

independent and dependent adjectives (i.e., for both ends of

the scale). Therefore, the Markus and Kunda finding is

enigmatic both in light of previous findings and in terms of

the proposed model. Though this study was not designed

explicitly to test this model or to validate certain

findings over others, it was expected to shed light on these


This thought brings us back to the problem discussed

earlier with regards to James's influential chapter. While

we recognize the multiplicity of social selves, we sometimes

construe this variety as occurring among distinct

dimensions, rather than along any one dimension. In light

of this discussion of the Markus and Kunda article, it

might be appropriate to wonder whether having subjects

respond to general traits is more appropriate when we are

interested in changes across dimensions rather than within a

particular one. Most studies investigating the

internalization of self-presentations are more interested in

the latter case, changes along (i.e., within) a given

dimension. Following is a description of a method that

allowed the use of reaction time measures (with their

attendant advantages) and yet addressed changes within a

dimension in a manner more precise than asking subjects to

accept or reject broad traits as self-descriptive.

Reaction Time for Responses to Scale Positions

In the previous section we outlined some possible

limitations to the measures used in Markus and Kunda study.

In that light it is ironic that one of their final

conclusions is that

Measures that assume the self to be a static
structure and require individuals to respond to a
very general description about the self are often

not adequate for revealing how the individual
adjusts and calibrates the working self-concept in
response to the social situation. (p.865)

They endorse the use of more subtle measures such as

reaction time, confidence of self-descriptions, and so on.

While the use of these measures can be helpful, it is

debatable whether they can adequately compensate for having

subjects "respond to a very general description about the

self," such as accepting or rejecting adjectives such as
"unique" or "similar," or their synonyms.

The practice of having the subject simply answer "yes" or
"no" to general questions (regardless of the domain) has

often been disdained, both for psychometric reasons (e.g.,

Nunnally, 1978) and in light of the subject's discomfort at

having to make a "forced choice." Preferred is the Likert

scale, which provides more information and is experientially

more acceptable to the subject, providing shades of gray in

place of the black and white world of dichotomous items.

Further, Sherif and Hovland (1961) argued that forcing

subjects to choose just one point on a scale is still rather

restrictive; they recommended allowing the subject to mark

the position he or she preferred, other acceptable positions

(forming the latitude of acceptance), other positions which

were clearly unacceptable (latitude of rejection), and still

other positions toward which he or she was ambivalent

(latitude of non-commital).

In most reaction time studies, subjects are limited to

accepting or rejecting a given adjective as self-

descriptive. This is no doubt due at least in part to the

desirability of having the subject keep a finger on each of

the response keys, thereby reducing error variance in the

reaction time data. This consideration would seem to render

the use of Likert scales cumbersome if one is interested in

reaction times. However, in the present study the following

modification of the latitudes of acceptance and rejection

paradigm allowed the subject the gradation of responses

afforded by the Likert scale yet preserved the rigor of the

reaction time task. As each point along the scale was

indicated, the subject responded by pressing one of two

keys, indicating whether or not that scale-position was

acceptably self-descriptive. The resultant output provided

not only the latitudes of acceptance and rejection but also

the time necessary to make the decision for each scale-


Of course, there were certain considerations and

compromises in such an approach. First, subjects did not

mark the most preferred position. For present purposes,

though, this was not of great concern. Our interest lay in

comparing the time to respond to descriptions which are

acceptable versus not, so which of the former was most

preferred was not crucial. Second, there were

considerations having to do with the different experiences

for the subject between the typical paper-and-pencil

application of the latitudes measures and that presented on

the computer monitor. Namely, to make the reaction time

data as confound-free as possible, it was decided to

randomize the order of presentation of the positions and to

not indicate the subject's response on the screen. It was

hoped that this approach would increase the likelihood that

the decision for each position would be as independent of

those for the other positions as possible, making the

reaction time data more meaningful. If the positions were

presented in order, there might be a sizable order component

in the reaction time data, which would likely vary in

combination with the subject's latitudes and thus be hard to

remove statistically. If the subject's responses were

indicated during the session, the presentation order might

influence the reaction time for later positions, as the

subject scanned his earlier responses. The chosen approach

avoided these problems but entailed the risk that as the

subject responded to randomly presented scale positions

without being able to see previous responses, the pattern of

responses might not have resulted in cohesive latitudes of

acceptance and rejection.

To address this concern, pilot subjects were presented

with the stimulus materials as described (i.e., randomly

ordered positions without identification of response).

Although there were occasional inconsistencies (e.g., a


rejected position between two accepted ones), for the most

part the response patterns formed cohesive latitudes of

acceptance and rejection. To minimize the risk of potential

inconsistent responses, it was decided that after the

reaction time data were collected for all nine scale

positions, the subject would be given the opportunity to

change any response he or she felt was "wrong." In this way

the reaction time data could be cleanly collected and we

could be relatively certain that the latitudes the subject

indicated were the same as he or she would have indicated on

a standard paper-and-pencil form. As a bonus, with this

approach it was possible to see if positions near the border

of the latitudes of acceptance and rejection underwent more

reversals--"corrections"--as one might expect from the

Rogers et al. (1979) study that showed that as the

adjectives in a paired comparison became more equally self-

descriptive, there was a greater likelihood of inconsistent

responses (that is, the adjective which had been rated less

self-descriptive was chosen).

To allow for more precise comparison of this study's

results to that of other studies, and to address some of the

ideas put forth in the critique of these studies, before the

subject responded to the scale positions for each trait he

first responded to the trait itself (e.g., "independent").

Also, in line with the working model discussed, the list of

traits that were presented for response (along with their

scales) contained antonyms of the traits of interest (e.g.,

"dependent"). Thus, we were able to look not only at the

endorsement of and the response time for two antonyms (cf.

Markus, 1977; Markus & Kunda, 1986) but also at these

measures for each of the scale points for these antonyms.

Of most interest, this approach allowed us to more finely

assess the effects of a self-presentation on subsequent

self-appraisals. After the subject indicated whether the

various degrees of the trait independent were self-

descriptive, he or she was asked to participate in an

interview in which he or she was asked to portray a

particular level of independence. This assignment was co-

determined by the individual's response pattern and random

assignment to a condition. The conditions were 1) the most

independent acceptable position; 2) the position slightly

more independent than that; 3) the least independent

position; 4) the position slightly less independent than

that; 5) the position three positions below the least

independent acceptable one; or 6) a no-presentation control


After the interview the subject again responded to the

scales on the computer and also completed another computer

task and a paper-and-pencil questionnaire. In the other

computer task the subject used a 9-point number pad to rate

him- or herself on each of 30 adjectives and to indicate how

certain he or she was of that rating, how consistent he or


she was on that trait, and how important that trait was to

his or her personality. The questionnaire obtained final

self-ratings on independence and also measured perceptions

of the interview presentation, self-esteem, and other self-



From the work on prototypes (Rosch, 1975; Schnur, 1977)

and, most germanely, that on the self as a prototype

(Kuiper, 1981), it has been found that elements which

clearly do or do not resemble the prototype should be

responded to more quickly than elements for which the

decision is not so clear. Applying this finding to a

latitude of acceptance and rejection framework, we might

predict that decisions for positions near the borders

between the latitude of acceptance and the latitudes of

rejection should require more time than decsions for those

positions clearly in the latitude of acceptance (e.g., the

most preferred position) or clearly not in it (e.g., the

most extreme positions). Graphically, such a prediction

might look like Figure 1.

Alternatively, following the Jones et al. suggestion that

the self is asymmetrically structured, with a sharp break at

the negative end and a fuzzy border at the positive end, the

pattern shown in Figure 2 might emerge.

e T
a i
c m X: X
t e X:X X:X
i X X:X X X:X X
o X X X:X X X:X X X
n X X X:X X X:X X X

Lat. of Rej. :Lat. of Acc: Lat. of Rej.

Figure 1: Predicted Pattern of Latencies

e T
a i
c m
t e X:X
i X X:X X
o X X X:X X X:X X X
n X X X:X X X:X X X

Lat. of Rej :Lat. of Acc: Lat of Rej.

Figure 2: Latencies Possibly Predicted by Jones et al.

If the latitude of acceptance is viewed as the available

self-knowledge (Rhodewalt, 1986), we might expect its

positions to be responded to more quickly than those of the

latiude of rejection; the phenomenal self--the position seen

as best describing the self at the moment--would

theoretically show the fastest response, but this effect may

be elusive since the other acceptances might be very quick.

This pattern is shown in Figure 3.

e T
a i
c m
t e
i X X X: :X X X
0 X X X:X X :X X X
n X X X:X X X:X X X
+-------+--- -------+--- -------+---

Lat. of Rej :Lat. of Acc: Lat. of Rej

Figure 3: Latencies Possibly Predicted by Rhodewalt

Also, since a narrower latitude of acceptance might

indicate greater certainty, we might expect the width to be

negatively correlated with the reaction times.

Of course, the initial patterns of responses and reaction

time are only of partial interest in this study. Equally or

more interesting are the changes in response and reaction

time following a self-presentation varying in its degree of

acceptability and positivity. Perhaps one of the simpler

areas of prediction involves the overall trait decisions.

As described previously, the interview presentation is not

expected to cause a change in the outcome of such a broad

decision. However, it is expected that the decision

latencies wil be affected. Most likely, presentations in

the latitude of acceptance will decrease this latency by

making the relevant self-images more accessible. Recalling

the Trudeau and Schlenker study, we might see differential

effects of presentations at the positive (more independent)

and the negative (less independent) ends of the latitude of

acceptance. In that study it was suggested that perhaps

negative presentations that fall in the latitude of

acceptance also elicit some active processing; in this case

we might expect longer reaction time for this overall

decision. The slightly incongruent presentations are

expected to increase decision latency, particularly the less

independent presentations. The extremely incongruent, less

independent presentations will probably be seen as so

threatening that the overall independence decision will be

rapidly affirmed, in a finding similar to that of the Markus

and Kunda study.

For the other measures, including the decisions for the

scale points representing degrees of the trait, the general

outlook is similar, though the specifics are more

complicated. First, consider the presentations in the

latitude of acceptance. All the theoretical approaches we

have reviewed would seem to predict that the relevant self-

images should become more salient, thus reducing reaction

time. The presentation might serve to consolidate the self-

image, perhaps tightening the latitude of acceptance or

decreasing the latency of decisions near the border--

possibly even for those rejected positions near the border.

Presentations in the latitude of acceptance should decrease

the latency for the 9-point self-rating without

substantially changing its value. The certainty, importance

and consistency measures should not be greatly affected,

since the presentation is just one more self-congruent

behavior; the latencies might be decreased. By the same

token, mood and self-esteem should not be affected, and the

presentations should be accepted as representative and


As discussed earlier, presentations in the latitude of

rejection just outside the latitude of acceptance might

elicit a number of possible responses. Consider the variety

of predictions one might generate concerning the effects of

a presentation in the latitude of rejection near the

latitude of acceptance (slightly incongruent presentation


1. Markus and Kunda's working self-concept
formulation would seem to predict faster
response times, as the self-verification
process would make the arguments against
accepting the position (i.e., self-
conceptions supporting the acceptable
positions) more accessible. The decision
to reject the position should remain the

2. Jones et al. and Rhodewalt and Agustsdottir
might predict that the position would now
be accepted, since the presentation was
under conditions promoting carry-over
(self-referencing, high choice). The
predictions for the response time are not
clear. If the position is now acceptable
(and particularly if it represents the
phenomenal self), reaction time might be
quicker. If, on the other hand, the
dissonance process is at work as the
subject is responding, reaction time might
be longer.

3. Schlenker's self-identification theory
would likely predict longer response time

while the subject considers the
presentation, probably counter-arguing the
implications or explaining it away. The
decision should remain the same--
rejection--but it should take longer than
before. As noted in the discussion, the
theory also addresses the increased
accessibility of self-information that can
result from intensified processing; thus,
the latencies could be reduced. It is
predicted here, though, that the accounting
process should increase the response
latency. Since the conditions of the
presentation should maximize its perceived
representativeness, and since it will
probably not be extremely threatening, the
position might possibly be accepted, but
this is not expected.

All three perspectives might offer slight adjustments

depending on whether the presentation is more independent or

less independent than the latitude of acceptance. For

example, a positive slightly incongruent presentation might

come to be accepted whereas a negative one might again be

rejected. In the Trudeau and Schlenker study, however, a

presentation in the latitude of rejection elicited basically

the same responses regardless of whether it was positive or


These slightly incongruent presentations may influence

subsequent measures other than the latitudes measures.

Although the position presented might be rejected, the self-

rating (on the second task) might be affected, either

towards carry-over or boomerang effect. If we see both

carry-over and boomerang effects, the latencies for each may

be informative (for example, the boomerang effect might be


the result of greater counterarguing, which might result in

longer latencies). Also, incongruent presentations may

decrease the certainty and possibly the consistency ratings,

or they may lead to longer latencies on these measures.

Finally, as in the Trudeau and Schlenker study, we might see

less willingness to accept responsibility for incongruent

presentations, or the tendency to rate them as less

representative of oneself.

For presentations that are deep in the latitude of

rejection (extremely incongruent presentation conditions),

all three approaches discussed above would probably predict

that the decision to reject would recur, though one could

argue that Jones et al. and Rhodewalt and Agustsdottir would

predict that the presentation would carry-over, since

participation was freely chosen. An interesting possibility

is that while the extreme position might still be rejected,

the latitude of acceptance might now lie closer to it,

either by stretching or by shifting. This is probably closer

to the carry-over effect that Jones et al. and Rhodewalt and

Agustsdottir would predict. Alternatively, in light of the

boomerang effects witnessed in some studies, the latitude of

acceptance might move in the direction opposite the

presented position. As mentioned previously, the outcome

here might depend on the positivity of the presentation:

Extremely positive presentations might not be accepted but

they might cause the latitude of acceptance to become more


positive; negative presentations, on the other hand, might

lead to a boomerang effect, with the latitude of acceptance

again becoming more positive.

For these extremely self-incongruent presentations, much

should be learned from changes in reaction time, both for

the position presented and for other positions. Since these

presentations are removed from the latitude of acceptance,

they may potentially influence the responses to the greatest

number of positions. Presentations near the border (i.e.,

slightly incongruent presentations) will probably not, for

the most part, substantially affect the responses to more

removed positions. Extreme presentations, on the other

hand, may affect the responses of positions between the one

presented and the latitude of acceptance, particularly with

regards to reaction time. Also, changes seen in response to

the extreme presentations should help to understand the

change in the slightly-incongruent presentation condition.

The effects on measures other than the latitudes measures

that were discussed for the slightly incongruent

presentations might also be expected for the highly

incongruent presentations; in fact, they might be

exaggerated. On the other hand, the presentations may be so

self-incongruent that they are not perceived of as a threat,

in which case the effects may not appear at all (other than

the reduced representativeness and responsibility).


There are some other topics that are fairly independent

of the type of presentation made. In the post-interview

session, antonyms (e.g., independent and dependent) were

included in the presented adjectives. As discussed in the

context of the Markus and Kunda study, the working model

would predict that the response latencies for one adjective

would resemble that for its antonym, since both decisions

rely on the same internal scale. Alternatively, the Markus

and Kunda findings would lead one to expect that the latency

for one adjective would not impact upon that of its antonym.

The data from the rating scales task should prove helpful

in better understanding the data from the latitudes

measurements (as well as the effect on these of the self-

presentation). First, does the scale rating fall within

the latitude of acceptance on the previous task? Is it the

position with the quickest reaction time in that latitude

(cf., Rhodewalt, 1986)? How do the other judgements for the

trait relate to both the self-rating and the responses in

the latitudes task? We might expect the latency of the

self-rating to be negatively correlated with the certainty

value and the consistency value, and to be positively

correlated with the width of the latitude of acceptance.

Similarly, the width of the latitude of acceptance should be

negatively correlated with the certainty and consistency

ratings. The relationship between these variables and the

importance ratings is less clear. Greater importance may

cause one to deliberate more before responding, but if it is

accompanied by greater certainty (which may or may not be

the case), it may lead to quicker responses; it may be

possible to partial out these influences.

Additionally, by combining the analyses just discussed,

we can compare responses on these scale ratings for the

independent versus the dependent adjectives. For each of

the three levels of adjective desirability, we would expect

the self-ratings for independent and dependent adjetives to

be negatively correlated. Also, the latencies for those

self-ratings should be positively correlated, as discussed

above in the context of the latitudes responses. The

certainty, importance and consistency ratings for

independence should be similar to those ratings for

dependence (a positive correlation).



The primary dependent measure in the study consisted of

pre- and post-interview assessment of the acceptance or

rejection (as self-descriptive) of the trait independent and

of each point on a 9-point scale for that trait, along with

the latency of those responses. In the interview subjects

presented themselves as being described by a position

falling either at one end of their latitude of acceptance or

just outside this latitude; this was crossed with the

position being either more or less independent than the

center of the latitude of acceptance. Another group

presented the third position below their latitude of

acceptance; the corresponding group above the latitude of

acceptance was not included in the design, for fear that

many subjects would not reject the three most independent

positions and thus would have to be discarded since they

would not be able to be randomly assigned to any condition.

Finally, a control group did not participate in the

interview. The basic design is a 2 (More/Less Independent)

X 2 (Latitude of Acceptance/Rejection) factorial, with an

offset Less Independent/Extreme Rejection group and an

offset control group.

Subjects and Procedure

One hundred forty-four introductory psychology students

participated in an experiment entitled "Computer-Aided

Personality Assessment" in partial fulfillment of course

requirements. Of these, 37 could not be used based on their

pre-interview response pattern. This 25% rejection rate

compares favorably to that of Fazio, Zanna and Cooper

(1977), who were unable to use approximately 36% of their

initial subjects. An additional 3 subjects declined to

participate in the interview, and the data from 2 subjects

were lost due to computer malfunction. Data from the

remaining 102 subjects were included in the study.

Before the subject arrived the experimenter randomly

assigned him to one of the six presentation conditions

described. This condition, in conjuntion with the subject's

latitude of acceptance, determined the presentation to be

assigned, as will be explained shortly. When the subject

arrived he was told that he would be asked to participate in

two unrelated tasks, since each required little time. The

experimenter explained that the experiment that he was

involved with concerned the development of new computer-

assisted measures of self-concept, and that in the other

task the subject would be asked to serve as the interviewee

in a simulated interview. This "other experiment" provided

the cover for the necessary presentation manipulation and

will be described shortly.

For the first task, the subject was seated in front of a

personal computer. On the keyboard the M and V keys were

labeled "YES" and "NO" (counterbalanced between subjects).

The subject read the following instructions:

In this task you are to decide whether each of a
series of words or phrases acceptably describes
you or not. You will record your response by
pressing either the YES or the NO key. During the
session, please keep the index finger of each hand
on these keys. After you respond, there will be a
very slight pause and then the next word or phrase
will be presented, so please stay alert. Respond
as quickly as possible without sacrificing your

In this first section you will be responding to
a total of ten traits. First, the trait by itself
will appear on the line to the right of the
screen, below the phrase "Are you: ?". You will
respond YES if you feel the trait describes you
acceptably well, NO if you feel it does not. Then
the following scale will appear to the left of the
screen; the scale refers to the trait, which will
remain on the right part of the screen.

: Extremely above average

: Very much above average

Moderately above average

Slightly above average


Slightly below average

Moderately below average

Very much below average

Extremely below average

Each of the nine positions on the scale will be
highlighted, one at a time, in random order. If
you feel that that degree of the given trait
acceptably describes you, hit YES; if not, hit NO.

For example, if the trait is FRIENDLY and the
highlighted position is "Very much below average,"
if you feel that you are very much below average
in friendliness, hit YES, otherwise hit NO. Then
another position will be highlighted, for
example, "Moderately above average." If you feel
you are moderately above average in friendliness,
hit YES; if not, hit NO. You may respond YES to
as many of the nine positions for each trait as
you feel acceptably describes you.

When you have responded to all nine positions
for a trait, you will be able to change your
responses if you want to. Your responses will
appear beside the scale: "Y" means that you hit
YES, "N" means that you hit NO. If you want to
change a response, use the arrow keys to position
the pointer next to that Y or N, then hit the
response you had intended. When all the responses
are as you had intended, press the Q key. The
screen will go blank, and after a slight pause a
new trait will appear and you will begin the cycle
over again. To familiarize you with the
procedure, there will now be a practice run. If
you have any questions at any time, please ask the

Following the practice run, after the experimenter was

satisfied that the subject understood the task, he cued the

computer to begin and left the room. He went to the

adjacent room, which contained another monitor hooked up to

the computer. The subject's responses were displayed on

it, allowing for the assignment of presentation (described

below). The pre-interview session started with a filler

adjective ("Practical"), followed by "Independent" and then

another filler adjective ("Sincere"). At the completion of

the response cycle for "sincere," when the screen went blank

it remained blank: the computer "broke."

When the subject told the experimenter what had happened,

he acted surprised and said that he would have to have the

computer consultant take a look at the computer. He

suggested that, to save time, the subject could participate

in the interview while the computer was being tended to, and

that hopefully they would be able to complete the task

later, if there was time.

He explained that he did not know much about the other

task, but that the graduate student conducting this study

agreed to help his professor recruit volunteers for a brief

task. Thus, subjects were being asked to volunteer to

participate in a "simulated interview" designed to give

students in a personnel psychology class experience

interviewing people. It was stressed that participation was

voluntary, since the interview was not part of the present

study, but that it would be greatly appreciated. That is,

all subjects were given "high choice" in deciding whether to

participate, to later increase the likelihood that the

interview behavior would be seen as representative of self.

The subject was given a sheet describing the purpose of

the interview and what would be asked of participants. More

specifically, it stated that in order for the interviewers

to obtain a wide range of experience, each interviewee would

be randomly assigned to the way he was to present himself.

Actually, each subject had been randomly assigned to

condition A, B, C, D, E, or F, which in conjunction with his

pretest latitudes scale determined the position he was

assigned to present, as follows. As an example, a subject


with the following pattern of responses could be assigned to

make a presentation of A, B, C, D, or E.

: Extremely above average

: Very much above average

_ :Moderately above average

+ : Slightly above average

+ : Average

+ : Slightly below average

-: Moderately below average

-: Very much below average


E : Extremely below average


+ : accceptable positions

- : unacceptable positions

A : most independent position in the latitude of acceptance

(More Independent/Latitude of Acceptance)

B : least independent position in the latitude of acceptanc

(Less Independent/Latitude of Acceptance)

C : more independent, slightly incongruent position

(More Independent/Latitude of Rejection)

D : less independent, slightly incongruent position

(Less Independent/Latitude of Rejection)

E : less independent, extremely incongruent position

(Less Independent/Extreme Rejection)

F : no-presentation control group



The sheet explained that the interview would focus on one

dimension (independence) in order that the interviewer get a

fairly in-depth impression of the interviewee on that

dimension, rather than a diffuse global view. It mentioned

that the interviewer would later have to justify his

perception of the interviewee, and therefore could not be

told that the subject had been assigned his presentation.

This precaution was included so that the subject would not

be able to quickly "take back" his behavior (deceiving the

interviewer); in this way the impact of the interview was

less likely to be diluted (Davis & Jones, 1960). Finally,

the sheet said that previous simulations had shown that in

order for the subject to be most natural and effective in

the interview, it was best if he thought of times when he

really saw himself as he was now to present himself. That

is, he was urged to really "get into the role." (These

instructions for the interview closely follow those of Jones

et al., 1981, Experiment 3.)

After the subject finished reading this sheet, the

experimenter asked if he would participate. Three subjects

declined; they were debriefed, given credit and dismissed.

The other subjects all agreed to participate, and the

presentation determined earlier was assigned as follows.

The experimenter told the subject how to present himself,

using the labels on the scale from the computer task. As an

example, if the imaginary subject above had been assigned to


the Less Independent/Latitude of Acceptance condition, he

would be told that he was to present himself as "slightly

below average in independence, compared to other University

of Florida students." As further points of reference, the

experimenter mentioned the adjacent scale points, indicating

that they were inappropriate. (For example, "It is

important that you try to come across precisely in this

manner; that is, not simply 'average' or 'moderately below

average', but 'slightly below average.'") No mention was

made of the subject's previous responses or how the assigned

presentation compared to them. After the subject understood

how he was to present himself, the experimenter briefly

described the content of the interview, so that the subject

would know what to expect, could think of responses

conducive to making the proper impression, and could begin

assuming the role. The experimenter repeated that the

subject could decline participating if he so desired. He

then took the subject to the interview room, saying that he

would now have the computer consultant look at the computer.

Actually, at this time he cued the computer to restart in

preparation for the post-interview session.

After the subject entered the interview room, the

interviewer (a research assistant) exchanged a few

pleasantries, then began the interview, which consisted of

four parts. (All the interview items are presented in

Appendix A.) First, the interviewer described some


hypothetical situations pertaining to independence; the

subject chose from possible responses. Second, the

interviewer read a pair of words and the subject apportioned

100 points between them, depending on how well each

described him. Based on pilot ratings these pairs each

consisted of one independent synonym and either a dependent

synonym or a neutral word. Moreover, the items in each pair

were matched as closely as possible on Anderson's (1968)

"Likeableness Ratings of 555 Personality-Trait Words" so

that the subject would respond on the basis of the

independence dimension rather than social desirability. In

the third part of the interview, the subject was asked how

much he agreed (on a 1 to 7 scale) that a certain word

(either independent synonomous or antonymous) described him.

Finally, the subject was asked to give a percentile ranking

of his independence, relative to the U.F. student body.

Throughout the interview, the interviewer recorded the

subject's responses, providing an objective measure of his

presentation. Also, after the subject left the interviewer

recorded his guess as to what position the subject had been

assigned, which he had not been told to reduce possible

demand effects. He also tried to guess whether the assigned

presentation was in the subject's latitude of acceptance or

rejection, and whether it was more or less independent than

the subject's typical self-rating. Interviewers were

instructed to be neutral in demeanor so as not to reinforce

the subject's presentation.


Control subjects were brought to the interview room

without having received a presentation assignment. The

experimenter said that he would return soon with someone to

explain the interview. This gave him a chance to supposedly

have the computer looked at and to cue it to begin the post-

interview session. After waiting ten minutes, about as long

as the interview would take, he returned and told the

subject that the interviewer was not available, and that the

computer had been fixed. From this point the procedure for

the control subjects was the same as that for the treatment


After the interview was finished the experimenter

informed the subject that the computer had been fixed and

that there was time to complete the task. He explained that

unfortunately the previous data were lost when the computer

went down. This justified collecting the data again and

reduced any perceived demand to respond consistently with

the pre-interview session.

The subject was reseated at the computer and was told

that the practice word would again be the same, since it was

programmed to always be first, but that the remaining ten

adjectives were randomly ordered and thus would likely not

be in the same order as before. In the post-interview

session the first trait to be presented was independent,

followed by sincere, conforming, consistent, practical,

dependent, non-conforming, insincere, and impractical. Thus


it was possible to assess changes not only on independent

but also on sincere and practical, which were also responded

to in the pre-interview session but were not the topic of

the interview. Non-conforming was included to see whether

any effects evidenced on independence would generalize to a

synonymous trait. Dependent was included (as were the

antonyms for the other traits mentioned) to allow comparing

the responses for antonyms and the reaction times for the

decisions (cf. Markus & Kunda, 1986).

After the subject responded to the scales for all 10

adjectives, the instructions for the next task appeared. In

this task, 30 adjectives were used (taken from Markus,

1977). Fifteen of the words were independent synonyms, 15

were dependent synonyms; within each group, 5 were in the

upper third of Anderson's (1968) list of "Likeability

Ratings for 555 Adjectives," 5 were in the middle third, and

5 were in the bottom third. For each adjective, the subject

rated 1) how well it described him, 2) how certain he was of

this judgement, 3) how important this trait was to him (as a

part of the self-concept), and 4) how consistent he

perceived himself to be on this trait. These responses were

on a 9-point scale, with the subject responding on a number

pad. The instructions read

In this task you will be asked four questions
about each of a series of adjectives. You will
respond by hitting one of the nine keys (I 9) on
the number pad at the right of the keyboard. 9 is
the high end of the sale, 1 the low end. Between
responses, please keep your finger on the 5 key.
These are the questions you will be asked for each

1. Where would you rate yourself on a 9-point
scale for this trait?

2. How certain are you of this rating?

3. How important is this trait to your

4. How consistent are you on this trait?

This is the format that will be used:

1. The adjective will be displayed on the line

near the top of the screen.

2. You will respond with your self-rating.

3. After a slight pause, the word "CERTAIN ?"

will appear in the middle of the screen.

4. You will respond with how certain you are

that the rating you just gave is accurate.

5. After a slight pause, the word "IMPORTANT

?" will appear in the middle of the


6. You will respond with your rating of how

important the trait is to your personality.

7. After a slight pause, the word "CONSISTENT

?, will appear in the middle of the


8. You will respond with your rating of how

consistent you are on the trait.

9. A new adjective will appear at the top of

the screen, and you will begin the cycle

over again.

When the subject was totally finished with the computer

task, the experimenter thanked him for participating and

asked him to fill out a questionnaire for each of the

projects, to provide feedback to the people in charge. The

first questionnaire was for the computer personality

assessment task, and was very brief (see Appendix B). It

asked how enjoyable the task was, how difficult it was

compared to similar paper-and-pencil questionnaires, and if

the instructions were clear. This information may be useful

in improving subsequent studies; the questionnaire was

intended mostly to make it more plausible that the

experimenter would be giving the second questionnaire, which

concerned the interview (see Appendix C). This

questionnaire assessed a number of possible reactions to the

interview: 1) self-ratings of independence (supposedly for

normative purposes), to see if any change in self-appraisal

had persisted; 2) ratings of mood and the Rosenberg self-

esteem scale, to see if the effects of the interview

generalized; 3) possible accounts, such as choice in

participating, responsibility for the presentation, and its

representativeness; 5) other ratings of the interview

behavior. The subject was also asked to recall the assigned

presentation position, to estimate the impression he


actually made, and asked how these descriptions compare to

the way he usually sees himself. After completing the

questionnaires the subject was probed for suspicion,

thoroughly debriefed, given experimental credit, and

dismissed. The experiment lasted approximately 50 minutes:

5 minutes for the initial instructions and the pre-interview

computer task, 15 minutes for the interview instructions and

the interview, 20 minutes for the post-interview computer

task, and 10 minutes for the questionnaires.


Manipulation Checks

Before addressing possible effects of the interview

presentation on subsequent measures, it is appropriate to

determine that the interview instructions themselves had the

desired effect. As described previously, the More

Independent/Latitude of Rejection group should have

presented themselves in the most independent fashion,

followed by the More Independent/Latitude of Acceptance

group, the Less Independent/Latitude of Acceptance group,

the Less Independent/Latitude of Rejection group, and the

Less Independent/Extreme Rejection group. An objective

measure of each subject's presentation was obtained by

calculating a z-score for each of the component scores of

the interview and then averaging the four z-scores. In a

between-groups ANOVA this average interview response score

showed a highly significant effect of the presentation

condition, F(4,84)=13.27, p<.O002. As Table 1 shows, the

interview responses were generally in the desired order; the

only exception was that the mean score for the More

Independent/Latitude of Acceptance group was non-

significantly higher than that of the More

Independent/Latitude of Rejection group.


As another measure of the effectiveness of the

presentation instructions, immediately after the interview

was completed the interviewer attempted to guess which

position on the 9-point scale the subject had been trying to

portray (with 9 being "extremely above average"). On these

estimates there was again a significant effect of the

presentation condition, F(4,84)=11.61, p<.0001. As shown in

Table 1, these scores follow precisely the same pattern as

the subjects' response scores.


Measures of the Interview Presentation

Response Interviewer's
Presentation Condition in z-score Estimate

More Independent/Rejection .51 6.93
More Independent/Acceptance .63 7.11
Less Independent/Acceptance .05 6.00
Less Independent/Rejection -.38 4.83
Less Independent/Extreme Rej. -.88 3.63

Related measures assessing the subject's perceptions of

and reactions to the presentation were collected in a

pencil-and-paper questionnaire at the end of the experiment.

Most of these measures will be discussed later, but one is

directly relevant now: This question asked how the

presentation compared with the way the subject usually saw

him- or herself on the trait independent. Again, there was

a significant effect of the presentation condition,

F(4,84)=3.91, p<.007. In Table 2 one sees that the mean

scores on this measure fall in perfect order with regard to

the presentation condition. (Note: 15="far better,"

8="same," 1="far worse.")


Comparisons of the Presentation with Usual Self-image

Presentation Condition Comparison Score

More Independent/Rejection 8.75 a
More Independent/Acceptance 8.65 a
Less Independent/Acceptance 7.78 a b
Less Independent/Rejection 6.44 b c
Less Independent/Extreme Rej. 4.58 c

(Means with the same letter are not significantly
different in Duncan's multiple range test, alpha=O.05.)

In light of these three measures, it appears that the

interview instructions had the desired effect: The subjects'

responses, the interviewers' perceptions and the subjects'

comparisons of their presentations with their usual self-

images were all in line with the presentation condition

assignments; however, the More Independent/Latitude of

Acceptance and the More Independent/Latitude of Rejection

conditions were comparable in self-presentations.

Dependent Measures

Since the presentation instructions were effective,

attention can now be focused on the influence the interview

presentations had on subsequent self-appraisals, as well as

on other effects not involving the presentation. These

self-appraisals occurred in several formats. In the first

computer task, subjects indicated whether or not a global

trait was self-descriptive, and then did the same for each

point on a 9-point scale representing degrees of that trait,

in effect marking latitudes of acceptance and rejection.

This first computer task was performed both before and after

the interview containing the self-presentation. In the

second computer task, subjects used a number pad to rate

themselves (from 1 to 9) on each of 30 traits and to

indicate their certainty of that rating, their consistency

on the trait, and the importance of the trait to their

personality. Finally, on a paper questionnaire subjects

made some additional self-appraisals and gave their

perceptions of and reactions to the interview presentation.

The second computer task and the paper questionnaire were

completed only after the interview.

First Computer Task

As expected, the interview presentation did not affect

the likelihood of subjects' accepting the general adjective

independent as self-descriptive: In the pre-interview

session only 9 of 102 subjects said "no." In the post-

interview session, only 8 of 102 said "no."

Table 3 shows the mean latencies for the overall

independent decisions for each of the presentation groups.

For the pre-interview session there was, as expected, no

effect of the interview assignments: In a 2 (Valence: More

versus Less Independent) by 2 (Latitude: Acceptance versus

Rejection) no effects approached significance, F(1,96)<1.0,

ns. For the post-interview session, the Valence by Latitude

interaction approached significance, F(1,96)=2.90, p=.09.

The simple effects for the interaction revealed that within

the Latitude of Acceptance, Valence had a strong effect on

the decision latency, F(1,96)=5.03, p<.03. The More

Independent group responded much more quickly than the Less

Independent group; respectively, the mean latencies were 83

and 129 sixtieths of a second (the unit of measurement in

which the decision latencies were recorded). In the

Latitude of Rejection the effect of the Valence of the

presentation did not approach significance, F(1,96)<1.0, ns.

Also in Table 3 one sees that the Less Independent/Extreme

Rejection group responded as quickly as the More

Independent/Latitude of Acceptance group. Duncan's multiple

range test showed that the Less Independent/Extreme

Rejection group differed from the Less Independent/Latitude

of Acceptance group but not from the Less

Independent/Latitude of Rejection group (alpha = .05). The


quicker decision latencies for the More Independent/Latitude

of Acceptance and the Less Independent/Extreme Rejection

conditions bring to mind previous studies showing quick

processing of schema-congruent information (Fazio, Herr &

Olney, 1984; McDonald & Kuiper, 1985) and extremely

incongruent or threatening information (Markus & Kunda,



Mean Latencies for Overall Independent Decisions

VALENCE Acceptance Rejection I Extreme Rejection
More 117 127 [ --
Less 119 132 I 123

Control 154

VALENCE Acceptance Rejection I Extreme Rejection
More 83 106 [ --
Less 129 103 [ 84

Control 121

The latencies for these and other overall adjective

decisions are useful in addressing other topics mentioned

earlier, such as differences between rather than within

traits, and the relationship of decisions involving antonyms

(e.g., independent and dependent). The mean latencies of

the overall decisions were submitted to a mixed ANOVA, with

single degree of freedom contrasts set up to test the

significance of differences between latencies of successive

adjectives. The effect of the presentation condition did

not approach significance, either alone, F(5,96)=1.05,

p<.40, or in interaction with the adjectives,

F(60,1116)<1.0, ns. The effect of the repeated measure

("adjective") was highly significant, F(12,1116)=9.65,

p<.0001. The adjectives and their mean latencies for the

overall decisions are presented in Table 4. The adjectives

are listed in the order in which they were displayed to the

subject, with the outcomes of the contrasts also displayed.


Mean Latencies for Overall Decisions

Practical 160 >
Independent 128 =
Sincere 128 >
Independent 106 =
Sincere 96 <
Conforming 163 =
Consistent 141 >
Practical 117 <
Dependent 144 =
Non-conforming 155 =
Insincere 142 =
Impractical 144 =
Inconsistent 151

( *: displayed for the second time; alpha=.01)

From these mean it appears that there was a substantial

practice effect, in that the means for the post-interview

independent, sincere, and practical decisions were

substantially reduced from the corresponding pre-interview

latencies. Repeated measures analyses showed the effect of

the session to be significant for each of the three

adjective pairs: For independent the means were 128 and

106, F(1,96)=10.7, p=.002; for sincere, the mean latencies

were 131 and 109, F(1,96)=21.7, p=.0001; for practical, the

latencies were 158 and 119, F(1,96)=14.9, p=.0002.

This practice effect extended only to the precise words

responded to previously. The latencies for the other words

returned to the higher (pre-practice) levels. That is, the

practice effect is due to the adjectives in question and not

merely to becoming accustomed to the task. Of more interest

is the fact that responding to a given adjective (e.g.,

independent) did not reduce the latency of the response to

its antonym (dependent) or to a synonym (non-conforming).

Separate repeated measures ANOVA's were performed to more

rigorously make these comparisons. For instance, when the

post-interview independent latency and the dependent latency

were included as repeated measures, the adjective effect was

highly significant, F(1,96)=13.01, p<.O01. The same effect

held for comparisons of the post-interview sincere latency

and the insincere latency, F(1,96)=31.07, p<.O001, and the

post-interview practical latency and the impractical

latency, F(1,96)=4.02, p<.05. These differences are due to

the reductions in latency caused by the practice effects for


the adjectives that had been responded to previously; the

latter adjective in each pair did not show this improvement

in decision latency. For antonym pairs in which both

adjectives were being responded to for the first time, there

was no significant difference in latency: For conforming and

non-conforming, F(1,96)<1.0, ns, and for consistent and

inconsistent, F(1,96)<1.0, ns. Responding to conforming and

consistent did not significantly reduce the latency of the

decisions for nonconforming and inconsistent.

These comparisons of latencies for antonyms and synonyms

are unfortunately somewhat limited, in that there were

usually intervening unrelated adjectives. If two successive

adjectives tap the same dimension, then, theoretically,

responding to the first could make accessible the same self-

images needed in responding to the second, in which case the

latter might show a decreased decision latency. In this

task, though, only in one instance did two successive

adjectives tap a common dimension: Dependent was followed by

non-conforming. The contrast between their latencies was

insignificant, as seen in Table 4. That is, responding to

dependent did not lead to a lower average latency for non-

conforming. Of course, there are several possible reasons

for this lack of effect, which will be discussed in the next


Turning to the task that required subjects to respond

yes or no for each of the 9 scale points, there were a


number of possible analyses, involving both the response and

the decision latency. It was anticipated that the

opportunity to change a response (for example, from yes to

no) would be utilized most frequently for decisions near the

borders of the latitude of acceptance. However, so few

decisions were changed that this hypothesis could not be


In order to make meaningful comparisons among subjects on

their responses and response latencies for the scale points,

it was necessary to devise an ideographic format that would

allow comparing, for example, the latency for each subject's

most independent accepted position regardless of where on

the 9-point scale this position fell. Accordingly, for each

subject the following positions were identified and given

the accompanying labels (which make reference to the

latitude of acceptance):
1. "High": The most independent acceptable point.

2. "Low": The least independent acceptable point.

3. "Mid": The midpoint between the the previous

positions; if there were an even number

of points and thus no true midpoint, the

latencies for the two points midway

between High and Low were averaged.

4. "Up One": The next point more independent than High.

5. "Down One": The next point less independent than


6. "Down Two": The next point less independent than

Down One.

All the analyses discussed in the next session refer to

these individualized positions, rather than the original

scale points.

Looking first at the post-interview independent scale

responses themselves, there were no differences among the

presentation conditions in the positions accepted, either on

the upper bound, F(5,96)=1.30, p<.30, or on the lower bound,

F(5,96)<1.0, ns, of the latitude of acceptance. That is,

the scale points that comprised "High" and "Low" did not

differ among the presentation conditions. Each subject's

pre-interview response pattern for the independent scale was

used as a baseline, but there was still no effect of the

presentation on the post-interview latitude of acceptance:

For the upper bound, F(5,96)<1.0, ns, and for the lower

bound, F(5,96)<1.0, ns. Just as the interview presentation

did not influence subjects' responses to the overall trait

"independent," it also did not significantly influence the

degree of independence they would accept as self-


The decision latencies for these positions, however, did

show many interesting effects. It was predicted that

decisions for positions close to the borders of the

individual's latitude of acceptance and the latitudes of

rejection would require more thought than decisions for

positions more removed from these borders (i.e., more

clearly acceptable or unacceptable). As can be seen in

Figures 4, 5 and 6, these predictions appear to have been

supported; subsequent analyses were performed to test the

statistical significance of these patterns.

For each of the three adjective scales presented in the

pre-interview session, the reaction times for the

ideographic positions were submitted to a mixed analysis of

variance, with presentation condition as a between-subjects

variable and position as a within-subjects variable. A

potential drawback to this approach is the loss of any

subject whose response pattern was such that he or she did

not have all the positions included in the repeated measures

analysis. As it turned out, this problem occurred most

often when "Mid," the midpoint of the latitude of

acceptance, was included in the analysis: For each

adjective a number of subjects had a latitude of acceptance

only two positions wide, thus having no midpoint. To

include the greatest number of subjects in the analyses,

only "Down Two," "Down One," "Low," "High," and "Up One"

were included in the repeated measures analysis. The other

positions are included in the graphs to give a more complete

picture of the latency patterns, but they are not included

in discussions of the statistical analyses. Since "Mid"

appears to differ from the positions that were included,

deleting it made the analysis more conservative. For each

of the adjectives in the pre-interview computer session

there was a strong effect of position on latency: For


F(4,308)=3.60, p<.007; for independent,

F(4,284)=2.62, p<.04; for sincere, F(4,244)=2.33, p<.05.

The longer decision latencies at the borders of the latitude

of acceptance are reminiscent of those of Rosch (1975) and

Kuiper (1981), who found an inverted-U relationship between

an item's prototypicality and the time needed to respond to



















down3 down2 down



high upl up2

Figure 4: Latencies for Positions of Practical

In the post-interview session the effect of position on

latency persisted for practical, F(4,312)=2.61, p<.04 and






70 63
60 50 rrrr
rrrr rrrr
50 rrrr rrrr
rrrr rrrr
40 rrrr rrrr

82 80 a
rrrr aaaa a
rrrr aaaa a
rrrr aaaa a
rrrr aaaa a
rrrr aaaa a
rrrr aaaa a
rrrr aaaa a
rrrr aaaa a
rrrr aaaa a

85 86 86
Laaa aaaa rrrr
aaa aaaa rrrr
aaa aaaa rrrr 68
aaa aaaa rrrr rrrr
aaa aaaa rrrr rrrr
aaa aaaa rrrr rrrr
aaa aaaa rrrr rrrr
aaa aaaa rrrr rrrr
aaa aaaa rrrr rrrr
aaa aaaa rrrr rrrr

down3 down2

down low

mid high

upl up2

Figure 5: Latencies for Positions of Sincere







60 54
50 rrrr
40 rrrr


95 rrrr
rrrr 89 rrrr
rrrr aaaa 83 rrrr
rrrr aaaa aaaa rrrr
rrrr aaaa aaaa rrrr 73
rrrr aaaa 68 aaaa rrrr rrrr
rrrr aaaa aaaa aaaa rrrr rrrr
rrrr aaaa aaaa aaaa rrrr rrrr
rrrr aaaa aaaa aaaa rrrr rrrr
rrrr aaaa aaaa aaaa rrrr rrrr
rrrr aaaa aaaa aaaa rrrr rrrr
rrrr aaaa aaaa aaaa rrrr rrrr
rrrr aaaa aaaa aaaa rrrr rrrr

down3 down2

down low


high upl up2

Figure 6: Latencies for Positions of Independent

was still close to significant for sincere, F(4,252)=1.97,

p<.10. For independent, however, the position effect did

not approach significance, F(4,300)<1.0, ns. For

independent there was instead a position by presentation

condition interaction, F(20,300)=1.55, p=.06. Examination

of the simple effects of the repeated measure "position" for

each of the presentation conditions revealed that the

differential latency effect persisted for the control group,

which did not participate in the interview, F(4,300)=4.12,

p=.Ol. None of the 5 presentation groups showed a

significant simple effect of position. When the control

group was excluded, the mixed ANOVA for these 5 groups did

not approach significance for either the position effect,

F(4,256)=1.28, p=.30, or the position by presentation

condition interaction, F(16,256)<1.0, ns. The mean position

latencies for the post-interview independent scale are shown

separately for the control group and the 5 presentation

groups in Figures 7 and 8. Appendix D contains the figures

for each of the presentation groups separately.

It is also useful to examine the position by presentation

condition interaction by looking at the simple effects of

condition for each of the positions. The reason for the

differential latency effect persisting for the control group

and not for the presentation groups can be found in the

extreme positions within the latitude of acceptance. For

the most independent acceptable position, there was a








40 rrrr

66 rrrr 64
rrrr rrrr 62 rrrr
rrrr rrrr aaaa 54 rrrr
rrrr rrrr aaaa 51 aaaa rrrr 47
rrrr rrrr aaaa aaaa aaaa rrrr rrrr
rrrr rrrr aaaa aaaa aaaa rrrr rrrr
rrrr rrrr aaaa aaaa aaaa rrrr rrrr
I I -

down3 down2 down low


high upl up2

Figure 7: Latencies for Positions of Independent
(Post-interview): Presentation Groups Only








50 45
40 rrrr

43 47
rrrr rrrr
rrrr rrrr

91 aaaa
aaaa aaaa
aaaa aaaa
aaaa aaaa
aaaa aaaa
aaaa aaaa 66
aaaa aaaa rrrr
aaaa aaaa rrrr
aaaa aaaa rrrr 47
aaaa aaaa rrrr rrrr
aaaa 39 aaaa rrrr rrrr
aaaa aaaa aaaa rrrr rrrr
_____________________________________________________________ I_____________________________________________

down3 down2 down low

mid high

upl up2

Figure 8: Latencies for Positions of
Independent (Post-interview):
No Presentation Control Group Only

significant effect of presentation condition on decision

latency, F(5,96)=2.68, p<.05. The means in Table 5 reveal

that all of the presentation groups responded more quickly

for this position than did the control group. Table 5 also

shows the mean latencies for the least independent

acceptable position. Again, the control group appears to

have required more decision time than the presentation

groups, whose mean latencies are all quite similar.

However, this effect is not significant, F(5,96)<1.0, ns,

probably because the error term is almost half again as

large as that of the previous effect. As one might expect

from looking at Figures 7 and 8, the simple effects of

condition were not significant for any of the other



Mean Latencies for Extreme Acceptable Positions

Most Least
Presentation Condition Independent Independent

More independent/Rejection 60 66
More Independent/Acceptance 49 70
Less Independent/Acceptance 54 62
Less independent/Rejection 39 53
Less independent/Extreme Rej. 61 57
Control (No Presentation) 101 91

With the persistence of the differential latency effect

for all presentation condition groups on the practical and


sincere scales and for the control group on the independent

scale, it is clear that it was not merely practice that

attenuated the effect on the independent scale for the 5

presentation groups. Rather, the interview presentation

involving the trait independent appears to have been

responsible for attenuating the effect.

Second Computer Task

In the second computer task (which occurred only post-

interview) the subject used a number-pad to rate himself on

an adjective displayed on the monitor. Then, cued by words

that appeared under the adjective, the subject rated how

certain he was of that rating, how important the adjective

was to his personality, and how consistent he was on the

trait. Then a new adjective appeared and these 4 decisions

were repeated. Thirty adjectives were presented in random

order, with the 4 decisions for each adjective always

occurring in the order described above. For each of the

resulting 120 decisions, a response (from 1 to 9) and a

reaction time were recorded.

The 30 adjectives were taken from Markus (1977). Fifteen

were independent synonyms and 15 were dependent synonyms.

Within each group of 15, 5 were in the upper third of

Anderson's (1968) list of "Likability Ratings for 555

Adjectives" (labeled "positive"), 5 were in the middle third

("moderate") and 5 were in the bottom third ("negative").


With these adjective categories it was possible to examine

the extent to which subjects' responses were influenced by

the desirability of an adjective as well as its degree of


As with the first computer task, the data collected in

the second computer task lent themselves to a variety of

analyses. The responses for the 4 decisions (rating,

certainty, importance, consistency) for each of the 6

adjective categories (positive independent, positive

dependent, moderate independent, moderate dependent,

negative independent, negative dependent) were examined for

possible influence of the interview presentation condition;

the latencies of these responses were similarly analyzed.

Also, these responses and latencies were submitted to a

variety of correlational analyses, both among themselves and

with data from the first computer task.

A separate analysis of variance was performed on the

responses for each of the 4 decisions for adjectives in each

of the 6 categories (i.e., 24 separate analyses), with

presentation condition as a between-subjects factor. None of

these analyses showed a significant effect of presentation

condition. Next, the ratings for the 3 independent

adjective categories were included as repeated measures

("desirability"), with presentation condition again a

between-subjects variable. There was no effect of

condition, either as a main effect, F(5,96)<1.0, ns, or in


interaction with the adjective category, F(10,192)<1.0, ns.

There was a significant effect of the adjective category,

F(2,192)=3.38, p<.04. Surprisingly, on a 9-point scale with

9 the highest, the positive adjectives received lower self-

ratings (M=5.6) than the moderate (M=6.0) or negative

adjectives (M=5.9). In a similar analysis for the dependent

adjectives, none of these effects approached significance.

The disappointing nature of the output from the second

computer task extended to the latency measures. Unlike

those from the first task, the reaction times for the

adjective ratings showed no interesting effects. A series

of ANOVA's showed no effect of the interview condition on

the decision latency for either the independent or the

dependent adjectives. Efforts to make the analyses more

powerful, such as using logarithms of the latencies and

devising average latency baselines, did not help in

uncovering previously undetected effects. Moreover, though

it was anticipated that the response on the certainty

measure would show a significant negative correlation with

the latency of the adjective rating, the correlation did not

approach significance (r=-.12, p<.30). Nor did the

adjective rating latency correlate with the consistency

ratings (r=-.002) or the importance ratings (r=.01).

The adjective ratings themselves (rather than the

latencies) did correlate with the certainty ratings (r=.29,

p<.Ol), the consistency ratings (r=.55, p<.O001) and the


importance ratings (r=.30, p<.01). While these correlations

could be meaningful, the following evidence suggests that

they may be due to response bias. The adjective ratings for

independent adjectives showed a high positive correlation

with the dependent adjective ratings (r=.46, p<.0001). This

correlation could indicate that subjects were responding to

the desirability of the adjectives more than to the degree

of independence (for example, giving similar ratings to

positive independent and dependent adjectives). However,

for the independent adjectives the ratings for the positive

adjectives and the negative adjectives were also

significantly positively correlated (r=.33, p<.01); for the

dependent adjectives this correlation was also positive but

it was not significant (r=.15, p=.21). If the desirability

of the adjective were the major influence on subjects'

responses, these correlations would be expected to be

negative. Therefore, it does not appear that the

desirability of the adjectives was overwhelming the degree

of independence. Rather, it appears that at least some

subjects were giving similar responses to all the

adjectives, independent or dependent, positive or negative.

The anomalous finding that the positive independent

adjectives received lower average ratings than the moderate

or negative adjectives (with no effect of presentation

condition), along with the puzzling positive correlations

between positive and negative independent adjectives and

between independent and dependent adjectives all suggest

that the data from the second task not be given much

credence. Perhaps having the subject keep the response

finger on the 5 key between responses led to that key

serving as an anchor; other possibilities will be addressed



The paper-and-pencil questionnaire that followed the

second computer task was substantially more informative.

The questions were designed to measure the subject's

reactions to and perceptions of the interview, and to assess

changes in self-appraisal on independence-related and

unrelated traits.

In the major measures in the questionnaire, subjects

rated themselves on independence on a 15-point scale, and

gave a percentile ranking of how independent they were

relative to the university student body. These two measures

were combined, since they were highly correlated (r=.66,

p<.001). Table 6 shows z-scores for each group calculated

from this combined score. To allow for a repeated measures

ANOVA addressing the effect of the interview on self-

ratings, a pre-interview z-score was also calculated. The

only pre-interview measure available came from the first

computer task, in which subjects delineated their latitudes

of acceptance and rejection. From this measure, the middle

of the latitude of acceptance (as described previously) was

used as the pre-interview measure of independence, and z-

scores were calculated. The mixed-model ANOVA, with the

self-rating repeated and Latitude and Valence as between-

subjects variables, was non-significant, F(1,96)=1.43,



Independence Self-ratings


VALENCE Acceptance Rejection I Extreme Rejection
More .27 -.39 I --
Less .00 -.16 I .06

Control .06


VALENCE Acceptance Rejection I Extreme Rejection
More .35 -.54 I --
Less -.31 -.10 I .17

Control .21

As expected, the interview presentation condition

significantly affected subjects' perceptions of their

interview behavior. The mean ratings of the degree to which

the subject felt the interview presentation was truthful,

representative, or typical are listed in Table 7.