Professional Documents
Culture Documents
JULIEM. D L J C KMICHAEL
,~ A. HOGG,AND DEBORAH
J. TERRY
University of Queen-dand
Bri.ybane. A ustraliu
People typically perceive negative media content (e.g., violence) to have more impact on
others than on themselves (a third-person effect). To examine the perceived effects ofpos-
itive content (e.g.. public-service advertisements) and the moderating role of social identi-
ties, we examined students’ perceptions of the impact of AIDS advertisements on self.
students (in-group), nonstudents (out-group), and people in general. Perceived self-other
differences varied with the salience of student identity. Low identifiers displayed the typi-
cal third-person effect. whereas high identifiers were more willing to acknowledge impact
on themselves and the student in-group. Further, when intluence was normatively accrpt-
able within the in-group. high identifiers perceived self and students (us) as n70re intlu-
enced than nonstudents (them). The theoretical and practical implications of this reversal
in third-person perceptions are discussed.
1879
given widespread public concern regarding the effects of a mass medium such as
television on a gullible and susceptible mass audience (Smith, 1986). people may
be strongly motivated to deny the personal impact of media messages. Moreover,
the ego-defensive motivation to deny personal persuasiblity may be particularly
pronounced when the media message promotes negative or socially undesirable
outcomes (e.g., violence. pornographytthe content that has traditionally domi-
nated discussion of, and research on. third-person effects (Perloff, 1993). In these
contexts, people may be motivated to view themselves both as less persuasible to
media influence and as less susceptible to negative outcomes (Gunther & Mundy,
1993).
As we have argued elsewhere (Duck, Terry, & Hogg. 1995), less is hnown
about the perceived impact of positive messages that advocate personally benefi-
cial and socially desirable outcomes (e.g., public-service advertisements for
health and safety issues) and, theoretically, the situation in such contexts i s more
complex. I n the interests of self-presentation, people may be motivated to per-
ceive themselves as relatively responsive to intelligent and beneficial mes-
sages-that is, they may display a reverse third-person effect in which self is
perceived as more, rather than less. influenced than others. It is also possible that
a general reluctance to acknowledge personal persuasibility w-orks against, and
perhaps outweighs, any such desire. Indeed, research comparing the perceived
effects of different types of persuasive content has shown that third-person per-
ceptions are more pronounced and more pervasive with respect to negative con-
tent. but it remains unclear whether third-person perceptions are evidenced in any
systematic way for positive content (Duck & Mullin, 1995; Gunther & Mundy,
1993; Gunther & Thorson, 1992; lnnes & Zeitz. 1988). For instance, o u r own
evidence (Duck, Terry, et a]., 1995) indicates that people may perceive them-
selves as more or less influenced than others by public-service advertiseincnts for
safer sex and protection against AIDSIHIV, depending on the strength of their
personal beliefs in the benefits of such advertising and on their perceptions of the
quality of the advertisements under consideration. Given the ambiguity of find-
ings in the context of positive message content, as well as the social importance
of research on the perceived impact of public-service advertisements. further
research of this type is required.
Drawing on insights gained from more traditional social influence topics.
recent research has reemphasized the role of social groups in the persuasion pro-
cess (e.g.. Mackie. Gastardo-Conaco, & Skelly, 1992: Mackie, Worth, & Asuncion,
1990; McGarty, Haslam. Hutchinson, & Turner, 1994; van Knippenberg &
Wilke, 1992; Wilder, 1991). Like Turner (1991), these researchers have stressed
that the process of persuasion involves more than just information processing,
because “information processing has a social-normative aspect . . . [and] . . . is
affected by social relationships between self and others within varying social
contexts, embodied in one’s social identity and group memberships” (pp. 172-
SOCIAL IDENTITY AND PERCEIVED PERSUASION 1881
I 73). More specifically, recognizing that in-group attitudes and opinions are par-
ticularly influential in determining how we construct and define our social reality
(e.g., Festinger, 1953; Heider, 1958), researchers have concentrated on specify-
ing the processes that mediate the persuasive advantage of messages representing
in-group opinions. According to current theories of group membership, such as
social identity theory (Hogg & Abrams, 1988; Tajfel & Turner, 1979) and self-
categorization theory (Turner, 1982, 1985; Turner, Hogg, Oakes, Reicher, &
Wetherell, 1987). the validity of information is psychologically established by in-
group norms, and the basic influence process is one where the normative position
of people categorized as similar to oneself is accepted as subjectively valid
(Turner, 199 1).
Recent research has also reemphasized the role of social groups in the third-
person effect (i.e., in perceptinns of the persuasive impact of messages on self
and other). The third-person effect has typically been demonstrated by compar-
ing people’s perceptions of media influence on self with their perceptions of
media influence on a single generalized other (e.g., “voters in general,” Davison,
1983; “other viewers.” Lasorsa, 1989; “most Americans,” Rucinski & Salmon,
1990). However, there is a growing recognition that the magnitude of perceived
self-other differences depends, at least in part, on who the comparison others are.
For instance, researchers have noted that perceived effects on others are niagni-
fied as the definition of others becomes progressively more broad, vague, or
socially distant (e.g.. Cohen et al., 1988; Duck & Mullin, 1995; Gibbon &
Durkin, 1995: Gunther, 1991). In our own research (Duck. Hogg, & Terry.
1995), we have drawn attention to the neglected “us-them’’ or in-group-out-
group distinction implicit in Davison’s ( 1983) original specification of the third-
person effect. Drawing on social identity theory and self-categorization theory,
we have reasoned that, as with the process of persuasion per se, perceived self-
other differences in persuasion reflect the social psychological relationships
between self and others within varying social contexts.
According to social identity and self-categorization theories. an important
part of the self-concept is derived from membership in social groups. Although
people often view themselves as individual or unique persons, different from all
others, there are many social contexts in which people view themselves predomi-
nantly as group members with a shared social identity. In these contexts, a pro-
cess of self-categorization “depersonalizes” the perception of self and other in
terms of relevant group prototypes (norms, stereotypes). People view themselves
and others as group members rather than as individuals (“I” becomes “we,” and
“me vs. you” becomes “us vs. them”), and they accentuate prototypical similari-
ties between self and in-group members and prototypical differences between in-
group and out-group members. Moreover, since social identities have important
self-evaluative consequences (e.g., Hogg & Abrams, 1990; Turner, 1982), when
social identity is salient, group members tend to evaluate self-categories
1882 DUCK ET AL.
positively and to make intergroup comparisons that favor the in-group (e.g.,
Brewer. 1979; Brewer & Kramer. 1985; Tajfel & Turner, 1979). Contextual fac-
tors influence the degree to which people identify themselves as group members
rather than individuals (Hogg, 1992)-the stronger the social identification, the
more pronounced the tendencies toward group-based perceptual accentuation and
in-group favoritism.
From this perspective, we have proposed that the third-person effect should
occur to differing degrees according to the categorization of self and other into
in-group and out-group categories and to the accentuation of similarities within
and differences between those categories. To the extent that comparison others
are judged as out-group members. they should be contrasted to the perceiver’s
identity, evaluated. and represented unfavorably-as relatively vulnerable to
(negative) media influence. In contrast, to the extent that comparison others are
judged as in-group members. they should be assimilated to the perceiver’s iden-
tity, evaluated, and represented favorably, like the self-as relatively invulner-
able to influence. Moreover, the extent of self-in-group assimilation and in-
group-out-group contrast displayed should vary according to the perceiver’s
level of identification with the relevant social category.
We found some support for these predictions in our research on perceptions
of media campaign impact during the 1993 Australian federal election (Duck,
Hogg, et al., 1995). Although respondents who identified with one of the two
major political parties perceived themselves to be less influenced than others
(political in-group, political out-group, voters in general) by campaign content in
general, perceived self-other differences were more pronounced with respect to
political out-group than political in-group others. Moreover. this pattern of in-
group-out-group contrast was exaggerated in respondents who identified
strongly with their preferred political party. Our results also indicated that,
although politically identified respondents perceived self and in-group others to
be less influenced than others by campaign material that explicitly favored the
“wrong” side of the political fence. they perceived self and in-group others t o be
more influenced than political out-group others by campaign content that esplic-
itly favored the “right” side (in a reversal of the typical third-person effect).
Taken together, these results suggest that perceived self-other differences in
persuasibility depend not only on how we perceive others in terms of relevant
group memberships (i.e., as in-group or out-group others), but also on the extent
to which persuasibility on a specific content dimension is considered normative
or desirable for the salient in-group. When it is socially acceptable to resist
persuasion, people should see themselves and members of their in-group as
highly resistant and others as less so (a third-person effect). But, when i t is
socially acceptable to think of oneself as influenced, people may see themselves
and members of their in-group as quite yielding and others as less so (a reverse
third-person effect). In contrast to other theoretical accounts which emphasize
SOCIAL IDENTITY AND PERCEIVED PERSUASION 1883
the role of social desirability in third-person perceptions (e g., Gunther & Mundy,
I993), this proposal emphasizes that the social desirability of influence on partic-
ular content dimensions is defined in terms of salient group norms and, hence,
that perceived self-other differences in media influence reflect social-normative
processes.
This study is designed to explore further these issues. We aim to examine uni-
versity students’ perceptions of the impact of televised AIDS advertisements not
only on self and on people in general (cf. Duck, Terry, et al., 1995), but also on
students in general (in-group others) and on nonstudents (out-group others).
AIDS is one of the most pressing current health problems and is an issue of par-
ticular relevance to university students. Research shows that students engage in
sexual practices that lead to increased exposure to HIV infection (e.g., Crawford,
Turtle, & Kippax, 1990; Rosenthal. Hall, & Moore. 1992; Struckman-Johnson,
Gilliland, Struckman-Johnson, & North, 1990). and many of the HIV campaigns
promoting safer sex have explicitly targeted people in this age bracket. The
issues of AIDS and safer sex are frequently discussed in student newspapers. and
student unions promote safer sex through the distribution of free condoms and
information about AIDS/HIV during university orientation weeks. Hence, we
reason that students who identify strongly with the student community would
perceive it as socially acceptable to acknowledge the impact of AIDS advertise-
ments. However, given a general reluctance to admit persuasibility, especially to
mass-media messages, we reason that students who do not identify with the stu-
dent community would perceive it as personally undesirable to acknowledge
such influence. This presents an interesting scenario where identification with a
relevant reference group might serve to reverse typical third-person perceptions.
Specifically, based on self-categorization theory, we predict that perceptions
of media influence on self and on others (students, nonstudents, and people in
general) would vary with the perceiver’s level of identification with the student
in-group; that is, with the salience of student identity. We expect that low identifi-
ers would perceive themselves as less influenced than others (students, nonstu-
dents, and people in general) in a typical third-person effect, whereas high
identifiers would perceive self and students (in-group members) as more influ-
enced than nonstudents (out-group members; Hypothesis 1). We expect that high
identifiers, compared with low identifiers, would perceive more influence on self
and students (Hypothesis 2), less difference between the level of influence on self
and on students (Hypothesis 3), and more difference between the level of
influence on students and nonstudents (Hypothesis 4). These predictions follow
from the assumptions that the salience of self-categorization accentuates the per-
ceived similarity between self and in-group others (an sssimilation effect); accen-
tuates the perceived differences between in-group and out-group (a contrast
effect); and motivates respondents to judge their self-category positively (posi-
tive distinctiveness).
1884 DUCK ET AL.
We also reason that the effects of social identity would be clearer the stronger
the student norm for acknowledgment of personal influence and, given wide vari-
ation in the quality and character of AIDS advertisements, that this norm would
be stronger for some advertisements than for others. Hence, we predict that our
hypotheses would be supported more clearly when we focus on the perceived
impact of a subset of AIDS advertisements which students identify as being
desirable to be influenced by. This improvement in support of our predictions
would be unique to this subset of advertisements and not hold for other subsets of
advertisements judged as, for instance. the better or more informative advertise-
ments.
Method
Participants
Materials
Procedure
)In studies of third-person perceptions, respondents are typically asked to estimate the extent to
which certain types of media content will influence their attitudes and the attitudes of others.
Although it is possible that the word “influence” encourages third-person perceptions. alternative
approaches to measuring perceived media impact (e.g., by asking respondents to estimate the percent-
age o f neutral viewers who would become more negative to Israel after viewing news coverage of the
Beirut massacre; Vallone et al., 1985) have also revealed third-person effects.
1886 DUCK ET AL.
scale ranging from 1 (not much at all) to 9 (ven, much). Previous research (e.g.,
Hogg & Hains, 1996; Terry & Hogg, 1996) has shown that this item is a good
indicator of identification. In factor analyses of multiple items measuring identi-
fication, this single item (with a factor loading of approximately 0.80) usually
defines the factor.
In a subsequent session, 1 week later, 40 students from the original sample
were also asked the extent to which university students in general would think
that it is good to be influenced by each of the 11 advertisements, on a %point
scale ranging from 1 (not much at alI) to 9 (very much). The task was introduced
by suggesting that although students may believe that AIDS commercials in
general are a good thing, they may think that some AIDS advertisements are
more relevant and appropriate than are others. Participants were asked to recall
each of the 1 1 advertisements in turn (prompted by short titles, e.g., “Grim
Reaper,” “Shotgun”) and to indicate the extent to which university students in
general would think that it is good to be influenced by each advertisement. We
were not interested in memory or recall per se. Rather, we reasoned that the col-
lection of these ratings of normative acceptability at a subsequent session. rather
than in the original testing session, would serve to reduce possible demand char-
acteristics. The stimulus advertisements were vivid, familiar, and limited in num-
ber, and students reported no difficulty in recalling and evaluating them in this
manner.
lMean responses to the same item obtained from an independent sample o f 140 first-year psy-
chology students (55 males and 85 females; Magc= 18.63 years. SD = 1.03) indicated a similar rank
order: Advertisement 10 (Testimonials, .\I= 7.32). 8 (Beds II. .&.I=
7.28). 6 (Beds I. M = 7.25). 9 (Sex,
.U = 6.91), 5 (Playing footsie, A4 = 6.81), 1 (Grim Reaper, M = 6.72), 2 (Shotgun, M = 6.63). I 1
(Experts. M = 6.37). 7 (Everybody. M = 5.71 ), 3 (Information Brochures, M = 5.22), and 4 (Beach
Party, M = 5.00). Using a median split. Advertisements 10. 8, 6, 5, and 9 were again classified as the
advertisements for which normative acceptance of influence was strongest-and this selection held
for both male and female participants. despite some significant gender differences in the ratings of
particular advertisements.
SOCIAL IDENTITY AND PERCEIVED PERSUASION 1887
Using a median split on scores on the item measuring identification with uni-
versity students ( M = 5.98, SD = 2.37, Mdn = 6), participants were classified as
low identifiers ( n = 30) or high identifiers ( n = 28). As intended, low and high
identifiers differed significantly in their strength of identification with the student
in-group (Ms = 4.13 and 7.96), ((56)= - 10.48, p < .001. Preliminary checks also
indicated that identification was correlated with age, r(S7) = -.47, p < .OO 1, such
that younger students identified more strongly with the student community. Thus,
in the following analyses, steps were taken to demonstrate that any effects of group
identification on perceptions of media influence were not confounded with age.
Results
Using influence ratings averaged across all 11 commercials, the effects of iden-
tification (low, high) on perceptions of influence on self, students, nonstudents,
and people in general (target) were examined in a 2 x 4 mixed ANOVA with
1888 DUCK ET AL
8
6'01
5.8
-2
2
B
5.6
.-
i
$1
5.4
'E
5.2
5.0
LOW High
Identification
Target 19 Self
69 Students
Non-suknts
Pboplc
repeated measures on the second factor.5 There was no significant main effect for
identification, F( I , 56) = 1.4 I , ns, or for target, F ( 3 , 168) = 2.03, ns. but as
expected, there was a significant Identification x Target interaction, F(3, 168) =
3.06, p < .05 (Figure I ). Results of planned contrasts indicate that low identifiers
displayed the typical third-person effect perceiving themselves ( M = 5.17) as less
influenced than the target others ( M s = 5.50 students, 5.66 nonstudents, and 5.57
people in general; p < .05; Hypothesis 1 ). By contrast, high identifiers perceived
self and others to be equally influenced by AIDS advertising, and there was no evi-
dence to indicate that high identifiers perceived self ( M = 5.7 l ) and students ( M =
5.66) as more influenced than nonstudents ( M = 5.63). As expected (Hypothesis
2), high identifiers were more willing than low identifiers to acknowledge the
influence of AIDS advertising on themselves (Ms = 5.7 1 and 5.17, p < .05) and, to
a lesser extent, on the student in-group ( M s = 5.66 and 5.50), although this latter
difference was not significant. Moreover, consistent with Hypothesis 3, high iden-
tifiers also differentiated less clearly between self and in-group others (Ms = 5.71
and 5.66, ns) than did low identifiers ( M s = 5.17 and 5.50, p < .05). However,
5Preliminary analyses including gender of the subject as a between-subject variable indicated a
significant main effect for sex, F( I , 53) = 1 0 . 9 4 . ~< .01. but no interactions with identification or tar-
get. Overall. female respondents perceived the AlDSiHlV advertisements to have more influence
than did male respondents (M.v = 5.74 and 5.19).
SOCIAL IDENTITY AND PERCEIVED PERSUASION 1889
contrary to expectations (Hypothesis 4), there was no evidence to suggest that high
identifiers differentiated more clearly than low identifiers between in-group and
out-group others (Ms = 5.66 and 5.63, ns, high identifiers; Ms = 5.50 and 5.66. p <
.lo, low identifiers).
When age was included as a covariate in this analysis, the effect of age was
nonsignificant, F( 1, 5 5 ) = 1.31, ns. Moreover, correlational analyses revealed
that age was unrelated to perceptions of influence on self (Y = -.Ol), students (Y =
-.23). and nonstudents (Y = -.22) and only weakly related to perceptions of influ-
ence on people in general ( Y = -.28, p < .05).
A second analysis focused on the perceived impact of the subset of five AIDS
commercials that students thought it was best to be influenced by, and, as
expected, patterns in this context were more consistent with theoretical predic-
tions. As in the first analysis, there was no significant main effect for identifica-
tion, F( 1 , 5 6 ) = 1.55, ns, or for target, F(3, 168) = 1.27, ns, but there was a
significant Identification x Target interaction, F(3, 168) = 3 . 5 1 , < ~ .05 (Figure
2). Results of planned contrasts indicate, as expected (Hypothesis I ) , that low
identifiers displayed the typical third-person effect perceiving self ( M = 5.69) as
less influenced than others by AIDS advertising ( M s = 6.21 students, 6.30 non-
students, and 6.25 people in general, p < .05). By contrast, high identifiers tended
to perceive self(M= 6.48) and students ( M = 6.46) as somewhat more influenced
than nonstudents ( M = 6 . 2 4 , ~ < .lo). As expected (Hypothesis 2). compared with
low identifiers, high identifiers were more willing to acknowledge the influence
of AIDS advertising on themselves ( M s = 6.48 and 5 . 6 9 , < ~ .05) and on the stu-
dent in-group ( M s = 6.46 and 6.21, ns), although this latter difference was not
significant. Consistent with Hypotheses 3 and 4, respectively, high identifiers
also differentiated less clearly between self and in-group others ( M s = 6.48 and
6.46, ns) than did low identifiers (Ms = 5.69 and 6.21, p < .05), and differentiated
more clearly between in-group and out-group others (Ms = 6.46 and 6.24, p < .05
high identifiers; Ms = 6.2 1 and 6.30, ns, low identifiers).
When age was included as a covariate in this analysis. the effect of age was
nonsignificant, F( I , 5 5 ) = 0.23, ns. Moreover, correlational analyses revealed that
age was unrelated to perceptions of influence on self ( Y = .05, ns), or on others
(Y = -.18, ns, students; r = -. 13, ns, nonstudents; and -.23, ns, people in general).
Similar analyses also were performed using influence ratings averaged across
the six informative and five emotional advertisements, respectively. The mean
1890 DUCK ET AL.
5.6
LOW High
Identification
Target 61 self
Students
Non-studcnu
People
Table 1
Target
Non- People in
Self Students students general
Informative advertisements
Low ( n = 30) 5.66 5.98 6.17 6.17
High ( n = 28) 6.15 5.99 6.00 6.16
Emotional advertisements
Low ( n = 30) 5.79 6.05 6.19 5.95
High ( n = 28) 6.54 6.42 6.29 6.42
Note. Scale range is 1 (not influenced at all) to 9 (extremely influenced).
directions, only one planned contrast was significant: High identifiers were more
willing than low identifiers to acknowledge the influence of AIDS advertising on
themselves ( M s = 6.54 and 5.79, p < .05).
Discussion
Taken together, these results provide evidence for the role of group processes
in third-person perceptions. Although based on a relatively small sample, our
findings suggest that perceived self-other differences in persuasion reflect both
the social psychological relationship between self and other and in-group norms
concerning the acknowledgment of persuasive influence. Indeed, by demonstrat-
ing perceptions of unique invulnerability to AIDS messages on the part of stu-
dents who did not identify strongly with the student community (low identifiers).
and a tendency toward perceptions of relative vulnerability on the part of
students who identified strongly with the student in-group (high identifiers), our
results provide some indication of the potential importance of the social-nornia-
tive context in determining both the magnitude and direction of perceived self-
other differences. To this extent, our results support and extend Davison’s (1 983)
early speculation that “the concept of reference groups may prove useful in
explaining the third-person effect” (p. 12) and accord with our theoretical empha-
sis (viz. social identity theory and self-categorization theory) on the interdepen-
dence of individual cognition and social influence (e.g., Turner, 1991).
As expected, analysis of respondents’ perceptions of the impact of 1 1 AIDS
advertisements on self and other provided some support for hypotheses drawn
1892 DUCK ET AL
References
Duck, J. M., Terry, D. J., & Hogg, M. A. (1995). The perceived influence of
AIDS advertising: Third-person effects in the context of positive media con-
tent. Basic and Applied Social Psychology, 17,305-325.
Festinger, L. (1953). An analysis of compliant behavior. In M. Sherif & M. 0.
Wilson (Eds.), Group relations at the crossroads (pp. 232-256). New York,
N Y Harper.
Gibbon, P., & Durkin, K. (1995). The third-person effect: Social distance and
perceived media bias. European Journal of Social Psychology, 25,597-602.
Goodwin, M . P., & Roscoe, B. (1988). AIDS: Students’ knowledge and attitudes
at a Midwestern university. Journal of American College Health, 36, 2 14-
222.
Gunther, A. C. (1991). What we think others think: Cause and consequence in the
third-person effect. Communication Research, 18,355-372.
Gunther, A. C. (1995). Overrating the X-rating: The third-person perception and
support for censorship of pornography. Journal of Con7mztnication, 45, 237-
238.
Gunther, A. C., & Mundy, P. ( 1993). Biased optimism and the third-person effect.
Journalism Quarterlv, 70, 58-67.
Gunther, A. C., & Thorson, E. (1992). Perceived persuasive effects of product
commercials and public service announcements: Third-person effects in new
domains. Communication Research, 19. 574-596.
Heider, F. (1958). The psychologv of interpersonal relations New York, NY:
John Wiley & Sons.
Hogg, M. A. ( 1992). The social psychology of group cohesiveness: From attrm-
tion to social identity. London, UK: Harvester Wheatsheaf.
Hogg, M. A., & Abrams, D. (1988). Social identifications: A socialpsycholo,g~
of intergroup relations and group processes. London, UK: Routledge.
Hogg, M. A., & Abrams, D. (1990). Social motivation, self-esteem, and social
identity. In D. Abrams & M. A. Hogg (Eds.), Social identity theory. Con-
structive and critical advances (pp. 28-47). London, UK: Harvester Wheat-
sheaf.
Hogg, M. A., & Hains, S. C. (1996). Intergroup relations and group solidarity:
Effects of group identification and social beliefs on depersonalized attraction.
Journal of Personality and Social Psychologv, 70,295-309.
Innes, J. M., & Zeitz, H. (1988). The public’s view of the impact of the mass
media: A test of the “third person” effect. European Journal of Social Psy-
chology, 18,457-463.
Lasorsa, D. (1989). Real and perceived effects of “Amerika.” Journalism Quar-
terly, 66, 373-378, 529.
Mackie, D. M., Gastardo-Conaco, M. C., & Skelly, J. J. (1992). Knowledge of
the advocated position and the processing of in-group and out-group persua-
sive messages. Personality and Social Psychology Bulletin. 92, 145- 1 5 1.
1898 DUCK ET AL.