The Third-Person Effect and Social Distance:
Exploring who are the "others"
Joan Conners
Doctoral Student
School of Journalism and Mass Communication
University of Minnesota
111 Murphy Hall
206 Church St. S.E.
Minneapolis, MN 55455
Submitted to the Communication Theory and Methodology
Division
AEJMC, 1994
Abstract
This study considered how "others" in third person
effect research are defined. It suggested that defining others
not by general groups but rather by categorical others, an
individual with a general label, or individuated others, an
specific individual with attributes, might lead to differences in
the estimates made in the effects of mass media upon others.
Little overall support for the third person effect was found,
although significant support was found when others were
individuated.
The Third-Person Effect and Social Distance:
Exploring who are the "others"
The third-person effect hypothesis predicts "that
people will tend to overestimate the influence that mass
communications have on the attitudes and behavior of others"
(Davison, 1983, p. 3). This hypothesis has been studied in a
number of different contexts over the past 11 years. Perloff
(1993) identified the four major factors that have been found to
facilitate a third-person effect: message topic, involvement
with topic, source bias and social distance.
It is the concept of social distance, or who the
third persons, or "others" are which has received little
attention in third-person effect research. While there has been
attention to different types of media messages and the presence
of the third-person effect, there has been little effort to
understand how one perceives others when attempting to estimate
the effects of mass media upon them. The third-person effect is
typically measured by the difference between the estimated effect
of a media message on oneself, and the estimated effect of the
message on each group of others provided (described typically
with little information, such as "other students in this class").
Third-person effect research has considered the effect on
"others" in general (Gunther and Mundy, 1993), or has provided
levels of broadly defined groups of others at different levels of
social distance, such as other students at the university, others
in the state and others in general (Cohen and Davis, 1991;
Gunther, 1991; Cohen et al., 1988).
Perloff (1993) suggests that social distance is a
"complex variable that includes various components, such as
perceived similarity, familiarity and identification" (p. 175).
Perloff distinguishes between two conceptualizations of social
distance that may apply to third-person effect research. One
interpretation suggests social distance falls along a continuum
which ranges from "just like me" to "not at all like me." A
second interpretation, which Perloff says has been used more
often in third-person effect research, "reflects the
heterogeneity and size of the audience or group in question" (p.
175-6). This continuum ranges from "my closest group or
community" to "my largest group or community."
A different approach to test the limits of the
third-person effect is to vary how "others" are described to
respondents upon who they estimate the effect of a message.
Previous research has looked at the social distance between
oneself and groups of others. The third-person effect, however,
may not occur when others are described more specifically, as
individuals. This project considers how the use of categorical
or individuating information in describing others affects the
prevalence of the third-person effect compared to how it is
typically studied by referring to others by broad and general
categories.
This paper considers if the third-person effect
remains when "others" are described by categorical or
individuating information. If greater information on "others" is
provided, or if others are identified as particular individuals,
will respondents still estimate greater mass media effect on
other people than themselves? Does more information provided
about others diminish the third-person effect, or is the
third-person effect prevalent regardless of the amount of
information on others? Research on the use of categorical and
individuating information suggests that when others are
individuated, the third-person effect may diminish. It may be
the case that when the "other" is a specific individual it may be
more difficult to make assumptions and generalize about the
person, and if the other individual is similar to the respondent,
subjects may not estimate media effects all that differently from
the effect on oneself.
This approach in defining others does not
necessarily fit into Perloff's two conceptualizations of social
distance. The conceptualization of the continuum ranging from
"just like me" to "not at all like me" may rely on perceived
similarity or identification with others more than on
familiarity. Measuring social distance by broad groups to tap
this conceptualization may not suffice; however, if others were
defined as specific individuals whom respondents can evaluated as
like or unlike themselves, this may more adequately measure this
concept. Evaluating how individuating and categorical
information differ in social psychological research may provide
clarification for how such information could be used to test the
third-person effect hypothesis.
Research conducted by Krueger and Rothbart (1988)
suggests differences may be found depending on the type of
information provided in third-person effect research. While
Krueger and Rothbart consider what information respondents use to
"make inferences about people's attributes or about the causes of
their behavior" (p. 187), third-person effect research asks
respondents to estimate the effect on one's attitudes, which may
be a more complex task. As Krueger and Rothbart suggest though,
people "typically rely more on the specific properties of the
stimulus person than on the general statistical properties of the
class" (p. 187). The same may occur when people are asked to
make estimates on the effect of mass media on individuals, in
that individuating information may be more useful than general or
categorical information. In using categorical and individuating
information in describing others in third-person effect research,
Krueger and Rothbart's findings suggest that individuating
information may be more influential than categorical.
Kunda and Sherman-Williams (1993) argue that
categorical information may affect the "construal" of
individuating information, which suggests that categorical
information does have influence even when individuating
information is provided. They say "judgments made by people who
base their impressions only on individuating information,
regardless of the stereotype, can nevertheless be influenced by
stereotypes if the individuating information is ambiguous and
open to multiple, stereotype-driven construals" (p. 91). They
found this to be the case and when individuating information was
relevant and unambiguous, stereotypes did not affect judgments.
Kunda and Sherman-Williams' findings might suggest if describing
individuals does lead to different third-person effects than
previously measured by general groups of others, then the
differences might be greatest between general others and when
unambiguous individuating information is provided.
Fiske, et al.'s (1987) approach to understanding
category- based and attribute- based reactions to others may also
help explain how the third-person effect may differ depending on
how "others" are described. They suggest categorization is
typically used but they found it is not successful when
attributes do not fit a particular category label, or attributes
do not cue a particular category. Their findings suggest when
only categorical information is provided, people use those
categories to make inferences. When categorical and attribute
information are both provided, a combination of categorical and
attribute-based processing occurs. When applying Fiske et al.'s
findings to the third-person effect, differences may be most
likely found in the perceptions of media effects on others
depending on the categorical and attribute based information
provided and the consistency between them.
This project constitutes a preliminary test which
considers to what extent different third-person effect results
are found depending on how the "others", the third persons, are
defined.
Support for a "strong" third-person effect would be
represented by significant differences between effect on self and
effect on others regardless of how the others are defined or
described. Others could be described as general others as has
been used in previous third-person effect research, or instead
others could be described categorically or are individuated as
specific people. This would suggest that regardless of who the
"other" is, a person will overestimate the effect of mass media
message on the "other".
On the other hand, if there are found to be
different patterns of results if others are described differently
than they have been defined in the past, this would question how
likely is the third-person effect to occur. If, for instance,
the third-person effect is supported when others are described as
general groups as has been tested in the past, but there is no
evidence support such an effect when others are described
differently, these results would suggest the third-person effect
is limited when estimating the effects of mass media upon general
groups of others rather than on specific individuals. It may be
the case that as the "other" is more individuated, made more
concrete, substantive, and realistic, the third-person effect
declines because generalization and estimation of effects upon
that other individual may be more difficult to do than upon
groups of general others. On the other hand, because the
individual is described and differentiated by details provided
about the individual other, respondents may be more likely not to
see that other person as similar in any way to themselves because
the differences between them are much clearer than in the case of
"other university students" or similarly broad groups in which
little differentiation is provided. These potential differences
are explored below.
The other questions examined in this study involve
the importance of the issue to the respondent, and the opinion of
the respondent prior to reading a particular message, and how
these factors may influence the likelihood of third-person
effects occurring.
Method
Procedure
Subjects read two newspaper editorials and
following each editorial they completed questions regarding the
effect of the editorial. Subjects were randomly assigned surveys
to complete, with questions of general others (N=26), categorical
others (N=22) or individuated others (N=28). In addition to
being asked to what extent the respective editorial affected
their own opinion on the topic of the editorial, respondents were
also asked how it would affect three "others" who varied by
perceived similarity with the respondent. Undergraduate
introductory journalism students completed the survey.
Editorials
Subjects read two editorials on recent events, in
which one discussed controversy surrounding the fate of the local
professional basketball team which may be moved to another city.
The other editorial discussed United States and Russian relations
following discovering that Aldrich Ames was a spy, and what the
United States' reaction should be, recommending aid to Russia not
be cut in response. One half of students read the basketball
editorial first, while the others read the spy editorial first,
although no differences in responses were found related to
editorial order.
Measurement
Effect of message on self. Once the first of the
two editorials were read, subjects responded to the question,
"How does the above editorial affect your opinion (about the
team/on reaction about the spy)?" and indicated responses on a
7-point scale which ranged from "more in favor of saving team" to
"more opposed to saving team" and from "more in favor in
punishing Russia" to "more opposed to punishing Russia."
Effect of message on others. Following estimating
the effect of the message on themselves, respondents then read
the statement "we are also curious how you think other people
would feel about this editorial", which was followed by a
description of the other group or person. Subjects were then
asked To what extent do you think this editorial would affect
this person's/their opinion(s). for each of the three others
described in the condition. They responded on 7 point scales
which ranged from "more in favor of saving team" to "more opposed
to saving team" and from "more in favor in punishing Russia" to
"more opposed to punishing Russia." For the "general others"
condition, respondents were asked to estimate the effect of the
editorial on other university students, others in the state, and
other people in general. For the "categorical others" condition,
respondents were asked to estimate the effect of each editorial
on the typical college student, typical housewife, and typical
business executive. For the "individuated others" condition,
respondents were asked to estimate the effect of each editorial
on "Jeffrey, a senior in history...", "Cynthia, a computer
programmer..." and "Joseph, retired from the military in
Phoenix...".
Opinion on subject. Prior to reading either
editorial, respondents were asked for their opinion on the topics
of the subsequent editorials and responded on 6-point scales
which ranged from "strongly in favor of saving the team" to
"strongly opposed to saving the team" for the basketball
editorial, and from "strongly favor cutting aid to Russia" to
"strongly opposed cutting aid to Russia" for the Ames spy
editorial.
Importance of subject. Also prior to reading the
editorials, respondents rated the importance of the two topics on
4-point scales which ranged from "extremely important to me" to
"not at all important to me."
Other measures. Respondents were also asked to
evaluate each editorial on how interesting it was, how strong an
argument it made, how clear was it, and how much of the editorial
would they have read if they saw it in the newspaper. These
items were included in an attempt to detract from the main
purpose of the study, the estimated effects on oneself and
others.
Results
Opinion and issue importance
Prior to reading the editorial, 88% of respondents
reported favoring saving the team to some degree, which only 12%
opposed saving the team. Regarding the spy topic, 52.4% favored
cutting aid to Russia, while 47.6% opposed it. Clearly these two
issues different dramatically in distribution of opinion about
the topic. The two also differed by reporter issue importance by
respondents. Only 7.1% reported the basketball team issue was
extremely important to them, while 16.7 % said the issue was
moderately important, 29.5% reported the issue was somewhat
important, and 45.2% said the issue was not at all important to
them. With regard to the spy issue, 14.3% reported it was an
extremely important issue, 36.9% reported the issue was
moderately important, 36.9% reported the issue was somewhat
important to them, and 11.9% reported the issue was not at all
important to them.
Effect on self
A large proportion reported no change in opinion
on the basketball team after reading the editorial. The large
proportion of support for the team which was reported prior to
reading the editorial may have resulted in little opportunity for
perceived change, as the editorial favored change. Regarding the
spy editorial, fewer reported no change in opinion, while more
than half reported some degree of opposing cutting aid to Russia,
the direction advocated in the editorial.
-----------------------------------------------------------------Table 1 about
here
-----------------------------------------------------------------
Effect on others versus effect on self
Paired sample t-tests were conducted to compare
the "effect on self" scores with "effect on others" scores. Few
significant differences in effect on self and effect on others
were found; support for the third-person effect was found only in
the "individuated others" condition for both editorials.
-----------------------------------------------------------------Table 2 about
here
-----------------------------------------------------------------
The original opinions respondents took of the two
issues, and the reported importance of the issues to the
respondent may identify differences in the relationship between
effect on self and effect on others, and the low support for the
third-person effect.
Prior opinion on topic
Regarding original opinion positions, there was a
significant difference in reported effect on self between those
who originally favored saving the team and those who opposed
saving the team to some degree. Mean "effect on self" scores for
those favoring saving the team was 3.66, while the mean for those
opposing saving the team was 4.63 (t=2.04, p=.045). Respondents
were then split according to their original position on the issue
(favor saving the team vs. oppose saving the team), and
difference scores (difference between effect on self and effect
on each "other") were then compared. No significant differences
were found, although two comparisons approached significance.
Due to the majority of respondents favoring saving the team, few
subjects opposed saving the team. Small sample sizes for each
condition may have also contributed to the lack of adequate
comparisons.
For opinions on the spy issue, the difference
between effect on self for those who favored cutting aid and
those who opposed cutting aid approached significance. Mean
"effect on self" scores for those favoring cutting aid was 4.18,
while the mean for those opposed to cutting aid was 4.81 (t =
-1.76, p = .083). Respondents were again split into two groups
(favor cutting aid vs. oppose cutting aid) and difference scores
were compared. While no comparisons were significant for the
general others and categorical others conditions, there was some
differentiation in the individuated others condition. While one
comparison was significant at the p<.01 level, the two other
comparisons in the individuated others condition approached
significance.
-----------------------------------------------------------------Table 3 about
here
-----------------------------------------------------------------
Issue importance
In considering how different levels of issue
importance might influence the relationship between estimated
effect on self and estimated effect on others, importance was
also collapsed into a dichotomous variable of "high importance"
and "low importance". With regard to the basketball team, nearly
75% saw the issue of low importance, while approximately 25%
reported it to be a highly important issue. With regard to
effect of message on oneself, issue importance appears to be
influential. Mean "effect on self" scores for those reporting
the issue to be highly important was 2.88, while the mean for
those reporting the issue was not important was 4.07 (t=-3.48,
p=.001). When considering the difference scores between effect
on self and effect on others, small sample sizes, in this case
those for whom the issue was important, was again problematic in
achieving significance. Significant differences were found only
in the general others condition, although the balance of the two
groups in terms of sample size was more disproportionate in the
other two conditions.
In the spy issue, there was a greater balance of
issue importance; when collapsed into high or low importance,
51.2% reported the issue was highly important, while 48.8% said
the issue was of low importance. With regard to effect of the
message on oneself, issue importance did not produce significant
differences. Mean "effect on self" scores for those reporting
the issue be highly important was 4.24, while the mean for those
reporting the issue was not important was 4.77 (t=-1.47, p=.15).
When evaluating difference scores for the two groups,
few significant differences were found, although in a couple
cases issue importance did make a difference. For this issue,
however, there was no consistency in the significant differences
all occurring in the same condition as was the case in the
basketball team topic, and in both topics regarding position on
the issue.
-----------------------------------------------------------------Table 4 about
here
-----------------------------------------------------------------
Conclusions
There was overall weak support for the
third-person effect hypothesis in this exploratory study,
although some results are of interest. When considering which
type of "other" brought about the third-person effect, it is
interesting to note that it was most likely in the individuated
others condition versus the general others or categorical others.
One explanation if these findings are confirmed is that when
"others" are individuated by great detail, people see less
similarity between themselves and these others, they realize they
are not familiar with these other people, and that they do not
identify with them. Possibly by individuating others
respondents see themselves as very different to these others and
therefore are more likely to estimate the effect of the media
upon themselves and these individuated others differently. It
could be the case that when others are broadly defined groups,
while one is not familiar with the individuals who compose that
group, they may be more likely to see themselves as more similar
and be more familiar with those in a less well defined group,
because it is then up to the respondent's interpretation to
define who might compose that group.
These results do support further analysis of issue
importance and original opinion position (and its congruity with
the message) when evaluating the third-person effect. While few
significant differences were found in this study, the topic of
the message manipulation and respondent involvement with that
topic may both be influential in the process of estimating the
effect of mass media upon others. This study confirms the value
the studying these variables further.
Limitations in this study may be responsible for
the lack of success of support for the third-person effect
hypothesis. The small sample size for each condition was
problematic, as was the design of considering the between subject
results (in the 3 conditions) versus evaluating within subject
results (which would have required all respondents to evaluate
all possible "others"; that is, general, categorical and
individuated others. Such a design may be more successful in
determining how "others" are evaluated differently by the same
respondent. While it was expected that the importance of the two
issues would differ, as would the degree of positions taken on
it, a lack of variability was also problematic. As most students
supported saving the basketball team prior to reading the
editorial, there may be little opportunity for opinion change
when so many respondents felt quite strongly about their opinion
already.
This study does suggest particular considerations
for future research. To adequately test which form of "other" is
most likely to produce a third-person effect, or the greatest
differentiation between effect on self and effect on other,
comparisons should be made in a within subject design in which
they estimate the effect of the media on various "others", some
who may be individuated and others who are more general persons
or groups.
Another important factor which may provide insight
into the process by which people estimate the effect of the media
differently is to probe why respondents estimate the effects as
they do. It could be valuable to find out that respondents most
often consider particular characteristics when asked to make such
estimates, such as "how similar is this person to me?", "how
important might this issue be to them compared to how important
it is to me?".
It would also be interesting in testing
individuated others to use real life others. Real others would
provide a greater familiarity for respondents to these others.
It might also be valuable to replicate an existing study which
successfully found support for the third-person effect, and test
the same process using those others defined by the original
research and testing other potential others, whether they be
categorical others, individuated and fictional others, or
individuated and real others.
Table 1
Reported effect on self
Basketball Editorial
No change in my opinion
55.8%
More in favor of saving team
26.0%
More opposed to saving team
18.2%
Spy Editorial
No change in my opinion
17.7%
More in favor of cutting aid to Russia
27.8%
More opposed to cutting aid to Russia
54.4%
Table 2
Paired comparisons of effect scores
Basketball Editorial Spy
Editorial
General others (N=26)
Mean(a) t-value Mean t-value
Effect on self 3.92 4.68
Effect on
students 3.84 .26 4.80 .51
residents 3.92 .37 4.60 .13
others 4.08 .84 4.68 .14
Categorial others (N=22)
Mean t-value Mean t-value
Effect on self 3.64 4.40
Effect on
student 3.36 1.30 4.35 .12
housewife 3.82 .64 4.30 .35
bus. exec. 3.59 .12 4.40 .00
Individuated others (N=28)
Mean t-value Mean t-value
Effect on self 3.96 4.60
Effect on
Jeffrey 3.33 2.35* 4.83 .74
Cynthia 4.59 2.35 4.72 .20
Joseph 4.26 1.28 3.03 3.73*
* p < .05, ** p< .01
(a) Note: Effect scales were 7-point scales, anchored
by "more in favor of saving team" at 1 and "more opposed to
saving team" at 7; and "more in favor of punishing Russia" at 1
and "more opposed to punishing Russia" at 7.
Table 3
Analysis of Opinion Position
by Difference scores (self-other)
Basketball Editorial Spy Editorial
General others
Favor save Oppose save T-value Favor cut
Oppose cut T-value
Self-students -.14 -.50 .32 -.50
.08 .90
Self-residents -.19 .00 .18 -.17
.17 .53
Self-others -.29 -.50 .20 -.17
.00 .28
Categorical others
Favor save Oppose save T-value Favor cut
Oppose cut T-value
Self-student .30 .00 .40 .38
-.17 .64
Self-housewife -.15 -.50 .35 .13
.08 .07
Self-bus. exec. .25 -2.0 1.73* .50
-.33 .96
Individuated others
Favor save Oppose save T-value Favor cut
Oppose cut T-value
Self-Jeffrey .52 .75 .29 -.89
.73 2.66***
Self-Cynthia -.70 .75 1.93* -.59
.64 1.77*
Self-Joseph -.35 .50 1.30 .89
2.45 1.86*
* p< .10
** p< .05
*** p< .01
Table 4
Analysis of Issue importance
by Difference scores (self-other)
Basketball Editorial Spy
Editorial
General others
High Imp. Low Imp. T-value High Imp.
Low Imp. T-value
Self-students -1.29 .47 2.22* .00
-.38 .58
Self-residents -1.00 .42 2.08** -.09
.08 .27
Self-others -1.14 .29 2.11** -.09
-.07 .02
Categorical others
High Imp. Low Imp. T-value High Imp.
Low Imp. T-value
Self-student .01 .33 -.60 -.80
.90 -2.32**
Self-housewife -.50 .11 -.52 -.20
.40 -1.04
Self-bus. exec. .00 .06 .05 -.20
.20 -.46
Individuated others
High Imp. Low Imp. T-value High Imp.
Low Imp. T-value
Self-Jeffrey -.17 .76 -1.44 -.56
.08 -1.04
Self-Cynthia -1.17 -.29 -1.33 -.63
.58 1.96*
Self-Joseph -.83 -.05 -1.42 1.75
1.15 .69
* p< .10
** p< .05
*** p< .01
References
Cohen, J. & Davis, R.G. (1991). Third-person effects
and the
differential impact in negative political
advertising.
Journalism Quarterly, 68, 680-88.
Cohen, J., Mutz, D., Price, V. & Gunther, A. (1988).
Perceived
impact of defamation: An experiment on
third-person
effects. Public Opinion Quarterly, 52, 161-173.
Davison, W.P. (1983). The third-person effect in
communication.
Public Opinion Quarterly, 47, 3-15.
Fiske, S.T., Neuberg, S.L. (1990). A continuum of
impression
formation, from category-based to individuating
processes:
Influences of information and motivation on
attention and interpretation.
Advances in Experimental Social Psychology. 23,
1-74.
Gunther, A., & Mundy, P. (1993). Biased optimism and
the third-
person effect. Communication Research, 18,
355-72.
Krueger, J. & Rothbart, M. (1988). Use of categorical
and
individuating information in making inferences
about personality. Journal
of Personality and Social Psychology, 55, 187-195,
Kunda, Z., & Sherman-Williams, B. (1993). Sterotypes
and the
construal of individuating information.
Personality and
Social Psychology Bulletin, 19, 90-99.
Perloff, R.M. (1993). Third-person effect research
1983-1992: A
review and synthesis. International Journal of
Public Opinion Research, 5,
167-184.
|
