Multiple self-presentations and the resiliency of the self


Material Information

Multiple self-presentations and the resiliency of the self
Physical Description:
vii, 125 leaves : ill. ; 28 cm.
McKillop, Kevin James, 1962-
Publication Date:


Subjects / Keywords:
Self-perception   ( lcsh )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1990.
Includes bibliographical references (leaves 116-124).
Statement of Responsibility:
by Kevin James McKillop, JR.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001640249
notis - AHR5334
oclc - 24229097
System ID:

Full Text






This dissertation is dedicated to Anita,
who eased my doubts and fears with her
warmth, understanding, and love.


I would like to thank Doctor Barry Schlenker for his support
and assistance in the preparation of this dissertation. I would also
like to thank Samuel F. Sears, Jr. and the rest of my research
assistants for their enthusiasm and persistence in tracking down
and running subjects.






Literature Review 2
Historical Perspectives 2
Symbolic Interactionist View of the Self-Concept_ 13
Self-Presentation _26
Effect of Self-Presentation on the Self-Concept_ 30
Overview and Hypotheses _______ 53
Overview_ 53
Hypotheses 54

2. METHOD 60

Pretest 60
Experimental Session and Cover Story 60
Independent Variables ____________________ 62
First Presentational Role and Choice
Manipulations 62
Second Presentational Role Manipulation___ 63
Interview Session_____ 64
Dependent Measures and Manipulation Checks 64


Design 68
Manipulation Checks _________ 68
Self-Presentations 68
Choice 69
Dependent Measures 70

Shifts in Sociability_____ 70
Self-Presentations 72
Representativeness 73
Perceptions of the Audience 76
Self-Esteem 77
Mood 79
Behavioral Recall Task _79


Effects of Single Self-Presentations on Self-Rated
Sociability 80
Effects of Multiple Self-Presentations on Self-Rated
Sociability ______________ ----83
The Role of Self-Esteem 84
Effects of Initial Self-Presentations on Subsequent
Self-Presentations 84
The Resiliency of the Self 87
Implications _89
Methodological 89
Theoretical _89



Cover Story 91
First Presentational Role Manipulations ___93
Sociable and Unsociable ___93
Accurate _____94
Second Presentational Role Manipulations _______94
Repeated ____94
Different _________ 95






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



Kevin James McKillop, Jr.

December, 1990

Chairperson: Dr. Barry R. Schlenker
Major Department: Psychology
This study examined the effects of single and multiple self-
presentations on self-ratings of sociability. Subjects were asked to
play the role of a job applicant in videotaped, simulated job
interviews and were instructed to present themselves in a sociable,
accurate, or unsociable manner. Following a single, unsociable self-
presentation, subjects increased their self-ratings of sociability in
an attempt to reaffirm their threatened self-image. This boomerang
effect was eliminated when subjects performed a second
presentation that was either sociable or accurate. Further evidence
of reaffirmation following an initial, unsociable presentation was
found as subjects perceived a sociable, second presentation to be
more representative of their true selves, and gave extremely
sociable second presentations when instructed to be accurate.

These results indicate that, due to active refutation of self-
threatening information, the self-concept may be more resistant to
change than has previously been suggested.


The self is the product of social interaction. This belief, first
expressed by William James (1890) and later expanded upon by
Cooley (1902) and Mead (1934), has generated a significant body of
literature and research in the fields of sociology and social
psychology. The present paper focuses on one of the research
programs derived from this basic belief and traces the history of
this program from James to the present.
In the last 25 years, several researchers have demonstrated
that our presentations to others influence how we feel about
ourselves. These effects have been found to occur across a variety
of self-beliefs and in a number of different situational contexts. To
date, all of this research has involved a single presentation followed
by an immediate assessment of the actor's self-beliefs. No attempt
has been made to more closely approximate en vivo social
interaction by exploring the impact of multiple presentations on
self-beliefs. The present study represents the first step in this

Literature Review
The study of the self was very popular when the science of
psychology was in its infancy. William James (1890) devoted a 111-
page chapter of The principles of psychology to "The Consciousness
of Self" and introspective musings about the nature of the self were
the cornerstones of theories developed by Freud and Jung. However,
as the young science struggled to gain legitimacy and behaviorism
began to dominate the field, the philosophical and often subjective
early writings on the self became less influential and interest in the
study of the self decreased.
In recent years, as the study of social cognition has gained
popularity, the self has re-emerged as a hot topic and is currently
one of the primary areas of social psychological research. With this
renewed interest in the self has come a renewed interest in the
early writings on the topic, many of which are still inspiring new
theories and research. Given the current importance and impact of
these early theories, a brief discussion of some of the more
influential works follows. While this is by no means intended to be
a complete or thorough review of the history of the self in social
psychology, it should provide an adequate foundation for the
discussion of self-presentation and self-concept change that

Historical Perspectives
Many authors agree that the line of development of the self
within social psychology begins with William James. James defined
the self broadly as the sum total of all a man can call his

(1890/1952, p.188). He divided the self into two classes: the pure
ego and the empirical self.
James conceptualized the pure ego as a "certain portion of the
stream [of thoughts] abstracted from the rest felt by all men as
a sort of innermost centre within the circle, of sanctuary within the
citadel" (1890/1952, p. 192). This portion of the stream of
consciousness is the "present judging Thought" which knows and
adopts all thoughts that went before and stands as the
representative of the entire past stream (1890/1952, p.218). It is
the active element of consciousness that seems to welcome or
reject other thoughts and feelings.
That Thought is a vehicle of choice as well as of
cognition; and among the choices it makes are these
appropriations, or repudiations, of its "own." But the
Thought never is an object in its own hands, it never
appropriates or disowns itself. It appropriates to
itself, it is the actual focus of accretion, the hook
from which the chain of past selves dangles, planted
firmly in the Present Anon the hook itself will drop
into the past with all it carries, and then be treated
as an object and appropriated by a new Thought in the
new present which will serve as living hook in turn.
James believed that while we can become aware of the
existence of the pure ego we cannot know about it until it has
passed on and been appropriated by a new "present Thought" (p.219).
Once it has been appropriated it can be viewed as an object and we
can learn about it as a portion of what James called the empirical
James appeared to be much more intrigued by the pure ego
while he spent relatively little time discussing the empirical self.

He defined the empirical self as all that a person is tempted to call
by the name of "me." When we ask ourselves "Who am I?" the
answers we give are components of the empirical self. While these
components may differ in the degree of their association with the
self and strength of the emotions they arouse, James believed that
they all affect one in much the same way: "If they wax and prosper,
he feels triumphant; if they dwindle and die away, he feels cast
down ." (p. 188). James divides the components of the empirical
self into three categories: the material self, the social self, and the
spiritual self.
The material self consists of our body, clothes, immediate
family, house, and possessions. Our possessions are more important
and central to our sense of self when they are the product of
personal labor.
The social self is the recognition a person receives from
others. According to James, people have an intense, innate desire to
be noticed favorably by others and this desire has little to do with
the rational worth of such recognition. To accomplish this we
present different social selves to different groups of people in an
attempt to make a favorable impression. Thus, "Properly speaking, a
man has as many social selves as there are individuals who
recognize him and carry an image of him in their mind" (pp. 189-
190). At first glance, this statement seems to indicate that James
believed we have multiple selves rather than one, true self.
However, James further explains that while we may present many
slightly different selves, we experience them as one identical "Self"
due to their resemblance in a fundamental respect, or their temporal


continuity. "Continuity makes us unite what dissimilarity might
otherwise separate; similarity makes us unite what discontinuity
might hold apart" (p.215). James illustrates the unifying effect of
temporal continuity through the metaphor of "dissolving views."
Dissolving views were similar to modern-day cartoons in that an
illusion of unity and motion was created by flipping rapidly through
a series of slightly different images. While any single image (self)
frozen in time may be different from the others, when viewed as a
part of the unbroken stream a sense of unity is created.
The final component of the empirical self--the spiritual self--
is defined as "man's inner or subjective being, his psychic faculties
or dispositions, taken concretely" (p.191). James argued that "these
psychic dispositions are the most enduring and intimate part of the
self, that which we most verily seem to be" and included among
them "our ability to argue and discriminate, our moral sensibility
and conscience, our indomitable will" (p.191). The spiritual self
differs from the pure ego in that the pure ego is the active "thinker"
while the spiritual self arises from reflectively thinking about
ourselves as thinkers.
The distinction between the pure ego and the empirical self,
first proposed by James (1890), is still popular in the current
literature on the self. Greenwald and Pratkanis (1984) have
developed a computer metaphor for the self which utilizes this
distinction. They compare the pure ego to the computer program and
the empirical self to the input and output or data. This type of
process/content distinction has also been employed by Nisbett and
Wilson (1977) who argue, as did James, that we have introspective

access to the outcomes of cognitive functioning (empirical self), but
not to the process by which they are formed (pure ego). Greenwald
and Pratkanis incorporate James' theory of introspective access into
their computer metaphor by defining the pure ego (the program
aspect of the computer metaphor) "as those aspects of cognitive
function to which we do not have introspective access" (p.142).
Since we cannot accurately assess introspective ability, "the
boundary between mental process and content--like the boundary
between computer program and data--is inherently fuzzy" (p.143).
Greenwald and Pratkanis (1984) further argue that if psychologists
had knowledge of memory and cognition as complete as computer
programmers' knowledge of programming, the process/content
distinction would disappear and the entire self could be viewed as
an object. They propose that the dichotomy within the self
experienced by many and described by James reflects a lack of
understanding and not an actual division of the self. The aspects of
the self which we understand and can describe are labeled the self-
as-known, or the empirical self, while the aspects of self which we
do not understand are labeled the self-as-knower, or pure ego.
Seymour Epstein (1973) recognized the utility of James'
process/content distinction and incorporated it into his depiction of
the self-concept as "a theory that the individual has unwittingly
constructed about himself as an experiencing, functioning individual
S. ." (p. 407). Like a scientific theory, the self-concept contains
constructs which organize and classify certain facts, such as those
pertaining to the material, social, and spiritual selves. These
constructs are related to each other and to future events by sets of

hypotheses which are constantly being tested. The data may support
our hypotheses and result in a strengthening of our self-theory, or
they may fail to support our hypotheses and result in their rejection
or modification, or the formation of auxiliary hypotheses to account
for the findings. Through the formation and testing of hypotheses,
the self-concept influences the acquisition of data and through the
assimilation of new information, it is influenced by data. This view
of the self-concept as a self-theory which is both the subject and
the object of self-knowledge nicely captures James' distinction
between the self-as-knower and the self-as-known.
With the publication of The principles of psychology in 1890,
James decided he had said all that he knew about psychology and
withdrew from the field to spend the remaining 20 years of his life
as America's leading philosopher (Allen, 1967). In his absence,
interest in the self within psychology waned. While behaviorism,
psychoanalysis, and the Gestalt school dominated mainstream
psychology, the study of the self in social life was continued largely
by sociologists such as Cooley and Mead.
Charles Horton Cooley (1902) focused his discussion of the
self on what James called the empirical self and completely ignored
the 'pure ego'--whatever that may be" (p.137). Cooley expressed
his view of the self as a social entity by quoting Goethe: "Only in
man does man know himself; life alone teaches each one what he is."
(Tasso, act 2, scene 3). He denied the existence of a self which is
not social stating: "There is no sense of 'I,' without its
correlative sense of you, or he, or they" (p. 151). Cooley proposed
that the self develops largely by imagining how one appears in the

minds of others. He compared this process to that of viewing
oneself in a looking-glass (mirror) and introduced the phrase,
"looking-glass self." While this metaphor is the most often cited
and well-known descriptor of his theory of self, Cooley noted that it
does not quite capture the essence of the social self.

A self-idea of this sort seems to have three principle
elements: the imagination of our appearance to the
other person; the imagination of his judgement of that
appearance, and some sort of self-feeling, such as
pride or mortification. The comparison with a looking-
glass hardly suggests the second element, the imagined
judgement, which is quite essential. The thing which
moves us to pride or shame is not the mere mechanical
reflection of ourselves, but an imputed sentiment, the
imagined effect of this reflection upon another's
mind We always imagine, and in imagining share, the
judgements of the other mind. (pp. 152-153).

We not only imagine others' judgements of our self, but in
many situations the others themselves may exist only in our
imagination. For Cooley, the "social" self may exist entirely within
the mind of the individual as he or she displays a self and imagines
the reaction of one or more internalized referents.
In tracing the development of the self in infants, Cooley
explained that self-feelings "exist in a vague though vigorous form
at the birth of each individual" (p. 139). He believed this initial
self-feeling was an instinct which evolved due to its motivational
and organizational importance to the individual. From this initial
self-feeling, the looking-glass self begins to develop as the child
becomes aware of the impact its behavior has on others and begins

to alter or manipulate its own behavior to obtain desired reactions
from an audience. As the audiences for the child's performance
change, "the young performer soon learns to be different things to
different people ." (p. 165), indicating that the child is beginning
to understand that others have unique and distinct selves which may
be different from its own. The child identifies certain others as
important and desires to interest and please them. Feedback from
this group of important others has a great impact on the child's
developing social self. As the child matures, it becomes less
interested in the observable reaction of others and focuses instead
on their "internal, invisible condition which his own richer
experience enables him to imagine ." (p. 168). The child also
becomes more subtle in its attempts to produce reactions in others.
At first, the child's quest for positive regard from others is quite
obvious and simple. In time the child begins to understand that such
obvious approval seeking is censured by our society and its motives
become less transparent.
For Cooley, the self arises through social interaction as
individuals present different selves for different audiences, imagine
the others' reactions to these presentations, and experience an
affective response. The individual may then accept the imagined
judgement of the other and alter its self-image accordingly, or more
likely, it will protect its social self "by some sort of withdrawal
[physical or mental] from the suggestions that agitate and harass it,
or in the positive way, by contending with them and learning to
control and transform them, so that they are no longer painful; most
teachers inculcating some sort of combination of these two kinds of

tactics" (p. 219). As we shall see later, this process of social self-
formation and change described by Cooley is central to the self-
presentational theory of self-concept change.
Although he was contemporary with Cooley, George Herbert
Mead's (1934) major work, Mind. self. and society, was published
after his death; this accounts for the discrepancy in the dates of
publication for Cooley and Mead. Working in the same historical.
time period, with many of the same issues, Mead approached the self
in social life from a slightly different perspective than James or
Mead (1934) defined the self as "that which can be an object to
itself" and believed that "it arises in social experience" (p. 140).
Mead's belief in the social genesis of the self is derived from his
social behaviorist orientation.
For the self to become an object to itself, there must be a
behavior which allows the individual to get outside himself, and a
situation which necessitates such behavior (Mead, 1934). Mead
explained that in social situations one must view oneself
objectively, from the standpoints of the other interactants, in order
to act intelligently or rationally. The behavior which allows one to
do this is communication. In conversation with others, one makes a
statement, listens to their response, and adjusts further statements
on the basis of the reply. A more advanced form of communication is
the internal conversation in which one says something, reacts to the
statement by taking an objective point of view, and alters the
statement due to this reaction. The importance of the internal

conversation to Mead's conception of the social self is illustrated by
the following quote:
I know of no other form of behavior than the linguistic
in which the individual is an object to himself, and,
so far as I can see, the individual is not a self in
the reflexive sense unless he is an object to himself.
(p. 142).
Mead attempted to distinguish between the internalized
conversation approach and what he criticized as the "internal and
individual" (p. 224) psychology of Cooley. Cooley focused on
imagining our appearance to others and their response to that
appearance while Mead believed that we can take the other's point of
view and objectively know what their response would be because of
the universality of symbols of communication (i.e., language,
gestures, facial expressions). Mead also criticized Cooley and James
for their "endeavor to find the basis of the self in reflexive
affective experiences, i.e., experiences involving 'self-feeling' .
(p. 173). For Mead, the essence of the self is cognitive and results
from "the internalization and inner dramatization, by the individual,
of the external conversation of significant gestures ." (p. 173).
With his focus on the internal representation of external behavior
Mead may be the earliest cognitive behaviorist.
Mead also differed with Cooley in his belief that the self is not
present at birth, but develops through social experience within the
existing social community. In describing the development of the
self in young children, Mead identifies two key activities: play, and
the game.

While playing, children often take the roles of others and play
at being a teacher, a mother, or a policeman. Children may say
something in one role and then respond to themselves by taking
another role; however, at this level of development they move in and
out of roles quickly and inconsistently with no unified conception of
the relationships between roles. As a result, the self which
develops is a loose organization of the internalized responses of
particular others to each of the many roles the child may play and is
necessarily fragmented and incomplete.
With the introduction of organized games, children must be
ready to take the roles of all of the individuals involved in the game
and understand how each of these roles are related to each other and
to themselves. To accomplish this task, they develop what Mead
called "the generalized other," which is an abstract generalization of
the attitudes of all the other participants in the game, or more
broadly, in the social community. Now, in addition to viewing
themselves from the standpoint of particular others, children are
able to view themselves from the standpoint of others in general.
The internalized responses of the generalized other provide the self
with the unity and stability previously lacking.
The mature self in Meadian terms is a product of one's internal
conversations with particular others and, more importantly, with
the generalized other. By viewing oneself objectively from the point
of view of the specific and generalized others, a unified conception
of self is formed and modified.

Symbolic Interactionist View of the Self-Concept
The work of Cooley and Mead led to the development of
symbolic interactionism--a broad theoretical framework which has
inspired much sociological and psychological research. The symbolic
interactionists view all meaning, including the meaning of the self,
as a product of the negotiation of reality which occurs in social
interaction (Stryker & Statham, 1985). Participants in an
interaction attempt to define the situation and each other through
the exchange of shared symbols (e.g., language). Thus, the symbolic
interactionists believe, as did Cooley and Mead, that the self arises
from social experience as it is defined and redefined based on the
responses of others.
The basic proposition of the symbolic interaction approach,
that social interaction is necessary for the formation of the self, is
difficult to test. A proper test of this hypothesis would require that
the self-concepts of infants raised under conditions of complete
social deprivation be compared to the self-concepts of infants
raised under normal conditions. A comparison of this type was
reported by Dennis (1960) who described an orphanage in Teheran
where each child was kept in a separate, almost soundproof, cubicle
and received little attention from caretakers. Virtually all of these
children were severely mentally retarded despite having come
almost exclusively from the literate population of Iran. Although
this study may provide partial support for the symbolic interaction
perspective, it does not directly test the necessity of social
interaction for the development of the self for at least two reasons.
First, Dennis made no attempt to directly assess self-concept

development among the orphans, and second, due to nonrandom
assignment of infants to the orphanage, confounding factors such as
prenatal care, abnormalities existing at birth, or child abuse may
have contributed to the observed deficits. To control for these and
other confounds, an experiment involving random assignment of
infants to social deprivation conditions would be required. While
such an experiment with human infants is out of the question, Gallup
(1977) was able to conduct several studies of this type with
Studies of self-recognition. The first step in Gallup's (1977)
research was to demonstrate the existence of a self-concept in
primates. Until this time most psychologists viewed the self as a
uniquely human attribute, largely because they lacked a method for
assessing the existence of a self in nonhumans. Gallup provided such
a method using what he termed "mirror-image stimulation," in which
he confronted animals with their reflections in a mirror. Most
animals, when faced with their mirror image, behave as if they are
seeing another animal and engage in other-directed social responses.
Even after prolonged exposure they fail to comprehend the
relationship between their behavior and their reflection and they
never learn to use the mirror for grooming, self-inspection, or other
self-directed behaviors (Gallup, 1968). However, Gallup (1970)
discovered that the great apes (chimpanzees and orangutans) do have
the capacity for self-recognition. When exposed to a full-length
mirror for 10 days, preadolescent chimpanzees initially exhibited
other-directed behaviors; but after the first two days the number of
other-directed responses declined rapidly and were replaced by a

self-directed orientation (Gallup, 1970). The chimps began to use
the mirror to pick food from their teeth and groom parts of their
bodies which they could not see directly. Although Gallup concluded
that these results demonstrated self-recognition in chimpanzees, he
conducted a second, more carefully controlled study to strengthen
his conclusions. Following the 10th day of mirror exposure, each
chimp was anesthetized and marked with bright red, odorless,
nonirritating dye on parts of their faces which they could not see
without a mirror. Once the chimps had recovered from the
anesthetic, a baseline for the number of times a marked area on the
face was touched was obtained by observing the chimps in the
absence of the mirror. After the reintroduction of the mirror the
chimps touched the marked areas 25 times more often than when the
mirror was absent. Given the location of the marks, the anesthesia,
and the properties of the die, the chimps could not possibly have
been aware of the markings unless they saw them in the mirror and
recognized the mirror image as their own.
Gallup (1977) argued that the capacity for self-recognition
presupposes the existence of a self-concept. In order to correctly
recognize its own reflection, an organism must already have at least
a rudimentary concept of self or it would not know who it was
seeing in the mirror.
Having established the existence of self-awareness and,
presumably, a self-concept among the great apes, the next step for
Gallup was to test the hypothesis that social interaction is
necessary for the development of the self-concept. Chimpanzees
reared in isolation and chimps maintained in group cages were

compared in the mirror-recognition task described above. The social
isolate chimps failed to show any evidence of self-recognition
(Gallup, McClure, Hill, & Bundy, 1971). However, when isolation-
reared chimps were given 3 months of social experience with other
chimps, they began to demonstrate the ability to self-recognize
(Hill, Bundy, Gallup, & McClure, 1970). These results provide strong
support for the necessity of social experience in the development of
Gallup (1985) also cites evidence of role-taking ability in
chimpanzees which further supports the process of self-concept
formation proposed by Cooley and Mead. Chimps have been observed
exhibiting deception, concealment, reciprocal altruism, empathy (de
Waal, 1982), and attribution of responsibility (Goodall, 1971). Such
behaviors imply the ability to anticipate the actions and reactions
of others. Even stronger evidence of role-taking ability in chimps
can be found in an incident described by Primack and Primack (1983)
in which a chimp removed a blindfold from her trainer's eyes in
order to lead him to food stored in a box which only the trainer could
open. When the trainer wore the blindfold around his mouth or hair,
the chimp made no attempt to remove it. The chimp had had prior
experience with the blindfold, but she had never seen anyone else
blindfolded, suggesting that she was able to use her own experience
to take the role of her trainer and anticipate his disability.
In a very similar study with human infants, Novey (cited by
Kagan, 1981) allowed 18-, 27-, and 36-month-old children to play
with either transparent or opaque goggles. One day later, when their
mothers put on the opaque goggles, the 27- and 36-month-olds acted

as if their mothers could not see while the 18-month-olds behaved
the same whether their mothers were wearing the opaque or
transparent goggles. It is interesting to note that at this period of
time (the end of the 2nd year) children also begin to exhibit
prosocial behavior, empathy (Zahn-Waxler & Radke-Yarrow, 1982),
and other characteristics which imply some degree of role-taking
ability. Research has also shown that children fail to self-recognize
(in a Gallup-type mirror recognition task) until 18 to 24 months of
age (see Anderson, 1984, for a review). Taken together, these
findings may provide some evidence of an association between role-
taking and the self-concept in humans; it seems unlikely that these
abilities emerge within the same developmental time span by
coincidence. Longitudinal studies of these abilities are necessary to
determine which appears first and whether role-taking ever emerges
without self-awareness or vice versa.
The research on the development of self-awareness in chimps
and humans is consistent with the symbolic interaction perspective.
Gallup (1985) has suggested that role-taking ability may have
evolved due to selective pressures to anticipate and influence the
behavior of others. According to the symbolic interactionists, the
ability to take the role of the other allows us to become aware of
ourselves and to develop a mature self-concept. Research has shown
that social interaction is necessary for the development of self-
awareness in chimps, and that role-taking ability and self-
awareness seem to emerge at about the same time in human infants.
Effects of feedback on the self-concept. Research using human
subjects has also tested various hypotheses derived from Cooley and

Mead's theories about the self. Several researchers have employed
naturalistic, correlational methodologies (Felson, 1989; Manis,
1955; Miyamoto & Dornbusch, 1956) while others have relied upon
experimental designs (Haas & Maehr, 1965; Hicks, 1962; Videbeck,
Correlational research has focused on the interrelationships of
1) the self-concept, 2) the actual responses of others, 3) the
perceived responses of others, and 4) the generalized other.
Miyamoto & Dornbusch (1956) surveyed 195 college undergraduates
in 10 groups ranging from 8 to 48 persons. Subjects were asked to
rate (using 5-point scales) the intelligence, self-confidence,
physical attractiveness, and likableness of themselves (self-
concept) and every other member of their group (actual responses of
others). Using the same scales, subjects also predicted how every
other member would rate them (perceived responses of others) and
how most people in general would rate them (generalized other). Due
to limitations in the experimental design and sampling procedure,
statistical tests of significance were not employed in this study.
The authors chose instead to inspect the data for consistent
tendencies which supported or failed to support their hypotheses.
Miyamoto and Dornbusch's (1956) first three hypotheses
concerned the relationship between the self-concept, the mean
actual responses of others, and the mean perceived responses of
others. As predicted, they found that subjects who actually received
high ratings from others and predicted that they would receive high
ratings from others tended to have high self-ratings. Further,

self-ratings tended to be closer to the mean perceived responses of
others than to the mean actual responses of others.
The fact that the perceived response of others is more closely
related to self-concept than the actual response of others is taken
by the authors as support for both Mead and Cooley's symbolic
interaction perspectives when, actually, it can be argued that this
finding is much more consistent with Cooley's perspective than with
Mead's. It was Cooley who emphasized the importance of the
imagination or perception of another's response while Mead believed
that due to the universality of symbols of communication we can
objectively know what another's response would be. Thus, for Mead,
another's actual response and our perception of that response should
be identical. Although Mead's writings on this issue are somewhat
ambiguous, the above interpretation is certainly consistent with his
belief in an objectively internalized conversation of gestures and
with his cognitive-behavioral point of view.
The final two hypotheses tested by Miyamoto and Dornbusch
involved Mead's concept of the generalized other. They found that
subjects who predicted that they would receive high ratings from
others in general tended to have high self-ratings. In addition, their
self-ratings tended to be closer to their perceptions of the
generalized other than to their perceptions of specific others. This
finding that the self-concept is more closely related to perceptions
of the generalized other than to perceptions of specific others
provides support for Mead's emphasis on the importance of the
generalized other in the formation and modification of the self-

Although this study appears to support several hypotheses
derived from the works of Cooley and Mead, statements of cause and
effect are not possible due to the correlational nature of the
experiment. It may be that one's perception of another's response
(especially the generalized other) does influence self-concept
formation and change; however, it is equally likely that one's self-
concept may influence one's perceptions of another's response and
that this process accounts for the relationships observed by
Miyamoto and Dornbusch. Both Manis (1955) and Felson (1989) have
conducted longitudinal studies of the relationship between self-
concept and the responses of others in attempting to establish the
direction of causation.
Manis (1955) surveyed 101 male dormitory residents 5 weeks
into their freshman year, and again 6 weeks later. Groups of
approximately 8 men each were formed by pairing adjacent dorm
rooms. Subjects were asked to rate each member of their group,
themselves, and their ideal self on 24 bipolar rating scales. Manis
(1955) found that the difference between subject's self-ratings and
their friends' ratings of them decreased from Timel to Time2. This
increase in agreement was found to be due to changes in self-
concept and not changes in others' perceptions. In other words, "the
S's self-concepts were significantly influenced by their friends'
opinions of them" during this 6-week period, but "the friends'
perceptions of the S's were not significantly influenced by the S's
self-images" (p. 369).
Felson (1989) surveyed 338 fourth- through seventh-grade
students and their parents. The children were asked to rate their

intelligence, athletic ability, attractiveness, and popularity.
Parents were also asked to rate their children on these
characteristics and the children were asked to guess how their
parents would rate them. Both the parents and the children were
asked the same questions approximately 1 year later. Using lagged
regression equations Felson (1989) found that children's perceptions
of their parents' appraisals at Timel affected their self-ratings at
Time2 and, to a lesser extent, that parents' actual appraisals at
Time affected their children's self-ratings at Time2.
Interestingly, perceived appraisals did not mediate the relationship
between actual appraisals and the self-appraisals; instead,
perceived and actual appraisals had independent effects on self-
appraisals. In addition, children's self-appraisals at Timel
influenced their perceptions of their parents' appraisals at Time2,
but not their parents' actual appraisals at Time2. These results
again support Cooley's proposition that the self-concept develops
and changes largely through the influence of perceptions biased by
the existing self-concept, and not, as Mead indicated, through
accurate perception of actual audience feedback.
Together, these correlational studies provide strong evidence
of a relationship between the self-concept, actual responses of
others, perceived responses of others, and the generalized other. It
appears that the existing self-concept influences one's perceptions
of the responses of others and these perceptions influence the self-
concept. While actual appraisals of others may also influence the
self-concept, their impact appears to be much weaker and
independent of one's perceptions of them. Finally, one's perceptions

of the responses of others in general is strongly associated with
one's self-concept, but further investigation is necessary to clarify
this relationship.
These studies may establish the existence and, to some extent,
the direction of the relationships among these variables, but, due to
the correlational methodology employed, they do not rule out the
possibility that some variable other than those assessed may be
accounting for the observed relationships. For example, both
Schlenker (1980) and Felson (1989) have proposed that actual ability
may affect both self-appraisals and the appraisals of others.
Although Felson (1989) tested and found no support for this
hypothesis with regard to intelligence, athleticism, and popularity,
it is still possible that some other factor could be influencing these
relationships. Numerous controlled experiments have been
conducted in an attempt to eliminate extraneous variables and
further clarify the effects of others' appraisals on the self-concept.
An early experiment by Videbeck (1960) typifies the
experimental approach to the investigation of the relationship
between others' appraisals and the self-concept. Videbeck (1960)
asked 30 superior students from introductory speech classes to
participate in an experiment purportedly investigating sex
differences in oral communication. Each student read six poems
aloud and then received evaluative feedback from a person who had
been introduced as a visiting speech expert. The evaluations were
actually standardized, prepared statements and half of the subjects
had been randomly selected to receive positive feedback while the
other half received negative feedback. Both prior to the

experimental session and after receiving the evaluations, the
students were asked to rate themselves on 24 items dealing with
oral communication. Some of these items assessed specific aspects
of oral communication criticized by the evaluators while other
items assessed characteristics which were either closely or only
marginally related to those criticized.
Videbeck found that students ratings of themselves increased
slightly (though not significantly) after receiving positive feedback
and decreased significantly after receiving negative feedback. In
addition, the effects of the feedback were confined to the aspects of
oral communication specifically mentioned by the evaluators,
although the closely related items did reveal slight, nonsignificant
trends in the expected direction. The considerably stronger effect
of the negative feedback may have been due to the biased sample of
speech students used in this experiment. These superior students
were probably accustomed to positive evaluations. Thus the positive
evaluation of the speech expert had relatively little effect on the
students' already high self-evaluations while the negative feedback
was quite discrepant with their initial self-beliefs, leading to a
downward revision. Subsequent research has shown that subjects
are generally reluctant to accept unfavorable (Harvey & Clapp, 1965)
or highly discrepant feedback unless the source of the feedback is
perceived to be highly competent and knowledgeable (Binderman,
Fretz, Scott, & Abrams, 1972). These findings are consistent with
Videbeck's given that the source was described as a speech expert.
Hicks (1962) and Haas and Maehr (1965) have demonstrated
that self-concept changes produced by evaluative feedback are

maintained for at least 6 weeks. Hicks (1962) found that positive
evaluations led to initial increases in self-ratings. Repeated
assessments 2 days and 8 days after the receipt of feedback
revealed that the initial increases had not only persisted, but had
generalized to a number of control traits about which subjects had
received neutral feedback. The author suggested that while the
recall of specific item ratings may have dropped over time, the
general tone of the feedback was retained.
In another test of the temporal stability of experimentally
induced self-concept changes, Haas and Maehr (1965, Exp. 1) asked
subjects to rate their general athletic skill and then had them
perform several simple physical tasks. Following the completion of
the tasks, judges described as "physical development experts" (p.
102) provided the subjects with either positive or negative
evaluations. Subjects' self-ratings were reassessed immediately
and 1 day, 6 days, and 6 weeks after the receipt of the feedback.
Self-ratings became more positive immediately following positive
feedback and more negative immediately following negative
feedback, and these changes were still significant (although
somewhat diminished) 6 weeks later.
In a second experiment, Haas and Maehr (1965, Exp. 2)
investigated the effects of repeated evaluations on the self-concept.
The procedure for the second experiment was identical to the first
except that subjects received a second evaluation from a different
"expert" within 48 hours of the initial evaluation. Self-ratings were
reassessed immediately following the first evaluation, immediately
following the second evaluation, 6 days later, and 6 weeks later.

Due to ethical concerns over the long-term effects of negative
feedback, only positive evaluations were given in the second
experiment. Haas and Maehr (1965, Exp. 2) found that subjects' self-
ratings became significantly more positive following the first
evaluation and even more positive following the second evaluation.
In addition, there was virtually no fading or diminishing of these
changes over a 6 week period. The repeated evaluation strengthened
both the immediate and long-term impact of the feedback, leading to
a relatively stable, revised self-concept.
Kinch (1968) also found that repeated evaluations had a more
powerful impact on the self-concept than single evaluations. Kinch
asked subjects to rate their own leadership abilities and several
weeks later placed them in an experimental setting where they were
required to assume a position of leadership and direct the activities
of others. Subjects then received positive evaluations of their
leadership ability from six supposed experts in organizational
dynamics. Some subjects participated in a second leadership
session and received a second set of positive evaluations from the
same experts. After receiving their last set of evaluations, all
subjects reassessed their leadership abilities. Kinch found that
positive evaluations led to increases in self-evaluations, and
repeated evaluations, even when delivered by the same group of
experts, led to greater increases self-evaluations than did the
initial evaluations alone.
The process of self-concept formation and change proposed by
Cooley and Mead is generally supported by the existing research.
However, much of the correlational and virtually all of the

experimental research reviewed thus far has focused on the effects
of direct feedback from other people. As Shrauger and Schoeneman
(1979) have suggested, this aspect of the looking-glass-self may
represent only one of the ways social interaction affects the self-
concept. In fact, two studies by Blumberg (1972) suggest that the
direct communication of either positive or negative evaluations may
be very rare in naturalistic settings. Other aspects of Cooley and
Mead's theories, such as the actor's manipulation of behavior to gain
desired outcomes and imagination of the responses of specific and
generalized others, must be explored to gain a more complete
understanding of self-concept formation and change.

Cooley's focus on the impact of manipulated performances on
the self-concept was largely neglected until Erving Goffman (1959)
renewed interest in the topic with the publication of The
presentation of self in everyday life. Goffman utilized a theatrical
metaphor to express his belief that consciously or unconsciously, we
constantly present ourselves to others and impress them in some
way. Like actors in the theatre, we engage in performances, which
Goffman (1959) defined as "all the activity of a given participant on
a given occasion which serves to influence in any way the other
participants" (p. 15). The breadth of this definition illustrates
Goffman's belief in the ubiquity of presentational behavior.
Although he did speak of a "front region" and "backstage" in social
life where presentational concerns may be more or less prevalent,
he also noted that even "backstage," when one acts out of character,

"this can come to be more of a pose than the performance ." (p.
134). It is, perhaps, unfortunate that Goffman chose the life-as-
theatre analogy to describe his position as this analogy seems to
overemphasize the manipulative, conscious, and deliberate nature of
presentations and fails to capture the less purposeful, unconscious,
and habitual types of presentations which may account for the
majority of presentational behavior.
Current self-presentational theorists are divided over what
should and should not be considered presentational behavior. Some
theorists believe that only intentional, manipulative, or deceptive
types of behavior should be labeled self-presentational (eg. Buss &
Briggs, 1984; Jones & Pittman, 1982) while others (Hogan & Sloan,
1985; Schlenker, 1980) prefer the broader definition proposed by
Goffman (1959).
Jones and Pittman (1982) defined strategic self-presentation
as "those features of behavior affected by power augmentation
motives designed to elicit or shape others' attributions of the
actor's dispositions" (p. 233). They allowed that self-presentations
are not necessarily false or distorted and they claimed that behavior
that is "purely expressive," habitual, task centered, or "authentic"
may be largely immune to self-presentational concerns.
Buss and Briggs (1984) have adopted a similar position in
attempting to specify what is self-presentational behavior and what
is not. They propose that genuine expressions of emotions and
personality traits, spontaneous behaviors, and habitual acts are not
self-presentational and, in fact, represent the opposite pole of a
continuum ranging from self-presentation to expressiveness.

Implicit in this model is the belief that self-presentations are not
accurate representations of the "true" self-concept; that the actor
is somehow deceiving the audience. However, as Hogan and Sloan
(1985) note, there is no reason to believe that expressive people are
any less deceitful than nonexpressive people. Expressive and
spontaneous behaviors convey at least as much self-relevant
information as any other social behavior and are just as likely to be
influenced, consciously or unconsciously, by impression management
concerns. The only difference between expressive and nonexpressive
behavior may be the image the actor seeks to create.
Schlenker (1986) has attempted to resolve this conflict by
proposing a two-dimensional classification scheme based on the
actor's purported motive for the activity (accuracy vs. personal
advantage) and its public or private nature. Behaviors such as self-
reflection and self-disclosure are considered to be attempts to
accurately define the self, while self-presentation and self-
deception are attempts to define the self in a personally
advantageous manner. In addition, self-reflection and self-
deception are considered to be private behaviors performed for the
self, while self-presentation and self-disclosure are performed for
a public audience.
As Schlenker (1986) points out, these dimensions are
extremely useful in classifying the various approaches to the self
that appear in the literature. However, to include these distinctions
in the language of a self-theory may create more confusion than
clarity. For example, it is misleading to label public behavior with
the purported motive of conveying accurate information

"self-disclosure" when such behavior is often used to gain personal
advantage. Also, in limiting self-presentation to public behavior
displayed for personal advantage, we lose much of the richness and
depth of the term implied by Goffman (1959) and supported by
Schlenker (1980) in earlier works. These dimensions seem to
reflect rather than resolve the current controversy over the
definition of self-presentation.
Schlenker (1986) also suggests that these four categories can
be integrated within the theory of self-identification (Schlenker,
1982, 1984, 1985, 1986, 1987; Schlenker & Weigold, 1989). Self-
identification is defined broadly as "the process, means, or result of
showing oneself to be a particular type of person, thereby specifying
one's identity" (Schlenker, 1986, p. 23). Self-presentation, self-
disclosure, self-deception, and self-reflection may all be viewed as
types of self-identification. Thus, the term self-identification is
proposed as a less value-laden substitute for the most general
definition of self-presentation. Schlenker's theory of self-
identification is an extremely useful and detailed theory of self and
will be outlined in later sections, but due to their similarity and in
order to remain consistent with past works, the terms self-
presentation and self-identification will be used interchangeably
throughout this report.
Research from many areas of social psychology demonstrates
that people can and do manipulate their self-presentations.
Baumeister (1982) found evidence of self-presentational behavior in
studies of altruism, conformity, attitude change, aggression,
attribution, interpersonal attraction, and emotion. He proposed. that

people vary their self-presentations either to please an immediate
audience or to construct a public self-image that is congruent with
their ideal self-image. Schlenker (1975) found that people enhance
their self-presentations when there is no threat of public
disconfirmation and present themselves accurately when
disconfirmation is possible. Jones and his colleagues (Jones, 1964;
Jones & Wortman, 1973; Jones & Pittman, 1982) have generated a
large amount of research involving ingratiation, a specific type of
self-presentational behavior "illicitly designed to influence a
particular other person concerning the attractiveness of one's
personal qualities" (Jones, 1964, p. 2). Given the abundance of
research support, few people would question the proposition that
people alter or manipulate their self-presentations. The impact that
these self-presentations has on the existing self-concept is a
subject of much less agreement among self-theorists.

Effect of Self-Presentation on the Self-Concept
In the past 25 years, several researchers have demonstrated
that self-presentations affect the self-concept. Studies have shown
that self-presentations can affect subsequent self-appraisals in
terms of shifts in global self-esteem (Gergen, 1965; Jones, Gergen,
& Davis, 1962; Jones, Rhodewalt, Berglas, & Skelton, 1981; Upshaw
& Yates, 1968) and more specific personality traits (Dlugolecki &
Schlenker, 1987; McKillop & Schlenker, 1988; Schlenker & Trudeau,
1990; Spivak & Schlenker, 1985). Several different perspectives
have been taken in the explanation of these internalization effects.

Gergen (1965) proposed a social reinforcement explanation for
what he termed the "generalization of responses" from presentation
to self-description. Based on a rather narrow interpretation of
Cooley and Mead, Gergen asserted that a person is reinforced by
others for providing certain behavior at certain times, and comes to
feel that the behaviors are representative of self. He provided
support for this proposition by instructing subjects to present
themselves accurately or to try to create the most positive
impression possible on an interviewer. The interviewer either
provided the subject with "reflective reinforcement" by agreeing
with positive statements about self and disagreeing with negative
statements, or provided no reinforcement. Gergen expected to find
positive shifts in self-ratings only when the subjects were
instructed to present themselves accurately and received reflective
reinforcement. However, generalization from presentation to
subsequent self-ratings occurred whenever subjects received
reflective reinforcement, regardless of the actor's goal in the
situation (accuracy, ingratiation). He interpreted these results as
strong support for the hypothesis that the self-concept is a result
of social learning. Jones' (1964) interpretation of these findings
was that, "persons tend to exaggerate the perceived
representativeness or felt sincerity of any performance which
elicits approval" (p. 67). In support of this explanation, an earlier
study by Jones, Gergen and Davis (1962) found that subjects rated
their prior self-presentations as more accurate when they believed
they had made a favorable rather than unfavorable impression. The
implication of Jones' (1964) interpretation is that

representativeness, or the degree to which a self-presentation is
perceived to be an accurate portrayal of one's self-concept, is a key
in determining when a presentation will be internalized.
Upshaw and Yates (1968) offer an alternative explanation for
the findings of Gergen (1965). They propose a "task success"
hypothesis whereby the successful completion of the task leads to
the increase in self-esteem. Experimental support for this
hypothesis was provided using a paradigm conceptually similar to
that used by Gergen (1965). Subjects were instructed to make the
best or worst possible impression when answering a personality
survey and were given either favorable or unfavorable feedback
(supposedly generated by a computer scoring of their survey
answers). Subjects who attempted to form a favorable impression
and received favorable feedback and subjects who attempted a
negative presentation and received unfavorable feedback scored
significantly higher on a subsequent test of self-esteem than did
subjects in the other conditions. No main effects of presentation or
feedback were found. The authors concluded that when feedback
indicates the attempted presentation has been successful (task
success), self-esteem increases. While these findings may appear
to be inconsistent with those of Gergen (1965) it must be noted that
the studies involved different contexts. Schlenker (1986) suggests
that "self-presentations may affect both corresponding self-beliefs
and feelings of self-efficacy, and different contexts may make one
or both relatively salient" (p. 38). The game context of fooling a
computer, used by Upshaw and Yates, may be the type of situation
that would emphasize succeeding and not the relevance of

self-presentations to self-beliefs, while the task used by Gergen,
creating a positive impression on an interviewer who provides
personal feedback after an interaction, may focus attention on the
relationship between self-presentations and self-beliefs
(Schlenker, 1986).
The findings and interpretation of Upshaw and Yates (1968) are
even more questionable in light of a recent study by Jones, Brenner,
and Knight (1990). Jones et al. (1990) audiotaped subjects
instructed to play the role of a morally reprehensible job
interviewee. One week later, subjects received feedback
(presumably from psychology students in a person perception
seminar) that indicated they had failed or succeeded in their
presentational attempt. While high self-monitors (Snyder, 1974) did
exhibit higher self-esteem following success feedback than
following failure feedback, low self-monitors displayed higher self-
esteem following failure feedback than following success feedback.
Dissonance theory has also been employed to explain the
effects of self-presentations on self-conceptions. The original
version of dissonance theory (Festinger, 1957) posits the existence
of a drive-like motivation to maintain consistency among relevant
cognitions. One prominent version of the theory (Wicklund & Brehm,
1976) states that dissonance arousal occurs only when people are
personally responsible for attitude discrepant behaviors that
produce negative consequences. Applied to the research on
internalization of self-presentation, inconsistency between prior
self-concept and self-presentation brings about a state of aversive
arousal and a motivation to reduce this arousal, if one is personally

responsible for the presentation and the presentation produces
negative consequences. Dissonance theory proposes that to reduce
arousal, people will change the cognitive element that is least
resistant to alteration (Wicklund and Brehm, 1976). Shifting one's
self-concept in the direction of the self-presentation is one way of
reducing dissonance, however, there are alternative modes of
dissonance reduction. If the actor does not appear to be responsible
for the behavior, then no change in the self-concept should occur
since the actor may blame the behavior on situational demands. In
addition, if the relevant aspect of the self-concept is particularly
strong or stable, the actor may reduce dissonance by perceiving the
presentation to be less discrepant than it actually is.
One problem with the dissonance explanation is that in order
for a self-presentation to be discrepant, the aspect of the self-
concept involved must be sufficiently clear and stable and not in
some way implied by the presentation. Given that dissonance theory
predicts that people will change the least resistant cognitive
element, and a clear and stable aspect of one's self-concept would
certainly be more resistant than a single self-presentation, when
and how can dissonance theory adequately predict the internaliz-
ation of self-presentations? Actors should be much more likely
retain their initial self-beliefs and instead alter their perceptions
of the discrepancy, representativeness, and situational control of
their self-presentation.
Another explanation for the internalization of self-
presentations can be derived from self-perception theory (Bem,
1972). Bem proposes that people may draw inferences about

characteristics of self "from observations of their own overt
behavior and/or the circumstances in which this behavior occurs"
(p. 2). In comparison to dissonance theory, self-perception theory
places less emphasis on the structure and stability of the self-
concept and, in fact, implies that the self-concept must be rather
fuzzy and mutable in order for self-perception processes to take
place. In addition, the self-perception explanation does not
postulate an aversive motivational pressure which drives the
individual to attitude change. Bem sees the actor, in many
situations, as being in the same position as any other observer of his
behavior; both are relying on the behavior and the conditions under
which it occurred to infer the actor's attitude. This inferential
process occurs only when internal cues are weak or ambiguous and
external contingencies are insufficient to account for the behavior.
When internal cues are strong and/or the external contingencies for
the behavior obvious, self-perception will not occur and self-
concept will not shift in the direction of the self-presentation.
Since Bem's (1967) proposal of self-perception as an
alternative to dissonance theory, a great deal of discussion and
research has been generated in the attempt to demonstrate the
superiority of one or the other of the competing theories (cf.
Greenwald, 1975). Despite numerous attempts at "crucial"
experiments (e.g., Snyder and Ebbeson, 1972; Swann and Pittman,
1975), neither theory has been proven to be relatively superior.
Several authors (Fazio, Zanna, and Cooper, 1977; Greenwald, 1975;
Jones et al., 1981) have suggested that we should no longer regard
the theories as competing, but as complementary.

Fazio, Zanna, and Cooper (1977) have suggested that the
theories of cognitive dissonance and self-perception each apply to a
different domain and, together, the two theories provide an
explanation for attitude change. According to Fazio et al. (1977),
attitude change following attitude-discrepant behavior is mediated
by cognitive dissonance. They derive this hypothesis from research
on the effect of misattribution cues on attitude change.
In one such study, Zanna and Cooper (1974) had subjects write
counterattitudinal essays under conditions of high or low choice
after they had been given a pill (placebo) which ostensibly produced
a side effect of relaxation, tension, or no side effect. In the no side
effects condition, subjects demonstrated the typical effect in which
high choice subjects changed their attitudes more than low choice
subjects, however, this effect failed to occur in the conditions
where subjects could attribute their arousal to the pill which
supposedly made them feel tense. The effect was accentuated in the
relaxation conditions, presumably because the high choice subjects
experienced arousal, despite believing they had taken a relaxing
drug. From this study Fazio et al. (1977) infer that aversive
arousal does occur when one performs a counterattitudinal behavior,
a finding which directly contradicts the self-perception view that
people infer their attitudes from counterattitudinal behavior
without experiencing aversive motivational pressure.
Fazio et al. (1977) further suggest that attitude change
following attitude congruent behavior is mediated by self-
perception. They point out that dissonance theory is not applicable
to situations in which proattitudinal advocacy occurs since the

theory predicts attitude change only if the behavior performed is
discrepant with the attitude. Self-perception theory can explain
changes following attitude congruent behavior because the theory
predicts attitude change if the advocated position falls anywhere
other than the person's preferred position on the attitudinal
In order to operationalize the concepts of attitude-discrepant
and attitude-consistent behaviors, Fazio et al. (1977) borrowed
from the work of Sherif and Hovland (1961) on latitudes of
acceptance and rejection. An attitude-congruent behavior was
defined as the endorsement of any position within an individual's
latitude of acceptance while attitude-discrepant behavior was
defined as the endorsement of any position within an individual's
latitude of rejection.
To test their hypothesis concerning the differential
applicability of dissonance and self-perception theories, Fazio et al.
(1977) first pretested subjects to obtain their initial latitudes of
acceptance and rejection and then gave them either high or low
choice to endorse (by writing an essay) either the most extreme
position in their latitude of acceptance, or the least extreme
position in their latitude of rejection. Half the high choice subjects
were told that any arousal they may experience might be due to the
small booths in which they were seated (misattribution cue). As
predicted, subjects who endorsed a position within their latitude of
acceptance shifted their attitudes toward that endorsement when
they had high choice, even when they were provided with an
opportunity to misattribute arousal. Since the opportunity to

misattribute arousal did not attenuate attitude change, Fazio et al.
(1977) proposed that dissonance arousal did not occur and self-
perception processes were responsible for the attitude change.
Subjects who freely chose to endorse a position within their
latitude of rejection shifted their attitudes toward that
endorsement only when they were given no misattribution cue,
indicating that dissonance arousal was responsible for the shift.
Recently, Jones, Rhodewalt, Berglas, and Skelton (1981)
conducted a set of studies designed to clarify the theoretical
processes involved in the internalization of self-presentations. In
the first study, subjects were shown videotaped portions of three
previous interviews in which subjects had been either self-
enhancing or self-deprecating. Half of the subjects viewed the
tapes before they participated in an interview, while the other half
viewed the tapes afterwards. When the tapes were viewed after the
interview, they had (of course) no effect on interview behavior and
no pre-post interview change in self-esteem ratings was observed.
Subjects who viewed self-enhancing or self-deprecating models
before their interview imitated this behavior during their interview
session. Also, subsequent ratings of self-esteem indicated a
significant carry-over effect, with subjects shifting their self-
esteem in the direction of their self-presentation. A second study
replicated these findings in a role playing context designed to
reduce the need for the radical deceptions which were used in the
first study.
A third study extended the argument of Fazio, Zanna, and
Cooper (1977) to attitudes toward the self, specifically

self-esteem. Jones et al. propose that, since people generally
regard themselves positively, a self-deprecating presentation
should fall in an individual's latitude of rejection and self-esteem
changes should be mediated by dissonance arousal, while self-
enhancing presentations should fall into an individual's latitude of
acceptance and invoke a biased-scanning variant of the self-
perception process. This biased-scanning approach views the self-
concept as "a complex set of alternative conceptions with
continuously shifting salience" (Jones et al., 1981, p. 408). Self-
presentation has the effect of making one of these alternative views
of the self salient, and this view will remain salient unless
something intervenes to redefine the self. This approach differs
from the self-perception assumption that the actor's prior self-
conception is essentially overwhelmed by the more salient self-
presentational behavior. Jones et al. propose that the biased-
scanning variant differs operationally from dissonance theory in
that "the crucial ingredient in biased scanning theory is not whether
an individual has or has not the freedom to engage in the suggested
behavior at all, but whether the behavior is seen as owned by the
actor and reflective of his or her phenomenal self" (Jones et al.,
1981, p. 419).
Jones et al. (1981, Exp. 3) provided support for the biased-
scanning hypothesis and the differential application of self-
perception and dissonance theories. In this study, subjects were
asked to act in a self-enhancing or self-deprecating manner in a
simulated job interview and were either given or not given a choice
to perform the task. Further, subjects were either allowed to

generate their own self-presentation within the bounds of the
instructions (self-referencing), or were asked to follow a script
based on the responses of a matched subject in the self-generated
presentation condition (yoked). Jones et al. found that subjects in
the self-enhancing condition raised their subsequent self-esteem
ratings when the presentation was self-referencing, but not when it
was yoked. The choice variable had no effect in the self-enhancing
condition, as the biased-scanning hypothesis would predict. In the
self-deprecating condition, only the choice variable had an effect.
Subjects given a choice whether or not to perform a self-
deprecating presentation later exhibited decreased self-esteem,
while those in the no-choice condition exhibited no change. The
self-referencing variable had no effect in the self-deprecating
The assertion by Jones et al. that choice and personal
responsibility can be used to distinguish self-concept change due to
dissonance from self-concept change due to self-perception has
been sharply criticized by Schlenker (1986). Schlenker provides an
alternative explanation of the findings based on the perceived self-
representativeness of the behavior. A presentation is perceived to
be representative of the self to the extent that it seems to be
descriptive of enduring personal characteristics and this perception
may be influenced by many factors, -including the choice and self-
referencing manipulations used by Jones et al. (1981). Schlenker
suggests that "subjects in the Jones et al. study may simply have
shifted their self-feelings in the direction of their behavior
whenever the behavior seemed to be representative of the self"

(Schlenker, 1986, p. 39). Self-deprecation is such an uncommon and
socially unexpected behavior that people may consider it self-
representative only when they have freely chosen to perform the
role. On the other hand, since self-enhancement is a far more
common and expected behavior, virtually everyone would agree to
perform a self-enhancing behavior. In this case choice may be less
relevant to attributions of self-representativeness while the
freedom to select the details of the performance (self-referencing)
may increase perceived representativeness.
The Jones et al. (1981) study has been conceptually replicated
by Rhodewalt and Agustsdottir (1986). The replication incorporated
the latitude of acceptance, latitudes of rejection methodology of
Fazio et al. (1977) by crossing level of depression (depressed vs
nondepressed) with valence of presentation. Rhodewalt and
Agustsdottir (1986) proposed that self-enhancing presentations by
nondepressed subjects and self-deprecating presentations by
depressed subjects would fall into the latitude of acceptance and
self-esteem change would be mediated by self-perception. Self-
enhancing presentations by depressed subjects and self-deprecating
presentations by nondepressed subjects were expected to fall in the
latitudes of rejection and cognitive dissonance would account for
shifts in self-esteem following the self-presentations. Self-
referencing and choice variables were manipulated as in the Jones et
al. (1981) study with the expectation that self-perception
processes would be affected by the self-referencing manipulation
while cognitive dissonance processes would be sensitive to the
manipulation of choice.

The findings for the nondepressed subjects replicated the
Jones et al. (1981) results. Self-enhancing interview behavior
influenced subsequent ratings of self-esteem only when the behavior
was self-referencing; the choice variable had no effect. Self-
deprecating behavior was internalized only under conditions
suggesting high choice; the self-referencing variable had no effect
here. The pattern of results for the depressed subjects was the
converse of those for the nondepressed. For the depressed subjects,
the self-enhancing behavior influenced self-esteem only under high
choice conditions, suggesting the operation of dissonance processes.
Self-deprecating behavior influenced self-esteem only when the
behavior was self-referencing, implicating self-perception
Although Rhodewalt and Agustsdottir's (1986) use of
depressed vs. nondepressed subjects is a clever attempt to
incorporate the latitudes of acceptance and rejection concepts,
manipulation checks reported in a footnote suggest that the attempt
was not entirely successful. Depressed subjects, for whom a self-
enhancing presentation was expected to fall in the latitude of
rejection, actually rated self-enhancing behavior as a truer
reflection of themselves than self-deprecating behavior. The
authors attempt to explain this contradiction by suggesting that,
since subjects responded to the manipulation check before the post-
interview self-esteem measure, they may have been reducing
dissonance by claiming that their interview behavior was consistent
with their true selves. If subjects reduced their dissonance in this
way, what produced their shift in self-ratings? Also, if depressed,

self-enhancing subjects reduced their dissonance in this manner,
why didn't nondepressed, self-deprecating subjects display the same
pattern on this manipulation check? The authors failed to address
these issues and instead continued to interpret their findings as if
their manipulations had been successful.
Rhodewalt and Agustsdottir (1986) did mention that the
ambiguity surrounding their attempt to indirectly incorporate the
latitudes of rejection and acceptance "highlights the need for
subsequent research to include independent assessments of
latitudes of the self" (p. 52). Such assessments were employed by
Schlenker and Trudeau (1990) in a study which conceptually
replicated Fazio et al. (1977) and extended their findings to the
domain of self-presentationally induced self-concept change.
Schlenker and Trudeau (1990) pretested subjects to determine
their self-ratings on the trait "independence", their behavioral
consistency on the trait, and their latitudes of acceptance and
rejection. One week later, subjects participated in the experimental
session. In the first part of the session, subjects were told that the
study was an investigation of the effects of a harmless drug on
mental efficiency. They were given the drug (actually a placebo) and
half the subjects were told that it may make them feel somewhat
tense while the other half were told that they would experience no
side effects. While they were waiting for the pill to take effect,
subjects were asked to participate in a brief, presumably unrelated
project in another lab. The "other project" involved presenting
themselves in a simulated job interview patterned after Jones et al.
(1981). Subjects were randomly assigned to present themselves as

either relatively independent or dependent within either their
latitudes of acceptance or rejection.
Schlenker and Trudeau (1990) found that subjects with strong
prior self-beliefs (as indicated by their ratings of high consistency)
shifted their self-ratings in the direction of their self-presentation
only when the presentation was in their latitude of acceptance.
Presentations in the latitudes of rejection had no impact on self-
ratings. The authors suggested that these subjects used their strong
beliefs as "anchors to evaluate the self-diagnosticity of behavior"
(p. 29). Behavior that was slightly incongruent with prior beliefs
led to shifts in self-ratings while extremely incongruent behavior
was dismissed. For subjects with weak prior self-beliefs (as
indicated by ratings of low consistency), self-ratings shifted in the
direction of self-presentations in either the latitude of acceptance
or the latitudes of rejection. These subjects inferred their self-
beliefs directly from their presentational behavior, becoming more
extreme in their ratings following more extreme behavior.
In addition to the findings regarding the strength of prior self-
beliefs, Schlenker and Trudeau also found that the misattribution
manipulation affected the rationalizations subjects employed.
Subjects who presented themselves in their latitudes of rejection
claimed less personal responsibility for their behavior when they
were told that the pill would have no side effects than when they
were told that the pill would have tension-producing side effects.
Extreme presentations, which should produce aversive arousal (Fazio
et al., 1977), are rationalized by claiming less personal
responsibility, except when the arousal could be misattributed to

the side effects of the pill. When subjects' presentations fell
within their latitude of acceptance, no aversive arousal should be
generated and self-ratings of personal responsibility were not
influenced by the misattribution cue.
Schlenker and Trudeau present these findings as support for
the two-process model proposed by Fazio et al. (1977) even though
they found a misattribution effect only on measures of personal
responsibility and not on self-ratings of independence. The authors
suggest that the aversive arousal resulting from the extreme
presentations can be reduced by either shifting self-beliefs or
denying responsibility for the behavior. Specific self-beliefs
concerning one's independence may be more resistant to change than
the more general, less self-related attitudes investigated by Fazio
et al. Therefore, while subjects in the Fazio et al. study reduced the
arousal by shifting their attitudes in the direction of their
presentation, subjects in the Schlenker and Trudeau study reduced
the arousal by denying responsibility for their behavior.
Schlenker's view of the two-process approach differs
somewhat from that of Fazio et al. (1977) and Jones et al. (1981).
Rather than employing both dissonance theory and self-perception
theory to account for the obtained findings, Schlenker incorporates
both explanations within his theory of self-identification.
Self-identification involves an attempt by the individual to
construct and express desirable identity images (Schlenker, 1980,
1982, 1985, 1986, 1987). These images reflect what people believe
they could and should be, and are influenced by characteristics of
the individual, the situation, and the audience (Schlenker, 1980,

1982, 1984, 1985, 1986, 1987). Desirable identity images must
meet situationally defined criteria of believability and
beneficiality. An image must be believable in the sense that it is
consistent with available evidence, and beneficial in that the image
helps satisfy the actor's goals.
Schlenker (1986) incorporates the two-process approach by
suggesting that self-identifications may be processed in a
relatively active or passive manner. In most everyday situations,
self-identifications occur without extensive prior thought and
planning. They are based on scripts which have been used
successfully in the past, and do not represent a clear break from the
current self-concept. Schlenker (1986) suggests that if such self-
identifications influence the self-concept at all, they are likely to
do so through passive processes, such as self-perception (Bem,
1972) or the biased-scanning variant (Jones et al., 1981) and people
will rely on situational cues to determine whether the behavior is
representative of self. If these cues (eg. choice, consistency)
indicate that the behavior is representative and prior self-beliefs
are weak or not salient, then self-beliefs will shift in the direction
of the self-presentation. If the situational cues indicate that the
behavior is not representative of self or prior self-beliefs are
strong or salient, then the self-presentation will have no effect on
self-beliefs. Passive processing seems to occur in situations where
the relevant identity-images are less important and there are few
impediments to the construction of desired identities (McKillop &
Schlenker, 1988; Schlenker, 1986).

There are occasions, however, where a great deal of thought
and planning precede a self-identification. When a presentation
involves important desired identity images, or impediments to the
creation of a desired identity image are encountered, a more active
assessment process will be generated (Schlenker, 1986, 1987).
According to Schlenker (1986), "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" (p. 41). These attempts at
reconciliation may take the form of either shifts in self-beliefs,
attempts to rationalize the behavior, or both.
When prior self-beliefs are weak or uncertain and the behavior
appears to be representative of self, self-beliefs may shift in the
direction of the presentation. However, when prior self-beliefs are
strong and well defined, active processing makes salient a wealth of
information with which to counterargue the implications of the
presentation. Counterarguing may involve the use of accounts to
minimize the actor's responsibility for the presentation (excuses),
or downplay the negative implications of the presentation
(justifications) (Schlenker, 1986). If these accounts successfully
eliminate the threat to the desired identity images, then self-
beliefs will not change. If the accounts do not completely eliminate
the threat, further counterarguing cause self-beliefs to shift in the
direction opposite the self-presentation. Both Spivak and Schlenker
(1985) and McKillop and Schlenker (1987) have found evidence of
these boomerang effects following negative presentations.

Spivak and Schlenker (1985) asked subjects to present a
negative image of their social sensitivity--an ambiguous trait
descriptor which most individuals believe is important. For half of
the subjects the instructions stressed that while playing their
assigned role they should be relatively truthful (representative).
The other half of the subjects were told that they could lie if
necessary to create the assigned role (unrepresentative). When
subjects gave a negative presentation following instructions
suggesting the role was unrepresentative of self, they did not
change their self-beliefs. However, when instructions suggesting
the role was representative of self preceded a negative self-
presentation, subjects increased their ratings of social sensitivity.
Spivak and Schlenker (1985) also found evidence of active
counterarguing when instructions suggested that a positive
presentation was unrepresentative of self. These subjects'
important, desired image concerning social sensitivity was
threatened by the unrepresentative nature of the presentation, so
they compensated by increasing their private self-ratings on the
trait. When the instructions suggested that a positive presentation
was representative of self, no threat was aroused and subjects'
ratings of social sensitivity did not shift.
In a more direct test of Schlenker's (1986) predictions
regarding levels of processing, McKiliop and Schlenker (1987)
generated active or passive processing by manipulating subjects'
perceptions of the importance of the trait practicality after
subjects had performed a negative presentation on this trait. When
subjects believed that practicality was a relatively unimportant

trait, their self-ratings of practicality decreased following their
negative presentation. This finding supports Schlenker's (1986)
prediction that self-presentations involving unimportant identity
images will be passively processed leading to shifts in the direction
of the presentation. When subjects believed that practicality was a
relatively important trait, their self-ratings of practicality
increased following their negative presentation, again providing
evidence of active counterarguing leading to a reaffirmation of a
desired identity image. Further evidence of active counterarguing
was revealed by subjects use of accounts to deny the
representativeness of the negative presentation when they believed
practicality was an important trait.
Although there is no direct evidence of a boomerang effect
following positive presentations, Schlenker (1986) suggests that
their are occasions when positive presentations may threaten
desired identity images and stimulate active counterarguing. "To
the extent that excessively positive self-presentations commit
actors to standards that they doubt they will be able to maintain,
they pose a threat and arouse anxiety" (p. 52). According to
Schlenker, such presentations are accounted for by rejecting their
representativeness. While Schlenker does not mention the
possibility that downward shifts in self-beliefs may also be
employed to counterargue the identity-threatening future
expectations generated by an excessively positive self-presentation,
it is certainly consistent with his theory.
As this review demonstrates, there is currently a wealth of
data indicating that self-presentations influence both specific and

global aspects of the self-concept. In addition, it is important to
note that in most of these studies, self-concept changes take place
in the absence of any direct audience feedback. These results, along
with the findings of Myamoto and Dornbusch (1956) and Felson
(1989), support Cooley's contention that it is the imagination of
other's reactions to our appearance that is critical in self-concept
formation and change. Clearly, direct feedback from others is not
necessary to motivate self-concept change.
Despite the strength of these findings, the self-presentational
research program can be criticized as lacking external validity.
While care is taken to maintain experimental realism in most of
these studies, one critical aspect of the process of self-concept
change has been neglected. All of the research on self-
presentationally induced self-concept change has involved a single
presentation followed by an immediate assessment of the actor's
self-beliefs. However, in the "real world" people are constantly
presenting themselves to others and reevaluating their own self-
beliefs. The interaction and combination of these multiple self-
presentations may affect self-beliefs in a manner that is
completely different from that of a single self-presentation.
Although no previous research has examined the effects of
multiple presentations, several self-theorists have questioned the
stability and longevity of self-concept changes based on a single

Although the subjects in this experiment were not
necessarily affected in any profound way, the importance
of the above findings would not seem to rest on the

longevity of the produced effects. Given that the
principle is valid, it would follow that with more intense
learning experiences more deep-seated effects could be
produced. (Gergen, 1965, p. 422).

Neither of these effects is expected to be permanent, of
course, but the results do have suggestive implications
for the resultant self-esteem of those who frequently
engage in either self-enhancing or self-deprecating
social maneuvers. it is possible to imagine
fortuitously initiated self-concept changes leading to
confirmatory actions that cement those changes. (Jones
et al., 1981, p. 421).

How long will the type of modification in self-concept
illustrated by the present experiment be maintained? .
We can only suggest, as do Jones et al. (1981), that the
change in self-concept is maintained until some
subsequent event or behavior implicates other
dispositions as being self-descriptive. (Fazio, Effrein, &
Falender, 1981, p. 241).

Fazio, Effrein, and Falender (1981) have suggested that while a
single presentation may produce only temporary shifts in the self-
concept, these temporary shifts affect subsequent behavior which,
in turn, may strengthen or exaggerate the initial self-concept
change. In support of this possibility, studies by Fazio et al. (1981)
and Dlugolecki, Doherty, and Schlenker (1990) demonstrated that
self-presentations influence not only self-ratings, but subsequent
behavior as well.
Fazio et al. (1981) induced subjects to present themselves in
an introverted or extraverted manner by asking them a series of
leading questions. Following the presentation, each subject was led
to another room and asked to wait for a few minutes. Behavioral

measures of extraversion were obtained by a blind, female
confederate already seated in the room who recorded how close the
subject sat to her, whether or not the subject initiated
conversation, and the amount of time the subject spent talking. The
confederate and two blind, independent judges also evaluated
subjects' extraversion based on their behavior in the waiting room.
Finally, subjects' self-reported extraversion was measured either
before or after the behavioral measures were obtained.
Subjects who engaged in extraverted self-presentations later
rated themselves as more extraverted than subjects who had
engaged in an introverted self-presentation. In addition, the
differential presentations affected subjects' later behavior in the
waiting room situation which was separated from the experiment.
Compared to subjects who had completed an introverted
presentation, subjects who had completed an extroverted
presentation sat closer to the confederate, were more likely to
initiate conversation, spent more time talking, and were considered
to be more extraverted by both the confederate and the independent
Dlugolecki, Doherty, and Schlenker (1990), have recently
conceptually replicated Fazio et al. (1981) with regard to sociable
self-presentations. Subjects who presented themselves sociably in
a simulated job interview later rated themselves as being more
sociable, generated more examples of past sociable behavior, and
behaved more sociably in an unrelated waiting room situation
similar to that used by Fazio et al. (1981).

The results of these studies provide further support for the
cyclic process of self-concept change explicitly or implicitly
described by every self-theorist mentioned in this paper. When in
social interaction, people present images of themselves to real or
imagined audiences and receive real or imagined feedback from
these audiences. This feedback affects the actor's existing self-
concept which influences future behavior (including self-
presentations) in the same or unrelated situations. These
subsequent self-presentations again elicit real or imagined feedback
and the cycle of self-concept formation and change continues. Each
component of the cycle has been investigated in isolation, and Fazio
et al. (1981) and Dlugolecki, Doherty, and Schlenker (1990) have
explored the interaction of two of the components, but no one has
studied the process as a whole. Thus, the primary purpose of this
study is to more accurately represent the process of self-concept
change by exploring the impact of multiple presentations.

Overview and Hypotheses

This study examined the effects of initial presentational role
on subsequent self-presentations and the combined impact of initial
and subsequent self-presentations on self-ratings of sociability.
All subjects were pretested in an unrelated context to
determine self-ratings of sociability and self-esteem. In the
experimental session, subjects were asked to create a positive,
negative, or accurate impression of their sociability in a videotaped,

simulated interview session. After the taping of the first session,
three-fourths of the subjects were asked to give a second positive,
negative, or accurate presentation while the remaining one-fourth
were not asked to perform a second presentation. Finally, all
subjects completed a questionnaire packet containing the main
dependent measures and manipulation checks. The design of this
study is thus a 3 X 4 factorial with first presentational role
(sociable, unsociable, accurate) and second presentational role
(sociable, unsociable, accurate, no presentation) as between
subjects factors.

Effects of single self-presentation on self-rated sociability.
The first set of hypotheses concerned the effects of the single
sociable, unsociable, or accurate self-presentation on subsequent
self-ratings of sociability. Previous studies have examined this
issue and have yielded inconsistent results.
Both Dlugolecki and Schlenker (1987) and McKillop, Berzonski,
and Schlenker (1990) have found that subjects' self-beliefs shift in
the direction of their presentation following a face-to-face,
sociable self-presentation in a simulated job interview setting.
However, following an unsociable self-presentation McKillop et al.
(1990) found lowered self-ratings of sociability while Dlugolecki
and Schlenker observed a boomerang effect with subjects raising
their self-ratings of sociability. Slight differences in the
methodologies employed by these researchers may account for these
contradictory findings. Dlugolecki and Schlenker introduced a

cognitive activity manipulation immediately after the presentation
and before the the completion of the dependent measures. This
manipulation successfully induced subjects to recall specific
behavioral examples of their sociability. Although the manipulation
had no direct effect on subjects' self-ratings, it may have
stimulated active processing and counterarguing of the unsociable
self-presentation, thus leading to the boomerang effect. Subjects in
the study by McKillop et al. completed the dependent measures
immediately after their self-presentations, passively processed the
self-relevant information, and shifted their self-beliefs in the
direction of their self-presentations.
No previous research has investigated the impact of a single,
accurate presentation of sociability on subsequent self-beliefs, but
Gergen (1965) did find that an accurate presentation involving
general self-esteem had the same impact on self-beliefs as a
positive presentation when both were followed by reflective
reinforcement (see pp. 30-31 for complete description of study and
findings). In contrast, when an accurate self-presentation was not
followed by reflective reinforcement, no changes in self-beliefs
were observed.
Based on prior research, it was predicted that a single,
sociable presentation would lead to increases in self-rated
sociability. The predictions concerning self-rating shifts following
a single, unsociable presentation were not as clear. If the self-
presentations were actively processed, counterarguing should lead
to a boomerang effect with increases in self-rated sociability; if
the self-presentations were passively processed, they should be

internalized and self-rated sociability should decrease (Schlenker,
1986). Given that the methodology used in this study was virtually
identical to that used in the face-to-face presentation conditions of
the McKillop et al. study and did not include the cognitive activity
manipulation employed by Dlugolecki and Schlenker (1987), it was
predicted that a single, unsociable presentation would lead to
decreases in self-rated sociability. Finally, since this study did not
involve the delivery of reflective reinforcement, it was predicted
that a single, accurate presentation of sociability would produce no
shifts in self-rated sociability.
Effects of multiple self-presentations on self-rated
sociability. The second set of hypotheses concerned the combined
impact of two self-presentations on subsequent self-ratings of
sociability. As noted above, no previous research has directly
examined this issue, and the few theoretical statements mentioning
multiple presentations (Fazio et al., 1981; Gergen, 1965; Jones et
al., 1981) are rather vague and contain few specific predictions. As
evidenced by the quotes cited above (pp. 50-51), these authors do
seem to agree that repeated self-presentations create similar,
though perhaps more powerful, shifts in the self-concept as single
presentations. This prediction received some indirect support from
two studies (Haas & Maehr, 1965, Exp. 2; Kinch, 1968) that showed
that repeated, positive feedback led to greater increases in self-
evaluations than a single, positive evaluation.
Based on this theoretical and empirical evidence, it was
predicted that following repeated presentations (e.g., sociable-
sociable, unsociable-unsociable, accurate-accurate) subjects would

shift their self-ratings in the direction of their self-presentation.
Further, these shifts should be equal to, or greater than, the self-
rating shifts following single self-presentations.
The remainder of the multiple presentation predictions
involve situations in which the subjects were asked to present two,
different self-images (e.g., sociable-unsociable, accurate-sociable).
Although no existing theoretical position even mentions the effects
of multiple, different self-presentations, it is possible to logically
derive several hypotheses by combining Schlenker's predictions
concerning active versus passive processing with the empirically-
based single self-presentation predictions.
When the initial self-presentation was accurate, it was
predicted that subsequent self-ratings and presentational behavior
would remain unchanged. Thus, the effects of a second, sociable or
unsociable presentation on self-ratings were predicted to be
identical to the effects of a single sociable or unsociable
When the initial self-presentation was sociable, it was
predicted that subsequent self-ratings and presentational behavior
would become more sociable. In this situation a second, accurate
presentation would be perceived as somewhat self-derogatory, yet
within the latitude of acceptance, and sociability self-ratings would
be lower than those of subjects who performed a single, sociable
presentation. In contrast, a second, unsociable presentation would
be perceived as extremely self-derogatory and within the latitude of
rejection. This unsociable presentation would stimulate active
counterarguing and a boomerang effect with these subjects rating

themselves as more sociable than subjects who performed a single,
sociable presentation.
Finally, when the initial self-presentation was unsociable, it
was predicted that subsequent self-ratings and presentational
behavior would become less sociable. In this situation a second,
accurate presentation would be perceived as somewhat self-
enhancing, yet within the latitude of acceptance, and sociability
self-ratings would be higher than those of subjects who performed a
single, unsociable presentation. A second, sociable presentation
would be perceived as extremely self-enhancing and within the
latitude of rejection. This sociable presentation would stimulate
active processing but because there was no threat of future
disconfirmation, subjects would not counterargue the implications
of the presentation and sociability self-ratings would again be
higher than those of subjects who performed a single, unsociable
Effects of initial self-presentation on subsequent self-
presentations. The final set of hypotheses involved the effects of
the first presentational role on the performance of the second
presentation. Although no previous research has directly addressed
this issue, the studies of Fazio et al. (1981) and Dlugolecki and
Schlenker (1987) would appear to be relevant. These researchers
found that following a self-presentation, both self-beliefs and
social behavior shifted in the direction of the presentation. The
social behavior observed in these two studies involved a waiting-
room interaction which could clearly be considered a form of self-
presentation. Thus, it was predicted that subjects in this study


would present themselves more sociably following an initial,
sociable presentation than following an initial, unsociable
presentation with subjects who performed an initial, accurate
presentation falling intermediate. In addition, since the sociable
and unsociable role-playing instructions might have constrained
subjects' second presentations, it was predicted that these effects
would be most pronounced when subjects were asked to give an
accurate second presentation.


During their normal class meetings, approximately 400
introductory psychology students completed a series of
questionnaires including an extended version of the Cheek and Buss
Shyness and Sociability scale (1981), and the Rosenberg (1965)
Self-Esteem scale. From this subject pool, 144 subjects were
randomly selected, contacted via telephone, and asked to participate
in a study entitled "Simulated Job Interviews" in partial fulfillment
of their course requirement. At this point subjects were randomly
assigned to one of the 12 possible experimental conditions.

Experimental Session and Cover Story
Upon arrival at the experimental session, subjects were
greeted by the experimenter and asked to read and sign a consent
form which explained that the study concerned simulated
interviews, and that their task would be to act as a stimulus person
in an interview session. The consent form also stated that the
subject was free to discontinue participation at any time and still
receive experimental credit. After collecting the consent forms, the
experimenter explained the purpose of the interview, gave the
subjects instructions on the role they were to enact and the tactics
they could use to create the desired impression (see Appendix A).

Subjects were asked to play the role of a student applying for a
research assistant's position during a simulated job interview
session. They were told that the session would be videotaped and
that these tapes would be shown to graduate students who were
studying personality-based job interview techniques. In order to
make the situation more realistic and involving, subjects were told
that the students who would be viewing their videotape would
believe that they were an actual candidate for a research assistant's
position and would be required to make a judgement about the
subject's adequacy for the position. The experimenter explained that
the interview would contain a few general questions, but that the
main focus of the interview would be on sociability, a trait being
studied by the graduate class. Subjects were told that the graduate
students would believe that they were observing an actual
interview, and would not be told of this deception until two weeks
later when their interview techniques were assessed in class. This
cover story is similar to one used by Jones et al. (1981, Exp. 3), and
is designed to justify the necessity of the interview, mask the
actual purpose of the experiment, and maximize the importance of
the interview situation for the subject. The purpose of announcing
the delay in debriefing the graduate students was to insure that the
subject's behavior could not be "taken back," thus diluting the impact
of the interview (Davis and Jones, 1960).

Independent Variables

First Presentational Role and Choice Manipulations
Next, the experimenter explained that in order to provide the
graduate student trainees with a wide range of interview behaviors,
subjects would be asked to adopt different goals and strategies
when answering the interview questions. Subjects in the positive
presentation condition were asked to present themselves in the
"best possible light" and to try to get the interviewer to form a good
impression of their sociability. If assigned to the negative
presentation condition, subjects were asked to present themselves
in a "negative light" and to try to get the interviewer to form a poor
impression of their sociability. Finally, subjects in the accurate
presentation condition were asked to present themselves honestly
and accurately and to try to get the interviewer to form an accurate
impression of their sociability.
After the experimenter finished reading the instructions, he or
she added: "We realize that some people might object to responding
in a way other than they really feel. You don't mind helping us out, do
you?" This statement and the choice manipulation contained in the
consent form were designed to emphasize that the subjects had free
choice about whether or not to participate in any portion of this
study. High choice has been shown to maximize internalization
effects in previous studies (Fazio et al., 1977; Jones et al., 1981).
Next, the experimenter reminded the subjects that they must
continue to play their assigned role as long as the videotape was
running, turned on the camera, and began the interview session.

Second Presentational Role Manipulation
Following the first interview session, some subjects were
asked to repeat their initial presentation. They were told that since
the graduate class was rather large, and we wanted to be sure that
each student had an opportunity to see their interview, we would
like to have at least two copies of their videotape. They were asked
if they would mind going through the interview one more time and
were reminded of their initial role instructions.
Subjects in conditions requiring them to present themselves
differently in the second presentation were told that in order to
expose the graduate students to a full range of interview behaviors,
we would like them to go through the interview one more time and
play a different role. They were given the new role instructions and
asked if they would mind going through the interview one more time.
In all conditions involving a second presentation, subjects
were told not to worry about remembering how they answered the
questions in the first interview since each student in the class
would view only one of their tapes and any inconsistencies between
their two interviews would not be noticed.
Some subjects were not asked to give a second presentation.
These subjects completed the dependent measures packet
immediately after the first interview. No mention of a second
presentation was made and there was no time delay. These first
presentation only conditions served as baselines for comparisons of
the effects of single versus multiple presentations.

Interview Session
In order to insure that each subject received the same
treatment in the interview situation and to minimize the potential
effects of each experimenter's personality, the experimenters were
instructed and carefully trained to maintain a neutral demeanor,
neither approving nor disapproving of the subject's presentation, and
the interview was based on a script (see Appendix B) which was
identical for each subject. The interview began with a greeting and
exchange of first names. This was followed by a series of
hypothetical situations, adjective descriptors, and self-ratings,
ostensibly designed to assess sociability. The experimenter
recorded the subject's responses to each item on an answer sheet
and, when the interview was completed, turned off the video
recorder and led the subject to a different location in the room.

Dependent Measures and Manipulation Checks
Finally, the subjects were given a packet of questionnaires
containing the dependent measures and manipulation checks (see
Appendix C) with instructions to disregard the role they were asked
to play in the interview and answer the questions honestly. To
convince subjects of the anonymity of their responses to this
questionnaire and reduce the likelihood that their responses
represented an attempt to please the experimenter, they were told
not to sign their name or place any identifying marks on the
questionnaire. Subjects were also be told that their data would only
be seen by an experimental assistant who would code it anonymously

and that the experimenter would receive feedback in the form of
group data only, and would not know how each individual responded.
The questionnaire began with three questions asking the
subject to rate, on 7-point scales, how enjoyable, interesting, and
worthwhile the simulated interview project was. On the next page
was the posttest extended version of the Shyness and Sociability
scale (Cheek & Buss, 1981).
The Shyness and Sociability scale developed by Cheek and Buss
(1981) contains only five items assessing sociability and nine
assessing shyness (the main focus of the authors' work). Because
scales consisting of so few items often have poor internal
reliability (Nunnally, 1978), seven additional items were added to
the original scale. Analyzing the pretest data for the 144 subjects
who participated in this experiment, it was found that each of the
new items was significantly (g.<.0001) correlated with the original
sociability scale (L's = .38 .68). Factor analysis of the revised
scale revealed one factor accounting for 44% of the variance with
each item loading substantially (.49 .83) on this factor. In
addition, Cronbach's Alpha for the revised scale = .88 as compared to
Alpha = .70 for the original scale (Cheek & Buss, 1981). Given the
improvement in internal reliability and the fact that previous
research (McKillop, Berzonski, & Schlenker, 1990) has yielded
virtually identical patterns of results using the revised and original
scales, this study utilized the revised scale as the primary measure
of sociability.
The next page of the questionnaire contained five, 15-point
scales assessing the subject's current mood (e.g., tense--relaxed,

sad--happy), and the 10 item Rosenberg Self-Esteem scale. The
following section consisted of nine 15-point scales dealing with the
subject's behavior and feelings during the interviews. On these
items and the remaining manipulation checks subjects were asked to
respond first concerning their first interview, and then concerning
their second interview. The next two pages of the questionnaire
contained checks of the manipulations of choice and role, as well as
items designed to assess subjects' perceived responsibility for their
behavior during the interviews and the representativeness of their
interview behavior.
Prior research suggests that one way for subjects to reduce
the arousal produced by a negative presentation is to attack the
audience for the presentation (Jones, Gergen, & Davis, 1962;
McKillop & Schlenker, 1988). To examine this effect, several items
assessing subjects' perceptions of the graduate students who would
be viewing their videotapes were included on the last few pages of
the questionnaire.
The final page of the questionnaire contained a behavioral
recall task. Subjects were asked to think of some incidents that
they felt were relevant to their sociability; these could be occasions
where they felt their sociability had been especially good, average,
or poor. Next, they were asked to briefly describe five such
incidents and rate each incident on a 7-point scale ranging from
extremely sociable (1), to extremely unsociable (7).
After collecting the questionnaire packet, the experimenter
carefully questioned each subject for suspicion regarding the true
nature of the study and debriefed them by following a standardized


debriefing script (see Appendix D). Subjects were asked to keep the
actual purpose of the study a secret until the end of the semester
due to the deception necessary, given their experimental credit, and
thanked for participating.


Unless otherwise stated, all major analyses were conducted
via two-way analyses of variance with first presentational role
(sociable, unsociable, accurate) and second presentational role
(sociable, unsociable, accurate, no presentation) as between subject
factors. Given that this study contained a detailed set of specific
hypotheses and the omnibus F-test is more appropriate in the
absence of specific hypotheses (Keppel, 1982), planned comparisons
were conducted without regard for the significance or
nonsignificance of the omnibus E-test.

Manipulation Checks
This study required that we manipulate subject's self-
presentations in the first and second interview as well as their
perceptions of choice to perform these behaviors. The following
analyses were conducted to ascertain the effectiveness of these

The effectiveness of the presentational role manipulations
was assessed through analysis of subjects' responses to the three
key sections of the interviews. Subjects' responses to the bipolar

adjective ratings, self-ratings, and percentile ranking were
converted to z-scores and averaged to provide an index of self-
presentational behavior.
First presentational role. Analysis of subjects' behavior in the
first interview revealed a main effect of first presentational role,
E(2,131)=143.58, .<.0001. Subjects instructed to present
themselves unsociably gave less sociable responses in the interview
session than subjects instructed to present themselves accurately,
F(1,131)=128.72, .<.0001, who, in turn, responded less sociably than
subjects instructed to present themselves sociably, E(1,131)=27.96,
.<.0001 (Ms=-1.039, .205, and .774, respectively).
Second presentational role. The second presentational role
instructions produced the same pattern of behavior in the second
interview. The main effect of second presentational role,
E(2,97)=252.69, g<.0001, was again clarified by planned comparisons
indicating that subjects instructed to present themselves
unsociably gave less sociable responses in the interview session
than subjects instructed to present themselves accurately,
E(1,97)=225.96, ,<.0001, who, in turn, responded less sociably than
subjects instructed to present themselves sociably, E(1,97)=51.04,
0<.0001 (Ms=-1.149, .240, and .902, respectively).

The choice manipulation was designed to ensure that all
subjects perceived a high degree of choice concerning whether or not
to participate in the interview sessions. The efficacy of this
manipulation was supported by subjects' responses to a question

asking them to rate the degree of choice they felt they had in
whether or not to participate. On 9-point scales with 1 labeled "no
choice at all" and 9 labeled "totally free choice", the mean ratings
were 7.64 for the first interview and 7.73 for the second

Dependent Measures

Shifts in Sociability
Analysis of pretest sociability self-ratings revealed no
significant main effects or interactions. Shifts in self-rated
sociability were measured by subtracting pretest sociability scores
(M=46.76, SD=8.81) from posttest sociability scores (M=48.86,
.SD=7.59). Thus, a positive change score indicates an increase in
sociability following the self-presentations while a negative score
represents a decrease in sociability. The overall mean change score
was 2.10 with a standard deviation of 6.53.
Effects of single self-presentations on self-rated sociablility.
The effects of a single self-presentation on changes in sociability
were examined via planned comparisons of change scores in the
conditions requiring no second presentations. A simple main effect
of first presentational role when no second presentation was given,
E(2,131)=3.56, .<.05, was clarified by pairwise comparisons which
indicated that subjects who presented themselves unsociably
showed a significantly greater increase in sociability than subjects
who presented themselves accurately, E(1,131)=5.98, 2<.05, or
sociably, E(1,131)=4.70, ,<.05 (see Table 1 for means). The change

scores in the accurate, no second presentation and positive, no
second presentation conditions did not differ significantly.

Sociability Change Scores

First Presentation

Second Presentation Negative Positive Accurate

Negative 2.27 3.75 3.92

Positive 0.27a 1.42 0.25

Accurate 1.08b -0.25 2.75

No Presentation 5.92abcd 0.85c 0.08d

Note. Positive sociability change scores indicate an increase in
sociability from pretest to posttest. Means with a common single-
letter subscript differ by at least p.<.05.

This boomerang effect following a single, negative
presentation replicates the findings of Dlugolecki and Schlenker
(1987) and seems to contradict the findings of McKillop, Berzonski,
and Schlenker (1990). Possible interpretations of these unexpected
results will be explored in the discussion section.
Effects of multiple self-presentations on self-rated
sociability. The hypotheses concerning multiple self-presentations
were tested by planned comparisons of change scores in each of the

multiple presentation conditions with scores in the appropriate
single presentation condition. This strategy resulted in the
following 9 single-df comparisons: accurate, accurate vs. accurate,
no presentation; accurate, negative vs. accurate, no presentation;
accurate, positive vs. accurate, no presentation; negative, accurate
vs. negative, no presentation; negative, negative vs. negative, no
presentation; negative, positive vs. negative, no presentation;
positive, accurate vs. positive, no presentation; positive, negative
vs. positive, no presentation; positive, positive vs. positive, no
Significant differences were found only for the negative,
accurate vs. negative, no presentation, E(1,131)=4.11, .O<.05, and the
negative, positive vs. negative, no presentation, E(1,131)=5.36,
9.<.05, comparisons (see Table 1 for means). These results indicate
that the increase in sociability following a single, negative
presentation was eliminated when subjects performed a second
presentation that was either accurate or positive.

The hypotheses concerning the effects of initial self-
presentations on subsequent self-presentations were tested by
assessing the simple main effect of first presentational role on
subjects' behavior in the second interview (as measured by the z-
score index described above). The simple main effect of first
presentational role was significant only when subjects were
instructed to give an accurate second presentation, E(2,97)=3.59,
.<.05. Further, planned comparisons revealed that subjects

instructed to give an accurate second presentation behaved more
sociably following an initial, unsociable, E(1,97)=5.92, .<.05, or
accurate, E(1,97)=4.81, .<.05, presentation than following an initial
positive presentation (see Table 2 for means). Additional planned
comparisons indicated that this effect occurred only when subjects
were instructed to give an accurate second presentation.


Behavior in the Second Interview

First Presentation

Second Presentation Negative Positive Accurate

Negative -1.05 -1.18 -1.20

Positive .79 .90 1.00

Accurate .38a .OOab .34b

Note. Scores represent the average z-score for subjects' responses
to the three portions of the interview. All pairwise comparisons for
the simple main effect of the second presentation are significant
(g's<.05). In all other cases, means with a common single-letter
subscript differ by at least p.<.05.

Although no specific predictions were made regarding
subjects' perceptions of the representativeness of their interview
behavior, analysis of these perceptions was undertaken due to their
important mediational role in prior research (McKillop & Schlenker,

1987; Schlenker & Trudeau, 1990). In this study, the degree to
which a presentation was perceived to be representative of self was
assessed by three, bipolar scales asking subjects to rate how
deceptive vs. truthful, fake vs. genuine, and unrepresentative vs.
representative they had felt during each of their interviews. These
three measures were highly correlated (rs>.92) and yielded identical
patterns, so they were averaged to simplify discussion of the
A significant main effect of second presentational role on
subjects' perceptions of the representativeness of their behavior in
the second interview, E(2,97)=120.23, 0<.0001, was further
elucidated by simple effects tests which showed that subjects who
performed a negative second presentation perceived their behavior
to be less representative than subjects who performed positive,
E(1,97)=90.81, .<.0001, or accurate, E(1,97)=236.16, .<.0001, second
presentations (Ms=3.43, 9.52, and 13.30, respectively). An accurate
second presentation was also perceived to be more representative
than a positive presentation, E(1,97)=33.26, S<.0001.
The first presentational role also had an effect on subjects'
perceptions of their behavior in the second interview, E(2,97)=7.46,
p<.001; an effect that is best understood in the context of a
significant first x second presentational role interaction,
F(4,97)=3.64, p<.01 (see Table 3 for means). Simple effects tests
showed that subjects who performed a negative first presentation
perceived a positive second presentation to be more representative
than subjects who performed a positive, E(1,97)=4.50, g<.05, or
accurate, E(1,97)=26.59, 9<.0001, first presentation. The positive

second presentation was also perceived to be less representative
following an initial, accurate presentation than following an initial,
positive presentation, E(1,97)=9.63, .<.005.


Perceptions of the Representativeness of the Second Self-

First Presentation

Second Presentation Negative Positive Accurate

Negative 4.27 2.97 3.11

Positive *12.33a 9.94a 6.53a

Accurate *13.72 12.92 13.25

Note. Higher numbers indicate greater ratings of representativeness
on a 15-point scale. All pairwise comparisons for the simple main
effect of the second presentation are significant (.'s<.05), except as
noted by asterisks. In all other cases, means with a common single-
letter subscript differ by at least ,.<.05.

Logically, the nature of the first presentational role should not
have any bearing on the representativeness of the second
presentation. These findings suggest a perceptual or memory bias
allowing subjects to reduce the threat of an initial, negative
presentation by exaggerating the representativeness of the positive
presentation that followed.

Perceptions of the Audience
The threatening nature of the unsociable presentation was
further supported by analyses of subjects' perceptions of how much
the graduate students (who would be viewing their videotapes)
would like them if they had an opportunity to interact at a later
date. A significant main effect of first presentational role,
E(2,131)=7.81, .<.001, was qualified by a marginally significant
first x second presentational role interaction, E(6,131)=2.11, .=.057
(see Table 4).

Subjects' Perceptions of How Much the Graduate Students Would
Like Them

First Presentation

Second Presentation Negative Positive Accurate

Negative 4.00 4.08de 3.33

Positive 3.91 2.58e 3.25

Accurate 3.58c 3.33 2.67

No Presentation 5.00abc 2.61ad 2.17b

OVERALLMEAN 4.13fg 3.14f 2.85g

Note. Lower numbers indicate greater perceived liking. Means with
a common single-letter subscript differ by at least .<.05.

Simple effects tests revealed a simple main effect of first
presentational role only. when no second presentation was
performed, E(2,131)=10.66, .<.0001. Subjects believed the graduate
students would like them less following a single unsociable
presentation than following a single accurate, F(1,131)=18.31,
.<.0001, or sociable, E(1,131)=13.49, ,<.001, presentation.

Changes in self-esteem were examined to determine whether
the observed changes in sociability self-ratings could have been
caused by shifts in global self-evaluations.
Shifts in self-esteem were measured by subtracting pretest
self-esteem scores (M=42.47, SD=7.81) from posttest self-esteem
scores (M=43.49, SD=5.33). Thus, a positive change score indicates
an increase in self-esteem following the self-presentations while a
negative score represents a decrease in self-esteem. The overall
mean change score was 1.16 with a standard deviation of 6.26.
Analysis of pretest self-esteem scores revealed only a
marginally significant main effect of second presentational
role, F(3,127)=2.46, .<.07. Although none of the pairwise
comparisons reached significance, inspection of the means indicated
that subjects in the accurate and negative second self-presentation
conditions had lower self-esteem scores than subjects in the no
second presentation and positive second presentation conditions
(Ms=41.09, 40.71, 44.31, and 44.44, respectively). This finding may
invalidate or at least cast suspicion upon any significant changes in
self-esteem as a function of second presentational role.

Analysis of self-esteem change scores yielded only a
significant main effect of second self-presentational role,
E(3,126)=3.00, .<.05, with subjects who gave an accurate second
presentation showing a greater increase in self-esteem than
subjects who gave a positive, E(1,126)=7.09, .<.01, or no second
presentation, F(1,126)=5.53, .<.05 (see Table 5 for means).


Self-Esteem Change Scores

First Presentation

Second Presentation Negative Positive Accurate MEAN

Negative .91 4.18 -.08 1.62

Positive -2.00 .25 -.18 -.62b

Accurate 1.90 2.67 4.42a 3.06bc

No Presentation -.33 .08 -.33a -.19c

Note. Positive self-esteem change scores indicate an increase in
self-esteem from pretest to posttest. Means with a common single-
letter subscript differ by at least g<.05.

The series of contrasts planned to test the hypotheses
concerning sociability shifts were also applied to the self-esteem
change scores resulting in only one significant difference. Subjects
who gave accurate first and second presentations showed greater
increases in self-esteem than subjects who gave an accurate first

presentation and no second presentation, E(1,126)=4.17, .<.05 (see
Table 5 for means).
It is possible that the opportunity to present themselves
accurately and consistently caused subjects to feel good about
themselves in general, however, as mentioned above, very little
confidence can be placed in any interpretation of these findings due
to the observed pattern of pretest self-esteem differences. In any
case, the pattern of self-esteem changes does not resemble the
observed changes in sociability indicating that the sociability shifts
were not caused by changes in global self-evaluations.

Further evidence that the shifts in sociability were not the
result of global affective shifts was found in subjects' responses to
five items assessing their postpresentational mood. Analyses of
these five items yielded no significant main effects or interactions
(ps>.10). In addition, the series of contrasts used to test the
hypotheses concerning sociability shifts were also applied to these
items resulting in no significant differences (.s>.05).

Behavioral Recall Task
Finally, analyses of the behavioral recall measure yielded no
significant main effects or interactions (.s>.70). It was expected
that subjects' behavioral recall would display the same pattern as
their shifts in sociability; however, the planned comparisons
revealed no significant differences (.s>.10).


The purpose of this study was to examine the effects of initial
self-presentations on subsequent. self-presentations and the
combined impact of multiple self-presentations on self-beliefs. In
the absence of any theoretical statements or empirical evidence
bearing directly on this issue, hypotheses were logically derived
from several related lines of research. The results generally failed
to support these hypotheses.

Effects of Single Self-Presentations on Self-Rated Sociability
It was predicted that a single, sociable self-presentation
would lead to increases in self-rated sociability; a single,
unsociable self-presentation would lead to decreases in self-rated
sociability; and a single, accurate self-presentation would produce
no change in self-rated sociability. As expected, subjects who
presented themselves accurately exhibited no change in sociability.
However, the results in the sociable and unsociable presentation
conditions did not support the predictions. Subjects who presented
themselves sociably reported no change in sociability while
subjects who presented themselves unsociably increased their self-
rated sociability. This boomerang effect following a negative self-
presentation is certainly not unprecedented (McKillop & Schlenker,

1987; Dlugolecki & Schlenker, 1987), but had not been predicted in
the current study due to the findings of McKillop et al. (1990) who
reported presentationally consistent shifts in sociability following
a face-to-face job interview simulation.
Why did McKillop et al. find a decrease in sociability following
an unsociable presentation while this study found an increase? The
most reasonable explanation involves the major difference in the
methodologies of the two studies. The simulated job interviews in
the McKillop et al. study were conducted by a single interviewer in a
face-to-face interaction with the subject. In this study, the
interviews were conducted by the experimenter and videotaped for
later evaluation by an entire class of interviewer trainees. At least
two aspects of this methodological difference may be responsible
for the discrepant findings. First, subjects in the present study
were told that their videotapes would be seen by a number of
students in the interviewing class while subjects in the McKillop et
al. study confronted only one graduate student interviewer. This
difference in the size of the audience may have made subjects feel
more evaluation conscious and threatened by the implications of a
negative presentation. This threat may also have been enhanced by
the relative permanence of the videotaped presentation. By
intensifying the threat to a desired identity image, the larger
audience may have stimulated active counterarguing (Schlenker,
1986, 1987) resulting in the observed boomerang effect.
In addition to the enlarged audience for the self-presentation,
the second important aspect of the videotape methodology which
may account for the discrepant results is the use of the video

camera. Several authors have suggested that video cameras may be
used to manipulate self-focused attention (Davis & Brock, 1975;
Duval & Wicklund, 1972; Geller & Shaver, 1976). More recently,
Carver & Scheier (1981) have claimed that cameras actually
manipulate public self-awareness and have generated some
empirical support for this claim (Scheier & Carver, 1980).
Individuals who are made publicly self-aware tend to be acutely
conscious of the aspects of themselves they are displaying to others
and of the impressions they are making on others. Fenigstein (1979)
found that persons high in public self-consciousness (the
dispositional equivalent of the state of public self-awareness) are
more sensitive to interpersonal rejection than those low in public
self-consciousness. The exposure of subjects to the video camera in
this experiment may have induced public self-awareness and an
increased concern with the creation of desired identity images. The
self-aware state may have magnified the threat of the unsociable
presentation leading subjects to counterargue and subsequently
increase their ratings of sociability.
Given that previous research has used face-to-face
presentations (e.g., Schlenker & Trudeau, 1990), videotaped
presentations (e.g., Kelly, McKillop, & Neimeyer, in press) and a
combination of the two (Jones et al., 1981; Rhodewalt &
Agustsdottir, 1986), these speculations concerning possible
differences between the two methodologies need to be explored in
future research.

Effects of Multiple Self-Presentations on Self-Rated Sociability
It is difficult to assess the degree to which the hypotheses
concerning the effects of multiple presentations were supported by
the data due to the unexpected results of the single self-
presentation manipulations.
It was predicted that repeated presentations (sociable-
sociable, unsociable-unsociable, accurate-accurate) would lead to
sociability shifts in the same direction as single presentations.
Although the single presentation shifts were not in the expected
direction, the shifts following repeated presentations did track the
single presentation results and none of the planned comparisons
differed significantly. None of the predictions involving
comparisons between multiple and single self-presentations were
confirmed, but this may have been largely due to the single self-
presentation results.
The boomerang effect resulting in substantial increases in
sociability following a single, negative self-presentation was
eliminated when subjects performed a second presentation that was
either positive or accurate. The opportunity to reaffirm desired
identity images by portraying oneself more positively in a second
presentation may have reduced subjects' need to counterargue the
implications of the initial, negative self-presentation. Having
repaired the threatened image through the second presentation,
there was no longer any need to bolster self-ratings of sociability.

The Role of Self-Esteem
Upshaw and Yates (1968) have proposed that the successful
completion of a task can lead to increased self-esteem even if the
task is to present oneself negatively. If this is true, then the
boomerang effect in the present study may simply reflect an overall
rise in self-esteem following the successful completion of an
unsociable presentation. For this reason it is important to note that
the pattern of changes in self-esteem was not similar to the pattern
of sociability changes and could not be responsible for the observed
boomerang effect.

Effects of Initial Self-Presentations on Subsequent Self-
The results concerning the effects of initial self-
presentations on subsequent self-presentations also failed to
confirm the hypotheses. Based on the studies of Fazio et al. (1981)
and Dlugolecki and Schlenker (1987), it was predicted that subjects
in this study would present themselves more sociably following an
initial, sociable presentation than following an initial, unsociable
presentation with subjects who performed an initial, accurate
presentation falling intermediate.
Initial presentational role did influence the performance of
the second presentation, but only in the accurate second
presentation conditions. This finding was not surprising in that it
was anticipated that the sociable and unsociable role-playing
instructions might overwhelm any effect of the initial presentation.
However, the pattern of the observed results was surprising as

subjects instructed to give an accurate second presentation behaved
more sociably following initial accurate or unsociable presentations
than following an initial, sociable presentation.
These findings may reflect an attempt by subjects who had
presented themselves unsociably in the initial interview to reaffirm
a desired identity image by presenting themselves sociably when
given the opportunity to do so later. In contrast, subjects who had
initially presented themselves sociably may have felt no need to
exaggerate their sociability in the second interview, or may have
even downplayed their sociability to avoid committing themselves
to an overly positive self-image which would have been difficult to
maintain in future interactions.
The inconsistencies between these findings and those of Fazio
et al. (1981) and Dlugolecki and Schlenker (1987) may be due to
methodological differences between these studies.
First, the behavior of subjects in the present study following
an initial, negative presentation was the opposite of the behavior of
the subjects in the Fazio et al. research. This may be due to the
subtle manipulation of self-presentational behavior employed by
Fazio et al. In asking leading questions to induce introverted or
extroverted presentations, it is likely that Fazio et al. were able to
manipulate subjects' behavior without making subjects aware that
their presentations were incongruent with their initial self-beliefs.
In this situation, self-beliefs shifted via a relatively passive, self-
perception process and these shifts were reflected in the
subsequent waiting room behavior. Since subjects were not aware
of any threat to a desired identity image and made no attempt to

counterargue the implications of their presentations, there was no
need to reaffirm their self-beliefs. In this study, self-
presentations were manipulated in a very direct manner which
forced subjects to become aware of the incongruency between their
presentations and their initial self-beliefs. The awareness of this
incongruency was strongest following unsociable presentations,
which subjects reported to be extremely unrepresentative of their
true selves. In this situation, the threat to a desired identity image
was obvious and subjects actively counterargued the implications of
their unsociable presentations. This counterarguing carried over to
their second presentation as they attempted to reaffirm the
threatened images.
The behavior of subjects in this study following an initial,
positive presentation was inconsistent with the findings of Fazio et
al. (1981) and Dlugolecki and Schlenker (1987). The reason for this
discrepancy is not as clear but may be related to differences
between the studies with regard to the audience for the second
presentations. Both Fazio et al. and Dlugolecki and Schlenker
assessed subsequent behavior by observing subjects' interactions
with a confederate in a waiting room situation. The confederate,
who comprised the audience for the subjects' second presentations,
clearly had no knowledge of subjects' initial presentation. In the
present experiment, although subjects' were told that each taped
presentation would be shown to different members of the
interviewing class, the experimenter was present during both
interviews. It is possible that subjects in this study were
concerned with publicly committing themselves to an overly

positive self-image in front of the experimenter, a person with
whom they expected to interact after the interview sessions. This
may account for the less sociable second presentation observed in
the present study while subjects in the two studies employing
clearly independent audiences felt free to aggrandize their self-
images in both presentational contexts.

The Resiliency of the Self
Taken together, the results of this study suggest a self-
corrective tendency which may render the self-concept more
resistant to change than has been previously suggested. Several
converging pieces of evidence indicate that when confronted with an
identity-threatening, unsociable self-presentation, subjects will
use any means available to counterargue the implications of this
First, when subjects were asked to perform a single,
unsociable self-presentation, they later inflated their private self-
ratings of sociability. Greenberg and Pyszczynski (1985) observed a
similar effect following public, failure feedback and labeled it
"compensatory self-inflation." The present study extends their
findings by demonstrating that direct, audience feedback is not
necessary to stimulate compensatory behavior. As Cooley (1902)
suggested, even in the absence of an observable response, we
imagine the judgement of the other. In addition, it should be noted
that the compensatory behavior was not directed toward the
external audience since self-ratings of sociability were obtained
under completely anonymous conditions; the inflated self-ratings

seem to have been intended to appease some private, internal
Second, when subjects were given the opportunity to present
themselves accurately following the threatening, unsociable self-
presentation, they bolstered their self-image by presenting
themselves more sociably than they had following an initial,
sociable presentation. Further evidence of reaffirmation can be seen
as subjects who presented themselves sociably following an initial,
unsociable self-presentation exaggerated the representativeness of
their sociable presentations. Both of these tactics appear to have
been successful in reducing the threat of the initial, unsociable
self-presentation as subjects who performed a second, accurate or
sociable presentation did not display the boomerang effect found
after a single, unsociable self-presentation. Steele and Liu (1983)
reported a similar finding following a standard forced-compliance
procedure. Subjects given the opportunity to affirm an important,
self-relevant value after writing a counterattitudinal essay
exhibited no attitude change while those who affirmed a value not
relevant to their self-concepts did show dissonance-reducing
attitude change. In the present study, subjects were able to affirm
their sociability by performing sociable second presentations and
perceiving these presentations to be extremely representative-of-
self. Having affirmed the self and removed the threat, subjects
were no longer motivated to counterargue and displayed no shifts in


The present study may demonstrate the sensitivity of
presentational research to minute variations in experimental
procedures. In future research exploring self-presentationally
induced self-concept change, the various methodologies described in
this paper (e.g., live audience vs. video, multiple audiences vs.
single, subtle vs. overt presentational induction) must be compared
to determine their unique impact on shifts in self-beliefs. Until
this information is available, any attempt to compare or generalize
from the findings of the methodologically diverse studies in this
area will be extremely difficult.

The importance of the well-documented shifts in self-beliefs
following single self-presentations seems to rest on the assumption
that these shifts will influence future self-beliefs. Gergen (1965),
Jones et al. (1981), and Fazio et al. (1981) have all proposed that
self-concept shifts resulting from single self-presentations are
likely to be maintained or even enhanced as a result of subsequent
self-presentations; but none of these theorists has made any
attempt to test these assumptions. The results of the present study
indicate that simplistic and global theoretical statements
concerning the impact of multiple self-presentations may be
inaccurate and misleading. While it is possible that some shifts
may be maintained or enhanced following subsequent presentations,


it is equally possible that initial shifts may be eliminated and self-
beliefs may return to their initial level. The conditions under which
each of these outcomes may obtain should be specified in future
theoretical statements.


Hello. Before we begin I would like you to read the informed
consent form I've given you and if you have no questions, sign it and
we'll get started. (Give subject 2 copies of informed consent)

Cover Story
The project in which you will be participating involves
simulated interview situations. One purpose of the project is to
provide practical experience for graduate students who are enrolled
in a class on Personality Assessment Techniques. As part of their
training in personality assessment, these graduate students must
have experience in different types of interview situations. At this
point in their training the students are learning personality-based
job interview techniques. What we would like you to do is play the
role of a student applying for an Undergraduate Research Assistant's
Interviewers are often asked to determine whether or not an
applicant for a position has the type of personality best suited for
the particular job. If employees with a certain trait are desired, the
interview may focus on that specific trait. The interview we will
be conducting today will focus on the trait "sociability," which is
one of the traits being studied by the graduate class. In the

interview, you will be asked a number of personality-type questions
regarding how sociable you are. The interviewer will begin by
asking a few general questions about yourself such as your hobbies,
favorite books, and intended major. However, the main focus of the
interview will be on your sociability. The object of the interview is
not to assess job-related skills, but personal characteristics.
To provide a uniform experience for the students in the class,
we will make a videotape of your interview. I will ask the questions
and your responses will be taped and later shown to the graduate
students for analysis and discussion. The students will watch the
tapes individually and form independent opinions. Later, their
evaluations will be discussed in the class. All of the questions used
in the interview were developed by the students themselves as part
of the training exercise. To make the situation more realistic and
involving for the students in the class, they will be told that you are
an actual candidate for a research assistant's position, and will be
asked to make a judgement about your adequacy for the position.
That is, the students who watch the videotape will believe that you
are actually applying for a job as an Undergraduate Research
Assistant, and that their evaluations will influence whether or not
you get the job. They will not be told of this deception until after
their evaluations have been reviewed and graded in class. Thus, we
must ask you to be careful, while we are taping the interview, not to
reveal in any way that this briefing has occurred. Do you think you
can help us out?
In everyday interactions, of course, people sometimes describe
themselves totally accurately, while other times they may slightly

or significantly misrepresent information about their attitudes,
preferences, skills and experiences. This range is particularly
noticeable during job interviews. In order to capture the wide range
of ways in which people describe themselves we are asking
different subjects in this experiment to adopt different goals and
strategies for responding during the interview. By including this
range, we can better assess the ways interviewers use information
when forming impressions about the people they interview.

First Presentational Role Manipulations
Sociable and Unsociable
In your case, we would like you to present yourself in the
"worst/best possible light" and try to get the interviewer to
form a bad/good impression about your sociability. Feel free to
exaggerate your weaknesses/strengths and underplay your
strengths/weaknesses, tailoring what you say to make a bad/good
impression. Your goal should be to make as negative/positive an
impression of your sociability as possible while still seeming
credible and sincere.
We have found that the best way to present yourself in the
"worst/best possible light" is to really get yourself mentally into
the role. Think of occasions when you really felt extremely
unsociable/sociable. Now, thinking of yourself in this way, simply
answer the questions so that your feelings are conveyed in both the
mood and content of your answers. We realize that some people
might object to responding in a way other than they really feel. Do
you mind helping us out?