Political Psychology, Vol. 28, No. 2, 2007
Social Influence on Political Judgments: The Case of
Presidential Debates
Steven Fein
Williams College
George R. Goethals
University of Richmond
Matthew B. Kugler
Princeton University
Four experiments investigated the extent to which judgments of candidate performance in
presidential debates could be influenced by the mere knowledge of others’ reactions. In
Experiments 1 and 2 participants watched an intact version of a debate or an edited version
in which either “soundbite” one-liners or the audience reaction to those soundbites were
removed. In Experiment 3 participants saw what was supposedly the reaction of their fellow
participants on screen during the debate. Participants in Experiment 4 were exposed to the
reactions of live confederates as they watched the last debate of an active presidential
campaign. In all studies, audience reactions produced large shifts in participants’ judgments of performance. The results illustrate the power of social context to strongly influence
individuals’ judgments of even large amounts of relevant, important information, and they
support the categorization of presidential debates as ambiguous stimuli, fertile ground for
informational social influence.
KEY WORDS: presidential debates, conformity, ambiguity, informational social influence, political
communication
Ronald Reagan leans back and enjoys the moment. It is the second of his two
presidential debates with Walter Mondale as they near the finish line to the 1984
U.S. presidential election. He has just delivered his famous line: “I will not make
age an issue of this campaign. I am not going to exploit, for political purposes, my
opponent’s youth and inexperience.” The question had been a difficult one, asking
165
0162-895X © 2007 International Society of Political Psychology
Published by Blackwell Publishing. Inc., 350 Main Street, Malden, MA 02148, USA, 9600 Garsington Road, Oxford, OX4 2DQ,
and PO Box 378 Carlton South, 3053 Victoria Australia
166
Fein et al.
about Reagan’s advanced age and the concerns raised by his performance in the
first debate, but his answer has satisfied the audience and they are cheering. The
moderator, Henry Trewhitt, says “I’d like to head for the fence and try to catch that
one before it goes over.” The newspapers soon report that Reagan’s response has
erased concerns over his age. Reagan’s declining poll numbers quickly returned to
their previously high levels. He wins reelection in 1984 in a 49-state landslide.
The above events could all be seen as independently stemming from Reagan’s
answer. The audience heard it, and applauded; the moderator heard it, and praised
him; the newspapers heard it, and stopped discussing the age issue; the public heard
it, and voted for him. But these events could also be seen as describing a causal path.
The audience heard it and they applauded, the moderator heard the applause and
praised Reagan’s answer, and so on. The difference between these two ways of
looking at what happened is that one, the latter, does not give the actors much credit.
After the initial applause, everyone is part of the herd, running smoothly down
preset tracks, conforming to the example set by a small studio audience that was
neither better nor wiser than they. In American culture, great value is set on
independence. One of the primary sources of enthusiasm for debates is the belief
that they help the public makes up its own mind by allowing them to cut through the
superficial. Our second, uncharitable, interpretation of these events is quite worrisome from this individualistic perspective, but that does not mean it is wrong.
While many studies have examined conformity in basic laboratory settings,
none exists that explores how it may affect assessments of such complex and
important real world events as presidential debates. Millions of people watch
debates, discuss them, and read and watch analyses of the candidates’ performances. They have been shown to play a role in the outcome of every campaign in
which they have been held, and they may have been decisive in several (e.g., 1960,
1980, 1992, and 2000). Is conformity an important determinant of people’s reactions to them?
While we do not know whether potential voters simply conform to the reactions of other viewers, we do know that contextual features of debates—for
example, what network commentators say about them and whether citizens watch
them on television or listen to them on the radio—make a difference (Kaid &
Bystrom, 1999; Schroeder, 2000). An overall conclusion from Kraus’s (1962)
volume on The Great Debates of 1960 between John F. Kennedy and Richard M.
Nixon was that the debates mattered whether people watched them or not. There
was so much discussion, especially of Kennedy’s performance in the first of the
four debates, that a social reality took shape holding that Kennedy had “won” the
debates, and this perception very likely was an important factor in the outcome of
a very close election. Similarly, media commentary following the Ford-Carter
foreign policy debate in 1976 contributed substantially to both the salience and
negative evaluation of Ford’s statement that there was no Soviet domination of
Eastern Europe (Steeper, 1978). McKinnon, Tedesco, and Kaid (1993) showed
that network commentary after the 1992 debates between George Bush, Bill
Presidential Debates
167
Clinton, and Ross Perot lessened Clinton’s perceived margins of victory, and
McKinnon and Tedesco (1999) showed that network commentary following the
1996 debates raised perceivers’ assessments of both Bill Clinton and Bob Dole.
While it is not entirely predictable how commentators’ appraisals might affect
voter opinion, it seems clear that they can and do.
Considering the influence of audience reaction specifically, some elements of
the social influence literature suggest that they would produce little conformity in
judging presidential debates. First, social influence often produces public compliance (Kiesler & Kiesler, 1969) through normative social influence, where people
conform essentially to be liked by their peers. Debate evaluations poorly fit that
mold. There is little normative pressure to conform to debate audiences; viewers
are not threatened for failing to cheer when their fellows do, especially not when
the fellows in question are a studio audience on the other side of the country. And
debates are important in that they can affect private voting behavior on Election
Day, when people will be answering to no one but themselves. When considering
voting behavior, we must limit ourselves primarily to informational social influence, conformity that works on a person’s desire to be right, not their desire to be
liked (Deutsch & Gerard, 1955).
Debate audiences supply little fodder for informational social influence. Their
reactions do little more than provide knowledge of others’ judgments. They do not
present arguments or reasons for those judgments. This stands in contrast to the
separate processes of pre- and postdebate spin, which do provide at least some
argumentation and analysis. Thus audience reactions provide information about
the judgments of other people, but not the reasons for their judgments, and, as
noted above, they apply little normative pressure to conform to or to adopt those
judgments. In such situations, the actions of the others provide only information
about the typical or predominant response to the situation and, therefore, what the
appropriate response for the individual might be.
How likely is it that this seemingly limited form of social influence can affect
a real-world judgment, one that we assume not only is important but also that most
people believe is important? While we are intuitively drawn to doubt that people
would conform on important tasks, conformity research does show that not only
trivial laboratory judgments about the length of lines (Asch, 1951) or the movement of lights (Sherif, 1936), but also substantially more important, real-world
questions, such as whether New Jersey is really being invaded by Martians
(Cantril, 1940), can be heavily influenced by other people’s reactions. Still, most
of the research on social influence concerns judgments made in the laboratory and
its external validity is often unknown. And even laboratory experiments show that
conformity is often reduced when judgments are made more consequential (Baron,
Vandello, & Brunsman, 1996). In a somewhat related vein, theory and research
supporting the Elaboration Likelihood Model of persuasion indicate that individuals are less likely to be influenced by factors peripheral to the content of a message
when the issue is an important one (Petty & Cacioppo, 1986).
168
Fein et al.
With these considerations in mind, the literature suggests that conformity to
studio audiences in debates should be weak. As noted, there is little interpersonal
or normative pressure, no explanations are provided to explain the audience
behavior, and the judgment is important, giving the participant every reason to
work for accuracy. Is there any reason to think this limited form of social influence
could affect such important judgments?
The results of McKinnon et al. (1993)’s research concerning the 1992 presidential debates suggested that viewers may typically process the debates at a
relatively shallow, heuristic level. To the extent that this is the case, individuals are
more likely to be susceptible to social influence and other factors peripheral to the
content of the debate. Another reason is suggested by a finding in the Baron et al.
study cited above. They showed that importance reduces conformity on clear-cut
judgments, but not when the judgment is ambiguous. Are presidential debates
clear stimuli? In reconstructing debates, people often see them as having clear
outcomes. Commentators tell narrative-style stories explaining how candidates
won or lost: Michael Dukakis as the unnatural “Iceman” in 1988, George H. W.
Bush as the disconnected elitist in 1992, John Kennedy as the unexpected professional in 1960. After the tales are told enough times, it becomes hard to imagine
that things could be seen any other way.
It is not obvious, though, that we should accept this retrospective consensus
as good evidence that debates are clear stimuli. As noted above, audience and
media reactions that accompany debates are a major part of the unfolding political
stories. While it could be that the debates really are clear, it could also be that
perceptions of debates are instead largely a function of how studio audiences,
debate panelists, and various professional commentators respond and the narratives they construct. Would Lloyd Bentsen telling Dan Quayle that he was “no Jack
Kennedy” in the 1988 vice-presidential debate (Germond & Witcover, 1989) have
been as memorable if the audience hadn’t reacted loudly? Polls show, for example,
that Al Gore in 2000 (Jamieson & Waldman, 2002) and Gerald Ford in 1976 (Sears
& Chaffee, 1979) both won debates in their immediate aftermath but lost them in
the postdebate spin wars, Ford to Eastern European immigrants who felt quite sure
that their homelands were under Soviet domination and Gore to parodies of him on
the television show Saturday Night Live.
A critical question, then, is whether presidential debates constitute an ambiguous stimulus. Certainly debates provide a rich and substantial physical reality.
Therefore individuals might not be much influenced by other people’s reactions.
Furthermore, people may have strong candidate preferences going into a debate and,
often, strong expectations that their candidate will win, with the result that they have
clear perceptions of exactly who won (their candidate) and little susceptibility
to social influence. Sigelman and Sigelman’s study (1984) on the Carter-Reagan
debates found many people do exactly that, as did Sears and Chaffee’s (1979)
review of studies on the 1976 debates. On the other hand, the complex physical
reality of debates might be difficult to interpret, leaving participants open to outside
Presidential Debates
169
“help” as has been seen in studies of pre- and postdebate spin (Fein, Frost, Goethals,
& Kassin, 1994; Kugler & Goethals, 2005; Norton & Goethals, 2004). This
perspective suggests a high potential for conformity effects. Because of the influence of debates in our modern political arrangements, it seems important to find out.
Four experiments are reported below. In the first three experiments, participants watched parts or all of the second 1984 Mondale-Reagan debate, approximately a decade (from 9 to 12 years) after the debate occurred. In the first of these
studies, participants watched parts of the debates under one of three conditions: a
pure, unaltered format; with two “soundbites” deleted; or with the soundbites
remaining but the audience reaction to them deleted. With this design, we could
isolate the effects of the content of the soundbites versus the effects of the audience
reaction to those soundbites. Experiment 2 used a similar design, but in this study
participants saw the entire 90-minute debate, rather than excerpts from the debate.
Participants in Experiment 3 watched the excerpts from this debate, but saw
superimposed on the video screen alleged real-time tracking of other people’s
reactions. The fourth experiment we report was actually conducted before the first
three, in October of 1992 on the night of the third debate between George Bush,
Bill Clinton, and Ross Perot. It is reported last because it addresses a key “external
reality” question lurking in the background of the first three. That is, can real
voters be influenced by the reactions of others when watching a debate in real time,
when it is actually broadcast, when there are real levels of engagement and
resistance? In this study, participants watched the debate either in the presence of
several confederates who cheered for pro-Bush remarks, in the presence of several
confederates who cheered for pro-Clinton remarks, or with no confederates.
In all four experiments the main dependent variables were participants’ judgments of each candidate’s overall performance and personal qualities. Because we
believe that presidential debates are, in fact, rather ambiguous stimuli and ripe for
social influence, we predicted for all these experiments that the cues from the
studio audience or from the participants’ peers around them would significantly
influence participants’ judgments and that this social influence would emerge
despite the large amount of relevant content expressed by the candidates during the
debate.
Experiment 1
With recent presidential debates including the presence of a live studio audience, political commentators have suggested that too often debates turn on simpleminded soundbites, one-liners, slogans, and canned mini-speeches designed to
elicit resounding applause (Sigelman, 1992; Zarevsky, 1992). In 1980, Ronald
Reagan buried Jimmy Carter in a televised debate the moment he shook his head
and, with a tone of sorrow in his voice, delivered the fatal blow, “There you go
again.” Four years later, with concerns being raised about his age and mental
acumen, Reagan turned around his faltering campaign with his one-liner noted
170
Fein et al.
above: “I am not going to exploit . . . my opponent’s youth and inexperience.”
Then in 1988, Lloyd Bentsen stopped Dan Quayle in his tracks by responding to
Quayle’s self-comparison to John F. Kennedy with the unforgettable line, “You’re
no Jack Kennedy.” In all instances, live audiences loudly erupted with some
combination of cheers, laughter, boos, and applause.
Our first experiment addressed two questions. First, how much impact do such
memorable moments really have on people’s judgments of a candidate’s overall
performance? It is part of the political lore that debate outcomes, and perhaps
elections as well, may turn on these isolated moments. However, it may be that
because one-liners constitute only a tiny fraction of what transpires in a debate,
they actually have little effect on overall evaluations. To answer this question our
first experiment presented a 40-minute debate segment either with or without
one-liners and the audience reaction they generated. Second—and in the context of
issues about social influence, perhaps a more interesting question—is it the content
of the one-liners themselves, or just the audience reaction to them, that is responsible for any impact the soundbite moments might have? To answer this question
we included a condition in which the candidate’s remarks, but not the reaction to
them, were presented. Thus we compared an unedited 40-minute segment with two
edited variations—one that deleted the soundbite exchanges in their entirety and
one that deleted just the audience reaction to the relevant one-liners.
Method
Participants and Design
Fifty-three introductory psychology students participated in exchange for
course credit. By random assignment, participants saw one of three versions of a
debate. One tape included two soundbites and the positive reactions they had
drawn from the audience. In a second version, the soundbites and reactions were
both deleted from the tape. In a third version, the soundbites were included but the
audience reactions to them were not.
Procedure
Participants completed a brief predebate questionnaire in which they
described their own party affiliation, level of involvement, and political orientation.1 Next they were shown a 40-minute segment of the second (and final) 1984
1
In each of the four studies, we analyzed the main and interactive effects of a variety of individual
differences, including participants’ gender, political party affiliation, degree of involvement and
interest in politics, and so forth. In these studies, there were approximately equal numbers of men and
women (all were undergraduates at Williams College). In each study, the majority of the participants
identified themselves as Democrats (ranging from 51% of the sample to 57%; the percentage
identifying themselves as Republican ranged from 12% to 26%, with the remaining identifying
Presidential Debates
171
debate between Reagan and Mondale. The focus of this debate was on foreign
policy. Participants were randomly assigned to watch one of three versions of the
videotape. In the control condition, the tape included two remarks by Reagan, both
of which drew loud cheers and laughter from members of the live audience (each
remark-reaction segment lasted for approximately 30 seconds). The first remark
concerned a TV ad in which Mondale projected an image of strength by standing
on the deck of the Aircraft Carrier Nimitz. Reagan quipped, “If he had had his way
when the Nimitz was being planned, he would have been deep in the water out
there, because there wouldn’t have been any Nimitz to stand on. He was against it.”
Later, in response to an expected question concerning his age and mental competence, Reagan delivered the memorable remark with which we opened this paper:
“I will not exploit, for political purposes, my opponent’s youth and inexperience.”
In the soundbite-deleted condition both one-liners—as well as the reactions they
had elicited—were deleted. In the reaction-deleted condition, Reagan’s remarks
were included but the audience’s reactions were deleted.2
Following the tape, participants rated the two candidates for their overall
performance on 0–100 point scales. They were also asked to indicate which
candidate (or neither) they thought won the debate. Participants then answered
additional questions concerning each candidate’s handling of key substantive
issues (Central America policy, the Middle East, and national defense) and personal character (sense of humor, intelligence, likeability, strength, competence,
sincerity, and leadership). Finally, participants were asked if they were transported
back in time to the 1984 campaign, who they thought they would have voted for.
Results and Discussion
Overall Performance
A series of one-way analyses of variance (ANOVAs) on the ratings of each
candidate’s performance, as well as the difference between these ratings, revealed
significant effects as a function of condition. As can be seen in Figure 1, participants in the control condition tended to rate Reagan’s performance much more
positively than they rated Mondale’s, reflecting the popular opinion of the debate
2
themselves as Independent or having no party affiliation). In less than half the analyses across the four
studies, a significant main effect for party affiliation emerged, but in the slight majority of analyses,
there was no significant difference as a function of party. No significant interactions between party
affiliation and any of our independent variables emerged. In each study, there were no reliable gender
differences, nor reliable differences as a function of any of the other measures of individual differences that we took. We therefore do not report the analyses as a function of these subject variables in
the Results sections of these studies.
There was occasional, very brief audience applause sprinkled in with the candidates’ responses during
other parts of the debate, but the reaction was always mild and fleeting. No sustained applause,
jeering, or any other reactions could be heard anywhere on the tape, with the exception of the two
soundbites in the control condition.
172
Fein et al.
Figure 1. Post-debate performance ratings of Reagan and Mondale in Experiment 1 as a function
of whether the clips contained the soundbites and the audience reaction (SB-AR), the soundbites
with no audience reaction (SB-NR), or no soundbites (NSB).
back in 1984. Indicating the importance of the two brief soundbites, however,
Figure 1 illustrates that this advantage enjoyed by Reagan disappeared and was
even slightly reversed when the soundbites were deleted (Reagan’s 7.50-point
advantage in the control condition switched to a 7.63 disadvantage in the
soundbite-deleted condition), F(1, 50) = 6.82, p < .02. Most interestingly, the
reversal is even more pronounced in the condition in which the content of
the soundbites was left in, but the audience reaction to them was deleted (Reagan’s
average rating was 23.57 points below Mondale’s), F(1, 50) = 24.29, p < .0001.
The results switched from an apparent victory for Reagan in the control condition
to a decisive victory for Mondale in the condition in which the content was
identical to the control condition but the audience reaction to the soundbites was
eliminated.
Examining the ratings of each candidate independently, ratings of Reagan’s
performance varied significantly as a function of our manipulation, F(2, 50)
= 4.71, p < .02. It is interesting to note that relative to the ratings in the control
condition of Reagan’s performance (M = 66.25), the ratings in the soundbitedeleted condition did not drop significantly (M = 65.40, F < 1), but the ratings in
the reaction-deleted condition did decrease significantly (M = 49.29, F(1, 50)
= 7.89, p < .008. It was the audience reaction to Reagan’s soundbites, rather than
the soundbites themselves, that made the critical difference in judgments of his
performance.
Presidential Debates
173
Although the soundbites were made by Reagan, evaluations of Mondale were
also significantly affected by our manipulations; in this case with participants
rating Mondale more positively in the soundbite-deleted and reaction-deleted
conditions than in the control condition (Ms = 73.03, 72.86, and 58.75, respectively), F(2, 50) = 8.59, p < .001. The effects of the manipulation on ratings of
Mondale’s performance were similar conceptually to results from other research
illustrating the implicit effects on a candidate whose opponent is the target of spin
(Kugler & Goethals, 2005; Norton & Goethals, 2004).
Participants in this study also responded to the question of which candidate
they thought won the debate. Here again the manipulation had a significant effect
on participants’ responses, X2 (2, N = 53) = 17.75, p < .002. A majority of the
participants in the control condition thought Reagan won the debate (60% versus
15% for Mondale and 25% judging that neither won), whereas most participants in
the soundbite-deleted condition thought either neither won (42.11%) or Mondale
won (42.11%), and, even more striking, a large majority in the reaction-deleted
condition thought Mondale won (71.43% versus 7.14% for Reagan and 21.43%
for neither).
For the question of who they would have voted for, there was not a significant
difference between the control and the soundbite-deleted conditions, but there was
a dramatic difference between these conditions and the reaction-deleted condition,
with the percentage of participants indicating they would have voted for Reagan
dropping from 51.28% to 14.29%, and those indicating they would have voted for
Mondale increasing from 33.33% to 71.43% (and the remaining participants
indicating no preference), X2 (2, N = 53) = 6.85, p < .04.
Candidates’ Qualities
The presence of the soundbites and audience reactions also affected perceptions of each candidate’s personal character. On a composite measure that combined mean ratings of intelligence, likeability, leadership, sense of humor,
strength, competence, and sincerity (a = .91), Reagan was viewed less favorably
in the reaction-deleted condition than in the control and soundbite-deleted conditions (Ms = 61.22 compared to 70.25 and 72.23, respectively), F(2, 50) = 2.81,
p < .07—a pattern of results that was strongest on ratings of Reagan’s sense of
humor (M = 83.50 in the control condition, 75.79 in the soundbite-deleted condition, 63.21 in the reaction-deleted condition; F(2, 50) = 6.11, p < .005) and intelligence (Ms = 67.75 versus 71.16 versus 55.71, respectively; F(2, 50) = 4.67,
p < .02). As with overall performance ratings, evaluations of Mondale’s personal
character (a = .87) were also significantly affected by soundbites and audience
reactions, F(2, 50) = 4.56, p < .02—particularly on ratings of his leadership, F(2,
50) = 5.30, p < .005. Remarkably, participants perceived Mondale as having significantly less leadership ability when they watched the original control debate
174
Fein et al.
(M = 62.75) than when they saw the tape without Reagan’s soundbites (M = 75.53)
or without positive audience reactions (M = 75.00).
Summary
Participants’ ratings of the candidates’ overall performance and specific
qualities were affected significantly by our manipulation. Reagan was seen as
the clear victor in the debate in the unedited, control condition, but this victory
was lost with the deletion of the two brief, but apparently critical, soundbites.
Deleting the reaction to the soundbites while keeping the soundbites intact
tended to have a much bigger effect on participants’ relative judgments of the
candidates, illustrating the relative importance of the social context over the
content of the soundbite.
Relative to the control conditions, ratings of Reagan tended to be affected
much more by the deletion of the audience reaction than the deletion of the
soundbite itself, whereas ratings of Mondale tended to be affected equally by
these two conditions. Although it is only speculation, perhaps the content of the
soundbites affected ratings of Mondale more because he was the target of them,
particularly of the one in which Reagan mocked Mondale’s record on defense.
That soundbite may have presented a strong negative for Mondale, although it
may not have counted for ratings of Reagan himself all that much. In any case,
and most importantly, it is clear that the relative ratings of Reagan and Mondale
were influenced strongly by both independent manipulations, and that, overall,
the deletion of the audience reaction had a particularly strong effect. The content
of Reagan’s responses may have hurt perceptions of Mondale, but it was only
with the strong validation of the responses by the audience and moderator that
they made participants see Reagan much more positively.
Experiment 2
The results of Experiment 1 indicated that viewers of the 1984 debate between
Reagan and Mondale were highly influenced not only by two remarks in the
context of a substantive 40-minute debate excerpt, but also by others’ reactions to
those remarks. When the remarks were heard but the audience reactions taken out,
the validation of these remarks as a significant “home run” for Reagan appeared to
evaporate.
In light of the provocative implications of these results, a second experiment
was designed to replicate these results and to address further three questions. One
question is whether these effects would emerge in the context of the full 90-minute
debate. With so much information presented in a full 90 minutes of debate, would
the effects of two brief soundbites, or the reactions to these soundbites, be strong
enough to influence participants’ ultimate judgments? Moreover, we felt that using
the full debate would increase the study’s external validity. A second question
Presidential Debates
175
addressed in this experiment was whether participants perceived the soundbites to
be memorable and noteworthy in the absence of audience reactions. That is, was
the content of Reagan’s remarks really effective and meaningful in the context of
the entire debate, or did they become so in part because of the snowballing reaction
of the moderator and audience? To examine this question, we added an item to the
postdebate questionnaire asking the participants to list several key moments in the
debate.
Third, we worried that having participants sit through an entire 90-minutelong debate would cause the participants to get antsy and stop paying attention
after a while. We thought it might help to interrupt the tape halfway through and
ask the participants some questions about the debate, and then resume the tape
after that break. This both would provide a break and would remind the participants that we would be asking them questions about the debate. We worried,
though, that this could reduce the external validity of the study, and, perhaps, have
some carry-over effects on the final measures. To allay these concerns, we randomly assigned half the participants to watch the entire debate without interruption, whereas the other half were asked questions approximately midway through
the debate.3
Method
Participants and Design
One-hundred and seven introductory psychology students participated in this
experiment for course credit. Participants were randomly assigned to one of six
cells produced by a 3 (control versus no soundbite-deleted versus no reactiondeleted) ¥ 2 (mid-debate ratings versus postdebate ratings only) factorial design.
Procedure
The procedure was similar to that of Experiment 1 with four exceptions. First,
as explained in footnote 3, participants filled out the Need for Cognition Scale as
part of their predebate questionnaire. Second, participants were shown a videotape
3
We originally had a fourth question in mind when designing this experiment: Would individual
differences in need for cognition moderate our effects? Prior research suggested that audience
reaction is a peripheral cue, influencing only participants who are low in involvement or in the need
for cognition (Axsom, Yates, & Chaiken, 1987). This same study, however, did not find such a
limitation on effects concerning participants’ perceptions of a persuasive speaker, as distinct from that
speaker’s message. We administered Cacioppo and Petty’s (1982) Need for Cognition Scale in
Experiment 2 in order to determine whether individual differences in need for cognition would
moderate our effects. As with the other measures of individual differences that we used in the various
studies, we did not find any significant relationship between participants’ scores on this measure and
their ratings of the candidates.
176
Fein et al.
of the entire 90-minute debate rather than the 40-minute excerpt previously used.
Third, for half the participants, we stopped the debate at the halfway point and had
the participants make ratings of the candidates’ performance up to that point,
explaining to them that these ratings were only their sense of how the debate was
going thus far and that their final judgments may or may not be quite different from
them. After a brief break, the rest of the debate was shown, after which participants
filled out the complete set of ratings. Fourth, to examine whether participants were
aware of the impact of the soundbites (i.e., for self-report purposes), they were
asked on the final questionnaire to describe two or three “highlights” of the debate.
We also shortened the final questionnaire a bit from the first study, in part to give
the participants more time to complete this new question.
Results and Discussion
The manipulation of whether or not participants completed mid-debate ratings
had no effect on any of the results. We therefore report only the main effects of the
manipulation of the presence or absence of the soundbites and audience reaction.
Overall Performance
As in our previous study, the postdebate ratings of each candidate’s performance constituted the primary dependent measure. Consistent with the results of
Experiment 1, the relative ratings of the candidates in this study varied as a
function of our manipulation. As can be seen in Figure 2, in this study Reagan’s
and Mondale’s performances were judged virtually equally in the control condition, but Reagan tended to be rated more negatively than Mondale in the other two
conditions—particularly in the condition in which the soundbite remained but the
audience reaction was deleted.
The ratings of each candidate’s performance were subjected to separate 3
(soundbite manipulation) ¥ 2 (obtaining measures twice versus only after the
debate) ANOVAs. The ANOVA revealed a significant main effect for the soundbite
manipulation on ratings of Reagan, F (2, 101) = 3.12, p < .05. More specifically,
Reagan received significantly higher ratings in the control condition than in the
reaction-deleted condition (Ms = 59.32 & 48.09, respectively; p < .05 via
Newman-Keuls test; M = 52.81 in the soundbite-deleted group). In contrast to
Experiment 1, however, ratings of Mondale’s performance were not significantly
affected by the manipulation, F < 1.
Candidates’ Qualities
As in Experiment 1, our manipulation influenced ratings of Reagan’s character. On the same composite measure used in our prior studies (combined ratings of
Presidential Debates
177
Figure 2. Postdebate performance ratings of Reagan and Mondale in Experiment 2 as a function of
whether the clips contained the soundbites and the audience reaction (SB-AR), the soundbites with
no audience reaction (SB-NR), or no soundbites (NSB).
intelligence, sense of humor, likeability, leadership, strength, competence, sincerity), there was a significant main effect, F (2, 101) = 3.21, p < .05, as Reagan was
viewed more favorably in the control condition than in the reaction-deleted condition (Ms = 65.94 and 55.85, respectively; p < .05 via Newman-Keuls test;
M = 62.03 in the soundbite-deleted condition). This pattern was particularly strong
in ratings of Reagan’s intelligence, sense of humor, and likeability. Once again in
contrast to the results of Experiment 1, the effect of the soundbites on evaluations
of Mondale’s character was not statistically significant, F (2, 101) = 2.79, p < .10.
Were Soundbites Perceived as Significant?
Finally, participants were asked on the postdebate questionnaire to describe
two or three highlights from the debate. These free-response data were coded for
whether or not the participants included a soundbite on the list. The result was
informative. Demonstrating the profound importance of audience reactions, 78%
of all participants in the control condition included a soundbite among the highlights, compared to only 15% of those in the reaction-deleted condition, X2 (2,
N = 74) = 29.17, p < .001. In short, participants were impressed by the soundbites
only when the remarks were followed by the strong reactions from the audience.
178
Fein et al.
Summary
Supporting the results of the first experiment, participants in this study tended
to offer very different relative ratings of the two candidates as a function of the
presence or absence of the soundbites and the reaction to them. Although the
results are more muted than in the first study, perhaps due to the extra 50 minutes
of material—much of it quite dry—diluting the impact of these critical moments,
it was still the case that participants’ relative judgments were affected significantly.
Whereas participants in the control condition perceived a very close debate, those
in the condition in which the reactions to Reagan’s two soundbites were deleted—
representing less than a minute of time in a 90-minute debate about some of the
most important issues facing the country—saw Mondale clearly outperforming
Reagan. Moreover, absent the applause, laughter, and general approval of
Reagan’s one-liners, these responses were not seen as particularly noteworthy by
the participants. With these audience reactions intact, however, the majority of the
participants saw them as a highlight of the debate.
Experiment 3
Although pivotal soundbites stand out remarkably well in debate lore, they are
rare, and often debates are won or lost in their absence. In the past several election
cycles it is hard to locate defining moments in debates, even in debates with clear
winners. The first Bush-Kerry debate of 2004, for example, was judged a clear
Kerry win despite its lack of “zingers.” Kerry’s successes were seen holistically
(as were his opponent’s foibles). The same could be said of Kennedy’s performance in the first debate of 1960. Therefore, conformity should also be examined
in the—arguably more common—context of these victories by attrition.
Experiment 3 used segments of the 1984 debate in that light using a 10-minute
tape that excludes the critical soundbites from the previous two experiments.
Unlike in Experiments 1 and 2, this study did not use studio audience behavior as
the independent variable. Rather, participants received false feedback concerning
the alleged reactions of their peers in the room with them as they watched the
debate together in a group.
The feedback was administered using a technology first displayed during the
televised debates of 1992. For those events, the cable television news network,
CNN, showed viewers continuous, real-time responses of focus-group members
who were watching the debate in an auditorium and using hand-held response dials
to record their changing impressions of the event. This computer-based technology
thus provided TV viewers with real-time, public-opinion poll data. We used this
technology to consistently insert audience feedback into each participant group,
with participants being told that the displayed “data” were their own. This procedure enabled us to assess whether judgments of the candidates would be influenced
by exposure to the group norm.
Presidential Debates
179
Method
Participants and Design
Ninety-four introductory psychology students participated in this study for
course credit. Participants were run in groups of approximately 15–20 students.
The groups were randomly assigned to either a pro-Reagan or pro-Mondale false
feedback condition.
Procedure
The stimulus tape used in this study was a 10-minute segment consisting of
excerpts of the second 1984 presidential debate between Reagan and Mondale.
Upon their arrival at the laboratory, participants completed a brief predebate
questionnaire and were then each given a wireless hand-held dial. Each dial was
equipped with a digital display that indicates the current numerical setting—in this
case, it was set to range from 0 (with the dial hand pointed to the far left) to 100
(with the dial hand pointed to the far right), and having a midpoint of 50 (with the
hand pointed up at a 90-degree angle). Participants were told that they would see
videotaped excerpts of a presidential debate and were instructed to set the dial at
50 and move it up or down to varying degrees to indicate their changing opinions
during the debate. Before seeing the debate, participants got some practice using
their dials, which also served to allay potential suspicions they may have had about
the validity of these dials—as the experimenters demonstrated how they could tell
what each participant’s dial was set to at any given moment.
Participants in both groups were informed that they would receive continuous
real-time feedback about their group’s average opinion in the form of a line graph
superimposed over the debate videotape. In fact, the feedback they received was
false, pre-programmed to indicate that Reagan or Mondale was gaining in support
over the other. In both groups, participants saw the same line graph begin at the
neutral midpoint of 50 and move gradually, in a fluctuating pattern, to a final value
of 85. For those in the pro-Reagan group, the 0–100 point response scale was
defined in such a way that 50 meant that neither candidate was outperforming the
other, that numbers higher than 50 meant that Reagan was outperforming
Mondale, and that numbers lower than 50 meant that Mondale was outperforming Reagan; for those in the pro-Mondale condition, the scale was reversed.
To guard against the possibility that a participant would turn his or her dial down
to 0 or up to 100 and become suspicious of the false feedback manipulation when
it appeared not to affect the group’s average, participants were told that the most
extreme high and low scores would always be excluded from the computation.4
4
It perhaps should be noted that participants found these dials to be easy to use and did not find them,
or the superimposed graph, to be particularly distracting. The first author has used this technology in
180
Fein et al.
After watching the tape, participants completed the same postdebate questionnaire as that used in the first experiment.
Results and Discussion
Overall Performance
Supporting the results of the first two studies, the results of Experiment 3
indicated that others’ reactions can have a significant effect on individuals’ judgments of the candidates’ overall performance and personal qualities. As can be
seen in Figure 3, participants who saw a graph suggesting that their peers saw
Reagan as winning the debate rated Reagan’s performance more than 15 points
better than Mondale’s, whereas they rated Reagan’s performance more than 20
points worse than Mondale’s if they saw a graph suggesting that their peers saw
Mondale perform better, resulting in a net difference of about 36 points, F(1,
92) = 51.74, p < .0001.
80
70
Reagan
60
Mondale
50
40
Pro-Reagan Feedback
Pro-Mondale Feedback
Figure 3. Postdebate performance ratings of Reagan and Mondale in Experiment 3 as a function of
whether participants were exposed to pro-Reagan or pro-Mondale feedback.
several studies in subsequent years (e.g., Fein, Hoshino-Browne, Davies, & Spencer, 2003), and it is
extremely rare for any participants to be suspicious of or very distracted by it.
Presidential Debates
181
On ratings of each candidate’s debate performance, the results again supported the hypothesis that participants would be influenced by the false normative
feedback concerning the opinions of others in the room. Reagan’s performance
ratings were significantly higher in the pro-Reagan feedback condition than in the
pro-Mondale condition (Ms = 67.98 versus 48.81), F (1, 92) = 31.67, p < .001. In
contrast, Mondale’s ratings were significantly higher in the pro-Mondale condition
than in the pro-Reagan condition (Ms = 69.17 versus 52.69), F (1, 92) = 30.03,
p < .0001.
A chi-square analysis on participants’ responses to the questions of who won
the debate (Reagan, Mondale, or neither) yielded additional support for our
hypothesis. Participants in the pro-Reagan condition were significantly more
likely to judge Reagan to be the winner of the debate (67.31% versus 7.69% for
Mondale and 25% for neither) than were participants in the pro-Mondale condition
(11.91% versus 71.43% for Mondale and 16.67% for neither), X2 (2, N =
94) = 43.61, p < .0001.
Further indicating the strength of the audience reaction effect was that there
was also a significant and sizable difference on the key question, “who do you
think you would have voted for?” (Reagan, Mondale, or neither), X2 (2, N =
94) = 9.88, p < .008. Participants in the pro-Reagan condition were significantly
more likely to indicate they would have voted for Reagan (46.15% versus 34.62%
for Mondale and 19.23% for neither) than were participants in the pro-Mondale
condition (16.67% versus 61.91% for Mondale and 21.43% for neither).
Candidates’ Qualities
As in the first two studies, audience reaction also had a marked effect on
perceptions of each candidate’s personal characteristics. On a composite measure
that combined mean ratings of intelligence, likeability, leadership, sense of humor,
strength, competence, and sincerity (a = .91), participants evaluated Reagan more
favorably in the pro-Reagan feedback group than in the pro-Mondale group
(Ms = 67.02 versus 58.57), F (1, 92) = 7.18, p < .009. Similarly, Mondale
(a = .86) was rated more favorably in the pro-Mondale condition than in the
pro-Reagan group (Ms = 65.50 versus 56.58), F (1, 92) = 14.52, p < .0003. Indicating the broad range of effects produced by audience reaction, individual analyses revealed that the two groups differed significantly in their ratings of Reagan’s
likeability, leadership, and sense of humor, and in their ratings of Mondale’s
likeability, leadership, intelligence, competence, strength, and sincerity (all at
p < .05). On estimates of how well the two candidates fared among voters in
general, audience reaction also affected mean estimates of Reagan’s popularity
(Ms = 74.23 and 57.98, respectively, in the pro-Reagan and pro-Mondale groups),
F (1, 92) = 35.58, p < .0001, as well as Mondale’s (Ms = 61.55 and 44.33, respectively, in the pro-Mondale and pro-Reagan groups), F (1, 92) = 37.56, p < .0001.
182
Fein et al.
Self-Reported Effects of Feedback
Finally, participants rated the extent to which their evaluations of the candidates were influenced by the online normative feedback (i.e., on a 0–100 point
scale, whether it led them to favor Reagan or Mondale, where 50 = no effect).
Interestingly, the pro-Reagan and pro-Mondale conditions did not differ on this
measure, F (1, 92) = 1.47, ns. With a combined mean rating of 51.20, it appears
that participants in general believed that they were not influenced by the feedback.
This self-reported lack of influence thus contrasts sharply with all of the above
results.
Summary
The results of Experiment 3 indicated that evaluations of the 1984 presidential
candidates were significantly influenced by fabricated real-time feedback in the
form of a line graph allegedly indicating the opinions of their peers in the room
with the participants. For example, seeing the graph of their peers’ alleged reactions changed the percentage of participants who thought Mondale had, in their
own minds, won the election from fewer than 8% to more than 71%. Overall, these
results, which were striking in their breadth and magnitude, suggest that the
practice of presenting TV viewers with continuous focus-group data (or, as can be
expected in upcoming elections, the practice of seeing others’ real-time reactions
on the internet as people watch the debate) constitutes a powerful source of social
influence information.
Experiment 4
The previous experiments all showed significant conformity effects. However,
the 1984 presidential election took place years before these studies were run. The
participants knew the outcome of the election, and their concern about the accuracy of their judgments might not have been particularly high. One might presume
that when citizens are watching an actual debate live, and sizing up the candidates
in order to decide how to vote, they may be much less influenced by the apparent
evaluations of others. Furthermore, there are other factors that might reduce
conformity effects in this situation—viewers may have well-formed impressions
of candidates, been subjected to months of predebate spin, been influenced more
strongly by personal biases, etc. In short, there may be something fundamentally
different about a debate preceding an election in which one will actually vote.
Experiment 4 was intended to address this concern by having research participants
evaluate a presidential debate on the night it actually took place, during the height
of the campaign. If strong conformity effects were found under these circumstances, a host of potential concerns relating to external validity could be
dismissed.
Presidential Debates
183
Participants and Design
Sixty-one introductory psychology students participated in exchange for
course credit. Participants were randomly scheduled to appear in one of three
classrooms to watch the third and final presidential debate of that year’s campaign.
By random assignment, one room was designated as pro-Bush, a second as
pro-Clinton, and a third as neutral.5
Procedure
The stimulus tape used in this study was the third and final 1992 presidential
debate involving George Bush, Bill Clinton, and Ross Perot. Upon recruitment, all
participants were asked to refrain from watching the debate live that evening, as
they would see it just minutes after its conclusion as part of the study (indeed, they
were scheduled to come to the lab several minutes before the debate would
conclude). This was a mere 15 days prior to the election. As soon as they arrived
at their assigned room, all participants completed a brief predebate questionnaire.
The entire 90-minute debate tape was then played simultaneously in the three
rooms. Afterward, all participants were administered a postdebate questionnaire,
fully debriefed, and thanked for their participation.
In this study, social influence was manipulated via audience reactions. In a
pro-Bush group, 21 participants watched the debate in the company of 13 student
confederates who had been instructed to quietly but audibly applaud statements by
Bush and disapprove (hiss, jeer) statements by Clinton. In a pro-Clinton group, 20
participants were joined by 12 confederates who were told similarly to applaud
Clinton and disdain Bush. All confederates were rehearsed to react in ways that
seemed natural. Their reactions increased gradually during the course of the
debate, but they were cautioned not to interfere with the participants’ ability to
hear the tape. In a third control group, there were 20 participants and no
confederates.
The dependent measures were similar to those used in the previous experiments, with the notable exception that in this study, at the end of the postdebate
questionnaire participants were asked also to estimate the mean 0–100 point
rating that each candidate would receive (1) from the other participants in the
room, and (2) from viewers all over the country. The participants were asked to
indicate which of the candidates, if any, they planned to vote for on Election
Day.
5
Although he also participated in the debate, third-party candidate Ross Perot was not considered to
have a reasonable chance of even coming close to winning the 1992 election, and so we focused our
manipulations on the two major candidates.
184
Fein et al.
Results and Discussion
Overall Performance
As shown in Figure 4, the effect of audience reaction on the relative performance ratings of Clinton and Bush was striking. In the control group, Clinton’s
rating was 23 points higher than Bush’s (and, indeed, Clinton was seen as the
convincing winner of the debate by analysts and opinion polls). Yet the difference
was up to 51 points in the pro-Clinton group and down to only 6 points in the
pro-Bush group, F(2, 58) = 20.59, p < .0001.
On ratings of each candidate’s debate performance, the results indicated that
the effects of audience reaction were significant. Specifically, Clinton’s ratings
were higher among participants in the pro-Clinton audience (M = 82.75) than in
the pro-Bush (M = 67.14) and control groups (M = 69.50), F(2, 58) = 12.99,
p < .0001. In contrast, Bush’s ratings were higher in the pro-Bush audience
(M = 60.95) than in the pro-Clinton (M = 31.50) and control groups (M = 46.25),
F(2, 58) = 9.94, p < .0002.
Participants’ postdebate voting preferences were relatively consistent with
these results, but the chi-square did not approach significance, X2 (8, N = 61)
Figure 4. Post-debate performance ratings of Bush and Clinton in Experiment 4 as a function of
whether confederates were exposed to pro-Bush, pro-Clinton, or no (Control) confederates.
Presidential Debates
185
= 6.95, ns, perhaps in part because there were five rather than only two options for
participants to choose—for Clinton, for Bush, for Perot, did not know, or did not
plan to vote.
There were two additional and interesting results on these overall evaluations
of the candidates. First, questionnaires were administered to the 13 pro-Bush and
12 pro-Clinton confederates, and an analysis of their responses indicated that they
too were significantly influenced by audience reaction in their ratings of both
Bush, F(1, 23) = 10.91, p < .01, and Clinton, F(1, 23) = 6.93, p < .02. Despite
knowing that audience reaction was systematically manipulated—indeed, despite
the fact that they themselves had to vary their behavior according to instruction—
the confederates exhibited the same effect. Second, although we manipulated
audience reactions only to Bush and Clinton, ratings of Perot’s performance were
also affected somewhat by the manipulation, F(2, 58) = 2.63, p < .09. Reflecting
the unanticipated fact that Perot had aligned himself more with Clinton in this
debate than with Bush, ratings of Perot were higher in the pro-Clinton group
(M = 76.75) than in the pro-Bush or control conditions (Ms = 65.00 and 66.10,
respectively), F(1, 58) = 5.21, p < .02.
Candidates’ Qualities
On the more specific evaluation measures, the results followed a similar
pattern as the overall performance ratings. Combining each candidate’s performance ratings on the economy, foreign affairs, education, health care, and crime,
results showed that Bush’s mean ratings (a = .83) were higher in the pro-Bush
group than in the other conditions, F(2, 58) = 3.70, p < .05, although no comparable difference was found in ratings of Clinton on this same constellation of
issues (F < 1). It is interesting that although audience reactions had no effect on the
Clinton composite measure (a = .73), participants did rate Clinton more favorably
on economic issues in the pro-Clinton group than in the other conditions, F(2,
58) = 5.17, p < .01. In fact, the effect of audience reaction on ratings of Bush was
also strongest on his economy performance, F(2, 58) = 9.41, p < .001. Although
this pattern was not predicted, the likely reason for it is that the debate focused
primarily on economic issues.
Overall, audience reaction effects were even more pronounced on ratings of
each candidate’s personal characteristics. Participants in the pro-Bush condition,
relative to those in the pro-Clinton and control groups, saw Bush as more trustworthy (F(2, 58) = 3.60, p < .05), more intelligent (F(2, 58) = 9.64, p < .001),
more caring (F(2, 58) = 3.51, p < .05), and as having a better sense of humor (F(2,
58) = 3.22, p < .05). Similarly, participants in the pro-Clinton condition relative to
those in other groups saw Clinton as more trustworthy (F(2, 58) = 5.64, p < .005),
more intelligent (F(2, 58) = 9.59, p < .001), more likeable (F(2, 58) = 4.91,
p < .01), and more mature (F(2, 58) = 2.98, p < .05).
186
Fein et al.
Estimates of Others’ Perceptions
Finally, participants were asked to estimate the mean ratings that each candidate would receive from the other participants in the room, and from all
viewers nationwide. On the first of these measures, the effect was striking in its
magnitude, as participants estimated the highest Bush rating in the pro-Bush
group and the lowest in the pro-Clinton group (Ms = 59.76 and 11.50, respectively; M = 41.00 in the control condition), F (2, 58) = 51.50, p < .0001. The
data were similar for ratings of Clinton, as participants estimated his highest
rating in the pro-Clinton group and his lowest in the pro-Bush group
(Ms = 89.75 and 62.38, respectively; M = 73.75 in the control condition), F (2,
58) = 33.20, p < .0001. Essentially, these results served as a check on the audience reaction manipulation. However, there was also a highly significant effect
on estimates of how the candidates fared more generally among viewers across
the country. On this measure as well, those in the pro-Bush group projected the
rest of the country to be much more positive about Bush’s performance
(M = 60.38) than did those in the control (M = 53.75) or pro-Clinton conditions
(M = 40.00), F (2, 58) = 7.85, p < .001. Similarly, those in the pro-Clinton group
projected the rest of the country to be much more positive about Clinton’s performance (M = 78.25) than did those in the control (M = 69.50) or pro-Bush
conditions (M = 63.33), F (2, 58) = 11.53, p < .0001. Thus, even though all participants saw the same debate (which included an actual live audience), they
inferred that the reactions of the confederates in their sessions were diagnostic
of national public opinion.
Summary
The results from Experiment 4 revealed that manipulated peer audience
reactions to a presidential debate strongly influenced participants’ judgments of
each candidate’s overall performance, performance on the focal substantive
issue, and personal characteristics, even under the most realistic circumstances
possible. The magnitude of this effect was astonishing (it produced a 45-point
shift in the relative ratings of Bush and Clinton), and was obtained not only with
naive participants who were low in political involvement, but with highinvolvement participants—and even confederates who knew that the cheers and
jeers were experimentally manipulated. This effect occurred even though the
stimulus event was a heated 90-minute presidential debate and despite the presence of a live, reactive (i.e., at the debate) audience. Indeed, our manipulation
led participants not only to alter their own judgments of the candidates, but to
infer that these judgments were diagnostic of national public opinion, as was
also seen in Experiment 3.
Presidential Debates
187
General Discussion
The results of the four studies reported here were highly consistent. In each of
them participants’ judgments about debate performance were strongly influenced
by their mere exposure to other people’s reactions. The magnitude of these effects
is striking. Our participants made very different judgments about debate performance as a function of the information we manipulated concerning the reactions
of others in the audience. In Experiment 1, participants watched Mondale and
Reagan. In one condition the audience reaction to Reagan’s winning “one-liners”
was removed. Compared to the control condition where the audience reaction was
presented intact, Reagan’s ratings relative to Mondale’s declined by 31 points. The
results of Experiment 2 were also significant despite the fact that the manipulation
was embedded within the full 90-minute debate. In Experiments 3 and 4, participants saw what they believed to be the reactions of their peers who were watching
the debate with them. In Experiment 3, as a function of this false group feedback,
Reagan’s ratings relative to Mondale’s varied by 30 points. In Experiment 4,
participants’ judgments about overall performances in a 1992 Bush-Clinton-Perot
debate, on the night of the live debate itself, were strongly influenced by whether
they watched the debate with pro-Clinton or pro-Bush confederates who audibly
vocalized their opinions. On a 101-point scale, Clinton’s advantage over Bush
increased by 45 points across conditions. Taken together, these studies all demonstrated that mere exposure to the judgments of others strongly influenced our
participants.
One striking feature of the results is that they were highly similar across a
number of dimensions. First, information about others’ judgments was presented
in a variety of ways: the recorded laughter and cheers of audiences attending actual
debates, the visual fabricated feedback of other group members, or the vocal
reactions of live confederates. Second, audience reactions significantly affected
perceptions of candidate performance across a broad range of participants, including Democrats and Republicans, those who defined themselves as politically
involved or uninvolved, and those who were high or low in need for cognition. In
each case there were similar effects for the different kinds of participants.6 Third,
effects were obtained across a range of dependent measures, including overall
performance ratings as well as perceptions of specific qualities such as intelligence
and likeability. Fourth, participants’ judgments about a presidential debate in a
current campaign cycle were influenced as much as, or even slightly more than,
6
It should be noted that the participants in our studies all were college undergraduates, and therefore
we cannot know whether older samples of people would respond similarly. We have conducted one
study in a different line of work concerning social context and presidential debates in which our
participants came from a large sample of public-school teachers, and the results were similar to those
we found with undergraduate samples. It is clear, though, that including older and broader samples of
participants should be an important goal for future research.
188
Fein et al.
their judgments about debates that took place nearly a decade earlier. The consistency of the influence effects is striking.
Our participants seemed to exhibit informational social influence and internalized conformity, reflecting, in Insko, Drenan, Solomon, Smith, and Wade’s
(1983) terms, a desire to be right rather than a desire to be liked. Their judgments
were given privately and anonymously. In Experiments 1 and 2 the source of
influence was simply a videotaped audience from a debate which had taken place
years earlier. In those cases, participants had no reason to be concerned about
pressure from others or others’ opinions about them. Importantly, then, the influence we obtained seems to be informational social influence affecting privately
held judgments. This is precisely the kind of influence one would expect in a
situation where physical reality is ambiguous but social reality is clear. Our
participants probably believed that there was in fact a right answer to questions
about how the candidates performed—and that the reactions of others signaled that
clear answer.
What is a likely account of the process by which informational social influence was produced in these studies? It may be that the identification and interpretation of specific behaviors in debates were influenced by others’ reaction. When
Ronald Reagan responded to a very serious question about his energy and intellectual capacity by joking that he would not exploit his opponent’s youth and
inexperience, there was room for considerable interpretation. Observers could
identify his reply as a feeble or even disgraceful attempt to dodge the issues or as
a deft and masterful turning of a hostile question to his advantage. The audience
reaction of strong approval did not recognize Reagan’s quip as a knockout punch
so much as it made it one. Throughout a 90-minute debate there are numerous
segments of behavior (Newtson, 1976), each of which can be framed and defined
by the reactions of others. Furthermore, once a particular expectation is adopted
about how each candidate is performing in a debate, individual behaviors can be
interpreted within that framework (Fiske & Taylor, 1991; Kelley, 1950; Kugler &
Goethals, 2005; Norton & Goethals, 2004). Positive or negative moments can
therefore define entire debates, or even campaigns.
It is perhaps worth underlining the fact that the audiences who produced the
influence in these studies were most likely perceived as having no intent to
influence. This is most clearly the case in Experiments 1 and 2 where the audience’s reactions were televised years before. Earlier studies of overheard communications suggest that they are perceived as not being designed to persuade and for
this reason often have more credibility and influence than communications delivered with intent to persuade (Brock & Becker, 1965). They also generate less
psychological reactance (Brehm, 1966).
What are the policy issues that are suggested by this research? Perhaps the
most fundamental is that those who stage, sponsor, and broadcast political debates
should take steps to minimize social influence, or psychologists should take steps
to publicize that influence, or both. Within the current debate format there are at
Presidential Debates
189
least three sources of influence that are under virtually no control. The first is the
reactions of live audiences. Since 1984 live audience reaction has been part of the
mix. In the vice-presidential debate in 1988 Lloyd Bentsen partisans laughed and
cheered at his “You’re no Jack Kennedy” put-down of Dan Quayle, while the
latter’s supporters competed with a loud chorus of boos. In several of the recent
debates, the moderator has strongly admonished the audience to remain quiet.
Since audience reaction has such a strong effect on viewers, this warning is well
justified.
A second possible source of influence is the behavior of the commentators
who moderate debates and ask questions. In 1984 journalist Henry Trewhitt
commented on Reagan’s age quip by saying “I’ll try to run to the fence to catch
that one before it goes out,” just in case anyone missed the fact that Reagan had
hit a home run. In 1988, Jim Lehrer implicitly supported Bush’s attempts to
make light of his persistent errors by quoting and endorsing the vice-president’s
“nobody’s perfect” response to Michael Dukakis’ derisive response to Bush’s
misstatements. Also in 1988, NBC newsman Tom Brokaw’s headshakes and
nods influenced Dan Quayle’s response as he walked into Lloyd Bentsen’s
“You’re no Jack Kennedy” put-down. Broadcasters should exercise caution and
restraint over the subtle ways in which they guide, diminish, or enhance a
debater’s remarks.
Third, networks now report on the results of instant polls and focus groups
that may have a significant effect on people’s judgments about how candidates
performed and who won. Some networks also give high-ranking campaign officials free reign to apply their “spin” to the evening’s debate performances. All this
information is consumed by viewers before they have had a chance to consolidate,
deliberate, and make up their own minds about the large amount of content they
have just seen. The potential for influence is therefore all the stronger. In this
regard, the news media might be more thoughtful about their role in channeling
social realities to viewers. While the news media may have no conscious intent to
influence, our research suggests that they may have great influence nonetheless.
Their good intentions aside, the power of the networks needs to be considered.
We can expect in future debates that communication via the internet will play
an increasingly important role. It is easy to imagine that very large numbers of
citizens, particularly younger voters, will provide spontaneous audience reactions,
along the lines of what we manipulated in Experiments 3 and 4, by posting to
blogs, electronic forums and boards, and so on. The normative information that
will be provided in this manner has tremendous potential to influence the perceptions and decisions of many individuals. When information such as this, or such as
the immediate postdebate spins presented by analysts and campaign officials,
bombards individuals before they have had the chance to reach their own interpretations and conclusions independently, individual voters may be unaware that
the privacy of their voting booth has been compromised by the long reach of the
opinions of a great many friends and strangers.
190
Fein et al.
In conclusion, we have shown that people’s judgments of presidential debates
can be strongly influenced by their knowledge of other people’s reactions. These
debates appear to be highly ambiguous events. For people to develop their own
independent opinions, great care must be taken on the part of organizers and
sponsors to limit the impact of irrelevant cues like the ones discussed herein. One
potentially useful direction for future research is to examine the factors that would
make people less vulnerable to such effects. Would being educated about the
results of studies such as the present ones reduce individuals’ susceptibility to the
kind of informational social influence we have documented? Of course, other
people can and often do provide important, useful, and even correct information.
In the theater of political campaigns, however, with spin doctors, hand-picked
audiences, and highly charged emotions galore, this may be less likely to be the
case than is usual. The often-adaptive practice of checking in with social reality
can be a trap in these orchestrated contexts. What is especially important, therefore, is for individuals to look at the social context around these debates with
vigilance and caution.
ACKNOWLEDGMENTS
We would like to thank Amelia Cottrell, Jessica Cross, Leigh Frost, C.J.
Gillig, and Lauren Parkhill for their help in collecting the data presented in this
paper. We would also like to thank Columbia Information Services for their
assistance with the Perception Analyzer technology used in Experiment 3. Correspondence concerning this article should be addressed to Steven Fein, Department
of Psychology, Williams College, Williamstown, MA 01267, or Al Goethals,
Jepson School of Leadership Studies, University of Richmond, Richmond, VA
23173. E-mail: sfein@williams.edu and ggoethal@richmond.edu.
REFERENCES
Asch, S. E. (1951). Effects of group pressure upon the modification and distortion of judgments. In H.
Guetzkow (Ed.), Groups, leadership, and men. Pittsburgh: Carnegie Press.
Axsom, D., Yates, S., & Chaiken, S. (1987). Audience response as a heuristic cue in persuasion.
Journal of Personality and Social Psychology, 53, 30–40.
Baron, R. S., Vandello, J. A., & Brunsman, B. (1996). The forgotten variable in conformity research:
Impact of task importance on social influence. Journal of Personality & Social Psychology, 71,
915–927.
Brehm, J. W. (1966) A theory of psychological reactance. New York: Academic.
Brock, T. M., & Becker, L. A. (1965). Ineffectiveness of “overheard” counterpropaganda. Journal of
Personality and Social Psychology, 2, 654–660.
Cacioppo, J. T. & Petty, R. E. (1982). The need for cognition. Journal of Personality and Social
Psychology, 42, 116–131.
Cantril, H. (1940). The invasion from Mars. Princeton: Princeton University Press.
Presidential Debates
191
Deutsch, M., & Gerard, H. B. (1955). A study of normative and informational social influences upon
individual judgment. Journal of Abnormal & Social Psychology, 51, 629–636.
Fein, S., Frost, L. A., Goethals, G. R., & Kassin, S. M. (1994). The effects of expectations on
perceptions of presidential debate performance. Presented at the 102nd meeting of the American
Psychological Association, Los Angeles.
Fein, S., Hoshino-Browne, E., Davies, P. G., & Spencer, S. J. (2003). Self-image maintenance goals and
sociocultural norms in motivated social perception. In S. J. Spencer, S. Fein, M. Zanna, & J. M.
Olson (Eds.), Motivated social perception: The Ontario symposium (vol. 9, pp. 21–44). Mahwah,
NJ: Erlbaum.
Fiske, S. T., & Taylor, S. E. (1991). Social cognition. New York: McGraw-Hill.
Germond, J. W., & Witcover, J. (1989). Whose broad stripes and bright stars? The trivial pursuit of the
presidency 1988. New York: Warner.
Insko, C. A., Drenan, S., Solomon, M. R., Smith, R., & Wade, T. J. (1983). Conformity as a function
of the consistency of positive self-evaluation with being liked and being right. Journal of
Experimental Social Psychology, 19, 341–358.
Jamieson, K. H., & Waldman, P. (2002). The press effect: Politicians, journalists, and the stories that
shape the political world. Cambridge: Oxford University Press.
Kaid, L. L., & Bystrom, D. G. (1999). The electronic election: Perspectives on the 1996 campaign
communication. Mahwah, NJ: Erlbaum.
Kelley, H. H. (1950). The warm-cold variable in first impressions of persons. Journal of Personality,
18, 431–439.
Kiesler, C. A., & Kiesler, S. B. (1969). Conformity. Reading, MA: Addison-Wesley.
Kraus, S. (1962). The great debates: Background, perspectives, effects. Bloomington: Indiana University Press.
Kugler, M. B., & Goethals, G. R. (2005). Trait-focused spin in presidential debates: Surviving the
kisses of death. Unpublished manuscript, Williams College.
McKinnon, L. M., & Tedesco, J. C. (1999). The influence of medium and media commentary
on presidential debate effects. In L. L. Kaid & D. G. Bystrom (Eds.), The electronic
election: Perspectives on the 1996 campaign communication (pp. 191–206). Mahwah, NJ:
Erlbaum.
McKinnon, L. M., Tedesco, J. C., & Kaid, L. L. (1993). The third 1992 presidential debate: Channel
and commentary effects. Augmentation and Advocacy, 30, 106–118.
Newtson, D. (1976). Foundations of attribution: The perception of ongoing behavior. In J. H. Harvey,
W. J. Ickes, & R. F. Kidd (Eds.), New directions in attribution research (pp. 223–247). New York:
John Wiley and Sons.
Norton, M. I., & Goethals, G. R. (2004). Spin (and pitch) doctors: Campaign strategies in televised
political debates. Political Behavior, 26, 227–248.
Petty, R. E., & Cacioppo, J. T. (1986). Communication and persuasion: Central and peripheral routes
to attitude change. New York: Springer-Verlag.
Schroeder, A. (2000). Presidential debates: Forty years of high-risk TV. New York: Columbia
University Press.
Sears, D. O., & Chaffee, S. H. (1979). Uses and effects of the 1976 debates: An overview of the
empirical studies. In S. Kraus (Ed.), The great debates: Carter versus Ford, 1976. Bloomington:
Indiana University Press.
Sherif, M. (1936). The psychology of social norms. New York: Harper.
Sigelman, L. (1992). There you go again: The media and the debasement of American politics.
Communication Monographs, 59, 407–410.
Sigelman, L., & Sigelman, C. K. (1984). Judgments of the Carter-Reagan debate: The eyes of
thebeholders. Public Opinion Quarterly, 48, 624–628.
192
Fein et al.
Steeper, F. T. (1978). Public response to Gerald Ford’s statements on Eastern Europe in the second
debate. In G. F. Bishop, R. G. Meadow, & M. Jackson-Beeck (Eds.), The presidential debates:
Media, electoral, and policy perspectives (pp. 81–101). New York: Praeger.
Zarevsky, D. (1992). Spectator politics and the revival of public argument. Communication Monographs, 59, 411–414.
Purchase answer to see full
attachment