When I Die, I Feel Small:
Electronic Game Characters and the Social Self
Daniel G. McDonald
Department of Communication
315 Kennedy Hall
Ithaca, NY 14850
Running Head: Game Characters and the Social Self
Correspondence should be addressed to the first author at the address above or
at [log in to unmask]
Paper submitted for consideration for presentation to the Theory and Methodology
the Association for Education in Journalism and Mass Communication, 1996
Electronic Game Characters and the Social Self
Theoretical perspectives on social orientations toward personality and
of the self are undergoing a resurgence (Rosenberg, 1988; Schwartz, 1990;
Moore, 1993). This paper explores children's use of mediated characters as role
development of their self concept and personality. Using the perspective of the
we examine data on children's use of video and computer games and perceptions of
game characters as comparisons to their own personality and other individual
Electronic Game Characters and the Social Self
Although eclipsed by the rise of behaviorism in the middle of the
perspectives on social orientations toward personality and the development of
the self are
undergoing a resurgence (Rosenberg, 1988; Schwartz, 1990; Thornton and Moore,
Indeed, much of the work in communication effects research since the 1960s
some aspect of social learning theory, and much of it includes a reference to
Soukup (1993) reviews the literature on the impact of media and suggests
number of research traditions in the field of communication point toward the
"social self as a
locus for the impact of media." Citing work by Mead (1934), Park and Burgess
members of the Frankfurt School (such as Adorno and Lowenthal), Duncan (1985),
Bandura (1977), Soukup suggests that all of these traditions point toward the
idea that the
media must be making an impact on the social self, going so far as to suggest
that nearly the
entire effects tradition in mass communication research presumes that
influence the social self.
The purpose of this paper is to explore children's use of mediated
characters as role
models for development of their self concept and personality. Using the
perspective of the
"social self," we examine data on children's use of video and computer games and
of electronic game characters as comparisons to their own personality and other
The Social Self
The phenomenon of the social self has been a concern for psychologists
communication theorists for at least 100 years (Cooley, 1902; James, 1890; Mead,
James notes that a person "has as many social selves as there are individuals
him." Rosenberg (1988) suggests that James was referring to both the idea that
individuals hold different perceptions of a person, and that individuals have
presentations and enactments for each person they meet. It is the latter idea
which forms the
core of theories of social personality (Rosenberg, 1988).
Baldwin (1897) went even further than James in describing a theory of the
social self), by making all aspects of the self and personality a social and
Baldwin suggested that a personality is formed primarily through imitation of
others, and said
that he could not see "...how the personality... can be expressed in anything
but social terms"
Mead (1934) suggested that the individual becomes conscious of others in
manner in which he becomes conscious of himself - through communication. Mead
the process of identification of the other with the self, accomplished through
taking the role of
the other. He suggests that it is through taking the role of the other that an
becomes integrated into society. Mead also suggests that it is through the
media that contemporary people enter into roles of the other.
Although the dominance of behaviorism in mid-century tended to obscure
the influence of early theory on the concept of the social self, the area of
(Schneider, Hastorf and Ellsworth, 1979), developed fairly sophisticated
theories and methods
for studying perceptions of others through impression formation, person memory,
Reviewing this literature, Rosenberg (1988) suggests that significant
components of the
stable views one has of oneself and others are the characteristics that a person
the more enduring aspects of the self and the other. Rosenberg suggests that
traits such as
physical characteristics, personality, attitudes and competencies, as well as
experiences associated with self and other, are stored as potential components
of the self and
are eventually manifest in descriptions of one's self and others.
Rosenberg (1988) uses a "set-theoretic" analysis to explain how such
contents may be
manifest in descriptions of others of importance to the self. Set-theoretic
analyses can be
developed Using free-response techniques, diaries, logs or novels to examine the
development of personality. Rosenberg (1988) examines Thomas Wolfe's
autobiography, Look Homeward, Angel for evidence of the development of his
personality over his life span.
Evidence for Social Development of Personality
Braun (1987) has suggested that individual personality progresses in a
fashion, with the social self and the personal self alternating in a kind of
spiral of dominance
over individual personality. Her study of the development of individuation in 6
to 9-year old
children suggests that these two subsystems of the self develop in a very
complex fashion and
over a number of years. She suggests that individuals' tendency to describe
terms of differences (rather than similarities) between oneself and others is a
indicating a striving to maintain a sense of self identity and individuality.
While Braun (1987) found lower distinctiveness in self-perceptions among
children (8-9-year olds) in her sample, Kardacz (1985) found greater
older participants (15-16-year olds) compared to younger (12-year olds) in his
suggests that these findings indicate differing dominance in cycles of social
A number of studies have found evidence that an individual's
affected by perception of others. Recent studies such as those by Thornton and
find contrast effects in that both men's and women's perceptions of their own
influenced by an apparent discrepancy between oneself and same sex others. Both
(negative) and downward (positive) contrast effects have been found, directed
appearance and social competencies. This research also suggests that situations
physical attractiveness becomes a salient dimension heighten states of public
consciousness, leading individuals to evaluate themselves in comparison to the
relevant standards, leading to shifts in evaluations of oneself. Study
experimental) have shown these shifts to be fairly situation-specific and
short-term. Few, if
any, long-term or nonexperimental studies have been attempted.
Citing a number of theoretical positions from Turner (1956), Lewin
(1952), Mead (1934), and others, Higgins (1988) suggests that by incorporating
between "own" and "other" as a feature for self/state presentations, different
conditions can be related to self-state conditions. Higgins suggests that we
are motivated to
reach a condition where our self-concept matches our personally relevant self
Mediated Role Models
Soukup (1993) notes that one of the most compelling aspects of the media
presentation of the human image. Feminist perspectives on the role of images
that these images teach, shape and distort social roles (Tuchman, Daniels and
Some time ago, Bronfenbrenner (1970) noted that children are turning more and
peers and television for their socialization, due to increases in a lack of
parental attention. A
number of empirical studies lend some support to these contentions. Sprafkin
(1978) review a number of studies and provide additional data which suggest
sex-role attitudes are influenced by televised sex-role portrayals.
Schreiber (1979) found that the age and gender of tv characters affected
positively people responded to the character, with young and middle-aged viewers
most favorably to characters in the same age groups as themselves, while older
responded least favorably to characters of the same age group as themselves.
An early study of the tv program "All in the Family (Tate and Surlin,
1976) found that
persons whose lifestyles and psychological characteristics were similar to the
character Archie Bunker were more likely to perceive the character and program
and to identify with that character.
Aside from the evidence that viewers identify with characters, several
found modeling effects of that identification. Leckenby (1977)found college
of distinguishing between tv characters in their degree of dogmatism, and found
dogmatic people tended to see less differences between themselves and tv
characters in their
level of dogmatism, suggesting that character dogmatism reinforces viewer
dogmatism. In a
followup study, Leckenby (1978) found that subjects low on dogmatism, the lower
perceived dogmatism of a character, the greater the perceived commonality, the
character is liked, the more the character is agreed with, and the greater the
influence of the
character on the viewer (measured through opinion change in Leckenby's study).
Meyer (1973) found that favorite tv characters were perceived as behaving
consistently with children's perceptions of their own behavior. Meyer concluded
provided primarily socially acceptable role models for children. Donohue (1975)
the extent to which tv provided antisocial behavioral models for children.
While finding that
the majority of role models were prosocial, a number were antisocial. In
addition, most of
the children reported that when confronted with an ethically ambiguous
situation, they would
behave in a manner similar to what they thought their favorite character would
Ellis, Streeter and Engelbrecht (1983) developed the idea of of role
taking of viewers
and television characters from the perspective of symbolic interactionism. They
viewers take the role of salient television personalities during viewing and
contexts, and may modify their behavior to conform to the imaginary evaluations
television characters. Indeed, the concept of "parasocial interaction"
developed around the
idea that viewers took the roles of tv characters and personalities defined by
(Horton and Wohl, 1956; Horton and Strauss, 1957).
The extent of research on identification with television characters is a
reflection of the
ubiquity of the medium since the 1950s. While a number of early studies on
focused on the extent of their use and the amount of time spent in use (e.g.,
Lambirth and Wilson, 1983; McClure and Mears, 1984), few studies have examined
computer game characters as role models (Selnow, 1993). Three characteristics
games suggest the importance of investigating identification with game
characters and that the
effects of these interactions my be different from the effects of interacting
characters: First, since the early 1980s, electronic games have become
in nature. Results of Selnow's (1993) preliminary investigation suggests that
enables players to become a "member" of the game unlike the opportunities
television, where viewers have no control over ongoing events.
Second, a very large percentage of electronic games involve violent
violent action as the primary activity of the game (Braun & Giroux, 1989;
Unlike most television content, in which there is a plot which unfolds before
the viewer, and
violence is incidental to the story, for many electronic games, the object is to
"kill" as many
of the enemy as possible.
Third, some players seem to perceive electronic games as a kind of
companion (Selnow, 1993), and their interactions with the games as social
this perspective, the games "impart lessons about people" (Selnow, 1993). Nass
1993) have shown that subjects, with fairly simple provocation, can begin to
between "self" and an electronic "other," and use certain social rules typical
interpersonal communication in interaction with the computer.
The present study examines whether or not children identify with and
electronic game characters and use personality and physical cues from those
characters as a
comparison base for themselves. If the concept of the "social self" can be
electronic game characters, we expect children to identify and compare
themselves to those
characters in terms of personality, physical attractiveness and general physical
We adapt the Ellis, et al. (1983) propositions related to tv characters and
role-taking to the
electronic game context and bring empirical data to three propositions:
H1. Players will evaluate their own characteristics and behavior from
of the electronic characters. In the present study, this will be evaluated by a
the type of characteristic (physical or mental/personality) of descriptions of
self and electronic
other. If players are evaluating their selves from the perspective of the
electronic others, we
should see similar dimensions of evaluation (e.g.,similar dimensions of both
characteristics and affective evaluation described for both self and other).
H2. Role-playing will occur in non-gaming as well as gaming contexts.
proposition will be evaluated by response to self-report questions related to
the extent to
which children "pretend" they are the character. We expect role-playing to be
associated with grade in school, so that younger children will be more likely to
are the characters.
H3. According to Braun's (1988) suggestion that the social self and the
alternate in a kind of spiral of dominance over individual personality, we
expect to find
differences by grade (nonlinear) in the extent to which identification takes
place and in the
impact of identification on self-descriptions.
H4. Electronic character role-playing results in modification of
perceptions to conform
to the perceived evaluation of the electronic other. This hypothesis will be
evaluated by the
extent to which descriptions of an ideal self match descriptions of electronic
closely than they match self-descriptions.
H5. In accordance with Higgins (1987), we test the proposition that
between self-descriptions and the description of an ideal self should result in
the inclusion of
dejection-related emotions in verbal descriptions of the self and in feelings
However, if the electronic character provides a model similar to the ideal self,
should hold for discrepancies between descriptions of emotions after a game is
over and for
descriptions of the electronic characters during the game.
In the spring of 1994, twenty-one elementary and high school teachers
enrolled in a
research methods class as part of their master's program in a private 5-year
college in New
York State's rural "southern tier" area participated in data collection for the
teacher distributed a questionnaire related to electronic game playing and
descriptions of self
and electronic characters. Teachers of students in younger grades conducted
interviews with the children. In grade levels in which the students could be
expected to read
the materials for themselves, teachers simply passed out the questionnaire
during class time.
No identifying information was recorded except grade and gender.
Open-ended descriptions of self and other were coded first as written or
the interview. The first two descriptions given were included in the coding.
After a listing of
all descriptions was obtained, coders reclassified the descriptions into a two
One dimension was the hedonic tone of the descriptor (positive, negative
while the other dimension was the type of description: behavior (e.g., runs
characteristic (e.g., short), or personality (e.g., shy).
For increased generalizability, three coders classified the descriptors
Original classification reliability for the questions was .91; all disagreements
were settled by
majority (2/3) category classification.
Data was collected from 303 students. Table 1 provides an overview of
distribution of background variables related to electronic game playing. As is
the table, more than 90% of students play video or computer games, with about
owning the game and about half owning a computer. Nearly all of those who play
games play them at home, and nearly three-fourths play games alone. A similar
(about three-fourths) indicate that there are rules in their home related to the
Table 1 about here
Hypothesis 1. The first hypothesis was that game players would evaluate
characteristics and behavior from the perspective of the electronic characters.
classification of descriptions of self by descriptions of the electronic game
some support for the hypothesis. There were greater numbers of responses
diagonal of the cross-classification of descriptions of self and electronic
characters than would
be expected if there were no relationship (Likelihood Ratio 2=8.46, 4 d.f., p =
is also a significant relationship between the dimensions used in the
description of feelings
the children felt when thinking of themselves and the electronic character
2=11.891, 4 d.f., p = .018).
Hypothesis 2. The second hypothesis was that role-playing would occur in
gaming as well as gaming contexts, with a negative relationship between
grade in school. Table 2 provides results for this test of the hypothesis, with
an analysis of
variance suggesting strong relationships between grade, sex of child, and
Table 2 about here
Additional illustration of the relationship is presented in Figure 1, in
playing is plotted against grade in school for both boys and girls in the study.
As evident in
the figure, girls report less role-playing than boys, but both boys and girls
role-playing throughout their school grades. The exception is for the extent of
for boys in 12th grade, but this was an anomaly of the sample, as only three
boys in the
sample were in 12th grade, and it may not be typical of what we would find if a
sample had been available.
Figure 1 about here
Hypothesis 3. Based on Braun's (1987) suggestion that the social self
and the personal
self alternate in a kind of spiral of dominance over individual personality, we
expected to find
nonlinear differences by grade in the extent to which identification takes place
and in the
impact of identification on self-descriptions. A simple ANOVA divided linear
components suggest that the nonlinear results are of marginal significance (F =
1.661, 11 d.f.,
p < .085), while the linear component is much stronger (F = 24.261, p < .000).
that, although the nonlinear component is apparently part of the relationship,
it is much
weaker than the linear component of the relationship.
Hypothesis 4. The fourth hypothesis centered on modification of
conform to the perceived evaluation of the electronic other, and was evaluated
the extent to which descriptions of an ideal self match descriptions of
more closely than they match self-descriptions.
Tables 3 and 4 provide information related to the test of this
hypothesis. A simple
measure of discrepancy in which all non-matching descriptions were coded as "0"
descriptions matching on the description dimension were coded as "1" was
subject to multiple regression analysis. Table 3 provides an indication of the
results of the
analysis for discrepancy between self-descriptions and descriptions of the
"Ideal Self." As
indicated in Table 3, no variable in the equation was a significant predictor of
between self and the ideal self.
Tables 3 and 4 about here
A somewhat different picture emerges for discrepancy between self and the
character, however, as role-playing is of marginal significance (p < .091), and
between role playing and sex of child is fairly strong (r2=.051, p < .035). The
weight for role-playing is negative, suggesting that the more role-playing that
occurs, the less
discrepancy is seen between self and the electronic character.
Hypothesis 5. In accordance with Higgins (1987),the fifth hypothesis was
proposition that discrepancies between self-descriptions and the description of
an ideal self
should result in the inclusion of dejection-related emotions in verbal
descriptions and in
expression of feelings one has after a game is over. If the electronic game
serving as a substitute or model for the player's perception of the ideal self,
should be more evident in discrepancy between self and electronic character than
and the ideal self. A dummy variable coded 1 for dejection-related emotion
words and 0 for
non-dejection emotion words (see Higgins, 1987) used in the descriptions was
subjected to a
paired-difference t-test. Table 5 presents the mean differences for both
feelings about the character and descriptions of feelings the player has after
the game is over.
As evident from the table, both comparisons were statistically significant,
for the hypothesis.
This paper has been necessarily exploratory, tapping a somewhat
the potential relationship between self-other comparison and electronic
results are somewhat tentative, there is considerable evidence here that
children identify quite
closely with electronic characters of all sorts, and that these identifications
important implications for their emotional well-being as well as for the
Evidence here suggests that identification with electronic characters is
an aspect of the
development of the individual -- as grade level increased, we found that
children were less
likely to report identification with the characters. However, it is important
to note that, even
though less identification is reported by the children as they mature, we found
no evidence to
suggest that they were less likely to use similar dimensions in describing
themselves and the
electronic characters. In other words, there may be less overt identification,
but the process
of making self-other comparisons, and using electronic characters as the "other"
throughout the developmental stages examined here.
We found several implications of this identification. We found that the
playing that occurs, the more likely the players are to use similar criteria in
themselves and the electronic characters. In addition, we found that a likely
outcome of this
self-other comparison is for the self to come up short -- the player is not as
tall, fast, funny,
smart or strong as the electronic character -- and the result appears to be
Higgins' (1987) hypothesis that the self will feel somewhat dejected as a result
There are still many unanswered questions related to self-other
comparisons and the
function of mediated characters. While identification with mediated characters
is a common
and much-discussed phenomenon of modern life, little research has been unable to
consequences of that identification. This research takes one step in that
examining the implications for emotional experience and the development of
it is clear that many of the experiences described here are a normal part of
development and do not necessarily imply negative effects on personality
development. If, as
Cooley, James, Baldwin and others have suggested, we store small components of
personalities in developing our own, it is important, and probably necessary,
for children to
store both good and bad aspects of personality so that they may assess the
effects of those
personalities on their environments. While it may be troublesome to some that
those bits and pieces of personality are electronic, programmed and planned,
developed from interactions in a real world, one might see them as simply
personalities - developed by a programmer or game designer who has had
experience in the
real world and communicates those experiences through the game characters. More
is needed to elucidate some of the implications of the role of electronic
characters and the
social self in the development of personality and human experience.
Distributions of Game Playing Variables *
Play Video or Computer Game
Ownership of Computer
Ownership of Video game
Parents' Rules on Games
Play Games at Home
Play Games Alone
*Total n=303; Table percentages reflect percent answering questions.
ANOVA Results for Role Playing by Grade and Sex
Sum of Squares
Regression on Discrepancy between Self and Ideal Self
Role Playing by Sex
Role Playing by
Hierarchical Regression on Discrepancy between Self and Electronic Character
Role Playing by Sex
Role Playing by
*p < .05
Results of t-test testing for inclusion of dejection-related emotions in verbal
Descriptions of Characters
Self-Ideal Self Comparison
Descriptions of Post-Game Emotions
Self-Ideal Self Comparison
Bandura, A. (1977). Social learning theory. Englewood Cliffs, NJ: Prentice-Hall.
Braun, B. (1988). Development of self-knowledge and the process of
Psychological Bulletin, 19, 257-262.
Bronfenbrenner, U. (1970). Two worlds of childhood. New York: Russel Sage
Cooley, C. H. (1918). Social organization: A study of the larger mind. New York:
Donohue, T. R. (1977). Favorite TV characters as behavioral models for the
disturbed. Journal of Broadcasting, 21, 333-345.
Duncan, H. D. (1985). Communication and social order (C. Wilder, Introduction).
Brunswick: Transaction Books.
Ellis, G. J., Streeter, S. K., & Engelbrecht, J. D. (1983). Television
characters as significant
others and the process of vicarious role taking. Journal of Family
Issues, 4, 367-384.
Gibb, G. D., Bailey, J. R., Lambirth, T., & Wilson, W. (1983). Personality
between high and low electronic video game users. Journal of Psychology,
Higgins, E. T. (1987). Self-discrepancy: A theory relating self and affect.
Review, 94, 319-340.
Horton, D., & Wohl, R. R. (1956). Mass communication and para-social
Observations on intimacy at a distance. Psychiatry, 19, 215-229.
Horton, D., & Strauss, A. (1957). Interaction in audience-participation shows.
Journal of Sociology, 62, 579-587.
James, W. (1948). Psychology. New York: World. (Original work published 1890)
Kardacz, A. (1985). Categories of self-description and perception of
self-features in others.
Unpublished M.A. thesis. Faculty of Psychology, University of Warsaw.
Kelly, H. H. (1952). Two functions of reference groups. In G. E. Swanson, T. M.
& E. L. Hartley (Eds.), Reading in social psychology (2nd ed., pp.
York: Holt, Rinehart & Winston.
Leckenby, J. D. (1977). Attribution of dogmatism to TV characters. Journalism
Lewin, K. (1935). A dynamic theory of personality. New York: McGraw-Hill.
McClure, R. F., & Mears, G. (1984). Video game players: Personality
demographic variables. Psychological Reports, 55, 271-276.
Mead, G. H. (1934). Mind, self and society. Chicago: University of Chicago
Meyer, T. (1973). Television's Behavioral models and children's perceptions.
Broadcasting Review, 7, 25-33.
Park, R. E., & Burgess, E. W. (1924). Introduction to the science of sociology
Chicago: University of Chicago Press.
Provenzo, E. F. (1991) Video Kids: Making sense of Nintendo. Cambridge, MA:
Rosenberg, S. (1988). Self and others: Studies in social personality and
Advances in Experimental Social Psychology, 21, 57-95.
Schreiber, E. S. (1979). The effects of sex and age on the perceptions of TV
inter-age comparison. Journal of Broadcasting, 23, 81-93.
Schwartz, T. J. (1990). Methodological advances in the study of self-concept.
at the 70th annual convention of the Western Psychological Association,
Selnow, G. (1984). The fall and rise of video games. Journal of Popular Culture,
Soukup, P. A. (1993). Media echoes in the development of the social self. Paper
the 79th annual meeting of the Speech Communication Association, Miami.
Tate, E. D., & Surlin, S. H. (1976). Agreement with opinionated TV characters
cultures. Journalism Quarterly, 53, 199-210.
Thornton, B., & Moore, S. (1993). Physical attractiveness contrast effect:
self-esteem and evaluations of the social self. Personality and social
Bulletin, 19, 474-480.
Tuchman, G., Daniels, A. K., & Benet, J. (1978). Hearth and home: Images of
women in the
mass media. New York: Oxford University Press.
Turner, R. H. (1956). Role-taking, role standpoint, and reference group
Journal of Sociology, 61, 321.