Unformatted Attachment Preview
Communication Modality as a Determinant of Persuasion:
The Role of Communicator Salience
Shelly Chaiken
Vanderbilt University
Alice H. Eagly
Purdue University
In two experiments, a likable or unlikable communicator delivered a persuasive
message via the written, audiotaped, or videotaped modality. In both studies, the
likable communicator was more persuasive in videotaped and audiotaped (vs.
written) conditions, whereas the unlikable communicator was more persuasive
in the written modality. Because of these opposing effects of modality on the
persuasiveness of likable and unlikable communicators, communicator likability
was a significant determinant of persuasion only in the two broadcast modalities.
Additional findings suggested that subjects engaged in more processing of communicator cues when exposed to videotaped or audiotaped, compared to written,
messages and that communicator-based (vs. message-based) cognitions predicted
opinion change primarily within videotape and audiotape (vs. written) conditions.
It was concluded that the videotaped and audiotaped modalities enhance the
salience of communicator-related information with the consequence that communicator characteristics exert a disproportionate impact on persuasion when
messages are transmitted in videotaped or audiotaped (vs. written) form. The
relevance of the findings to understanding "vividness" phenomena is also discussed.
Previous research on media effects in persuasion has proven largely inconsistent. Although the modal finding in this literature
appears to be that live or videotaped messages
induce greater opinion change than do audiotaped messages (Frandsen, 1963), which,
in turn, induce greater change than do written communications (Cantril & Allport, 1935;
Haugh, 1952; Knower, 1935, 1936; Wilke,
1934), a number of studies have obtained
either no differences in persuasiveness as a
function of communication modality or
greater opinion change with written messages
(McGinnies, 1965; Tannenbaum & Kerrick,
1954; Werner, 1978; Werner &Latane, 1976).
In an earlier article (Chaiken & Eagly,
1976), we suggested that at least some of the
inconsistency in the modality literature might
be explained in terms of message compreWe thank Eugene Borgida, Patricia Sorce, and Wendy
Wood for their comments on a draft of this article and
David Gibo for delivering one of the persuasive messages.
Requests for reprints should be sent to Shelly Chaiken,
Department of Psychology, Wesley Hall, Vanderbilt University, Nashville, Tennessee 37240.
hensibility. Based on research indicating that
good comprehension of persuasive argumentation often facilitates opinion change (e.g.,
Eagly, 1974; Eagly & Warren, 1976) and research suggesting that written (vs. videotaped
or audiotaped) messages enhance comprehension, especially for complex material (e.g.,
Beighley, 1952; Harwood, 1951; Westover,
1958; Wilson, 1974), we predicted and found
that modality differences in persuasion favored the written mode with inherently.difficult-to-understand material, whereas the
more usual advantage of videotaped and audiotaped modalities was found with easy
material.
Although the Chaiken and Eagly (1976)
study demonstrates that message comprehensibility can moderate the persuasive impact
of modality, two facts suggest that comprehensibility differences alone cannot fully explain inconsistencies in this literature. First,
the slope of the function relating message
comprehension to persuasion is relatively flat
(Calder, Insko, & Yandell, 1974; Insko, Lind,
& LaTour, 1976). Thus, a fairly large difference in comprehension would be required to
Journal of Personality and Social Psychology, 1983, Vol. 45, No. 2, 241-256
Copyright 1983 by the American Psychological Association, Inc.
241
242
SHELLY CHAIKEN AND ALICE H. EAGLY
exert a detectable effect on persuasion (Eagly,
1978). Second, message comprehension may
be an important determinant of opinion
change only in persuasion contexts that maximize the tendency for recipients to focus
attentively on persuasive argumentation and
to base their opinions on such content cues
(Chaiken, 1978, 1980; Chaiken & Eagly,
1976; Miller, Maruyama, Beaber, & Valone,
1976).
To further explore media effects in persuasion, the present research examined the
idea that the persuasive impact of communication modality is contingent on the valence of a communicator's personal characteristics or "image."1 The persuasive impact
of communicator characteristics should be
affected by modality because nonverbal communicator-related cues are present in videotaped and audiotaped messages but are
absent in written messages. Because these
nonverbal cues (e.g., the communicator's
physical appearance and/or voice) may draw
message recipients' attention to the communicator, his or her personal characteristics
may be more salient and therefore exert a
disproportionate impact on persuasion when
messages are transmitted in videotaped or
audiotaped rather than written form. This
idea is consistent with research on the salience of persons, which indicates that salient
individuals are remembered better, viewed as
more causally prominent, and have their personal attributes rated more extremely
(McArmur, 1981; Taylor &Fiske, 1978; Taylor & Thompson, 1982). Whether the greater
persuasive impact of communicator cues in
videotaped and audiotaped modalities is positive or negative should depend on whether
the communicator cues made salient are
themselves positive or negative. For positive
cues conveying, for example, that a communicator is likable or expert, increased salience should enhance persuasiveness. However, for negative cues conveying that a communicator is unlikable or inexpert, increased
salience should decrease persuasiveness.
Assuming that communicators with positive attributes are generally chosen to deliver
messages in persuasion experiments, the
communicator salience argument implies
that the modal finding in past literature
should be (as it is) greater persuasion for vid-
eotaped and audiotaped (vs. written) messages. Further, in studies employing communicators whose attributes differ markedly,
the persuasive effect of media should depend
on the valence of these attributes. Consistent
with this expectation, Andreoli and Worchel
(1978) found that a newscaster and a former
politician (presumably trustworthy sources)
were more persuasive on television than over
the radio or in a newspaper story, whereas
a political candidate (a presumably untrustworthy source) was more persuasive in a
newspaper story or over the radio than on
television. Similar though weaker findings
were obtained by Worchel, Andreoli, and
Eason (1975).
Other researchers have discussed media
differences in terms compatible with our
communicator salience argument. For example, Keating and Latane (1976) argued
that a "live speaker's" persuasiveness decreases progressively with televised, oral, and
written presentations because the media become "less rich and transmit less information." Worchel et al. (1975) and Andreoli and
Worchel (1978) suggested that television is
more "involving" than radio or newspaper
and, further, that the more-involving media
highlight communicator characteristics. Also,
Short, Williams, and Christie (1976) argued
that the media vary in "social presence,"
which they defined as the salience of the other
person (e.g., the communicator) and the consequent salience of interpersonal relationships. According to these authors, social presence is greatest in face-to-face communication, followed by videotape, audiotape, and
written communications. Although the notions that media differ in richness, involvement, and social presence are quite global,
they are consistent with the idea that the persuasive impact of modality is contingent on
the valence of communicator characteristics.
Another approach to understanding the
effects of communication media stems from
1
To exclude message reception from consideration as
a mediator of persuasion, the present research employed
relatively easy-to-understand messages (Chaiken & Eagly,
1976) and a cover story ("people's reactions to speeches")
that we hoped would not lead subjects to focus excessively on persuasive argumentation (Chaiken, 1978,
1980).
COMMUNICATION MODALITY AND PERSUASION
recent theorizing and research on "vividness" effects. Vivid information is information that presumably attracts and holds people's attention because it is, for example, concrete and imagery-provoking or proximal in
a sensory, temporal, or spatial way (Nisbett
& Ross, 1980). By this definition, videotaped
and audiotaped modalities present information more vividly than does the written modality. Because vividly presented information
is hypothesized to exert a greater impact on
judgments than nonvividly presented information (Nisbett & Ross, 1980; Taylor &
Thompson, 1982), it might be predicted, as
Taylor and Thompson (1982) recently suggested, that the more vivid videotaped or audiotaped modalities should effect greater persuasion than the less vivid written modality.
Yet, according to our analysis, in many persuasion settings, it is not the persuasive message itself that is made more vivid in videotaped and audiotaped (vs. written) modalities, but, rather, information about the
communicator. Given this clarification, our
ideas about the greater salience and thus the
greater persuasive impact of communicator
characteristics in the broadcast media are, in
fact, consistent with the general hypothesis
that vivid (vs. nonvivid) information exerts
a disproportionate impact on people's judgments.
To develop the communicator salience explanation more thoroughly, it is useful to employ Chaiken's (1978,1980, Note 1) heuristic
versus sytematic analysis of persuasion. In
the systematic view of persuasion, message
recipients focus primarily on persuasive argumentation in forming their opinion judgments, whereas in the heuristic view, recipients engage in little, if any, detailed processing of message content and, instead, tend to
employ simple decision rules often (though
not exclusively; Chaiken, 1980, Note 1) based
on cues such as the communicator's identity
in judging message acceptability. In the initial
test of this analysis, Chaiken (1980) found
that high (vs. low) levels of response involvement or issue involvement tended to foster
systematic (vs. heuristic) information processing: In two experiments, the opinions of
high-involvement subjects were more strongly
influenced by the amount of argumentation
contained in a persuasive message than by
243
the communicator's likability (vs. unlikability), whereas the opinions of low-involvement
subjects were more strongly affected by the
communicator's likability than by the amount
of argumentation he or she provided. As further specified by Chaiken's analysis, opinion
change was found to be mediated primarily
by message-based (vs. communicator-based)
cognitions for high- (vs. low-) involvement
subjects. More recent findings reported by
Petty, Cacioppo, and Goldman (1981) also
supported the heuristic versus systematic
processing distinction.
Of primary interest here is the idea that,
aside from a motivational variable like involvement, contextual features of a persuasion setting, such as communication modality, may influence the degree to which recipients process and utilize communicator cues
in making their opinion judgments. As argued earlier, the broadcast modalities may
increase the salience of communicator cues,
with the consequence that communicator
characteristics should exert a greater (positive
or negative) persuasive impact when messages are transmitted in videotaped or audiotaped, rather than written, form. The heuristic versus systematic analysis implies further that recipients of videotaped or
audiotaped messages should predicate their
opinions primarily on their reactions to the
communicator and less on their reactions to
message content. In contrast, recipients of
written messages (for whom communicator
cues may be less salient than message content) should show a greater tendency to predicate their opinions on their evaluation of
message content.2
To test these ideas, subjects in two experiments were exposed to a videotaped, an audiotaped, or a written persuasive message
from a likable or unlikable communicator.
We anticipated that with a likable communicator, videotaped and audiotaped messages
2
The current predictions are expected to hold primarily for moderately involving message topics such as
the ones employed in the present research. As suggested
by the heuristic versus systematic analysis, when involvement is extremely high, recipients may attend primarily
to message content (vs. communicator cues), regardless
of modality. And, when involvement is extremely low,
recipients may attend primarily to communicator cues,
regardless of modality.
244
SHELLY CHAIKEN AND ALICE H. EAGLY
would be more persuasive than written messages, whereas with an unlikable communicator, videotaped and audiotaped (vs. written) messages would be less persuasive. Because the video and the audio modalities both
differ from the written modality by the addition of nonverbal communicator-relat9d
cues, and because prior research has typically
not found large or systematic persuasion differences between the two broadcast modalities (e.g., Chaiken & Eagly, 1976), we predicted similar findings for videotaped and audiotaped messages. The present research also
attempted to obtain relatively direct evidence
that recipients pay more attention to the
communicator (and perhaps less attention to
message content) with videotaped and audiotaped (vs. written) messages. To accomplish
this goal, a thought-listing task (Brock, 1967;
Osterhouse & Brock, 1970) was included in
both experiments, and subjects' cognitive responses were classified according to whether
they were communicator oriented or message
oriented. Finally, Experiment 2 also explored
opinion persistence. Based on the heuristic
versus systematic analysis of persuasion and
research indicating that opinion change persists longer to the extent that it is bolstered
by topic-relevant cognitions (Chaiken, 1980;
Cook & Flay, 1978), we hypothesized that,
regardless of communicator likability, the
presumably source-mediated opinion change
manifested by subjects within videotaped and
audiotaped modalities would show less persistence than would the presumably contentmediated opinion change manifested by subjects within the written modality.
Experiment 1
Method
Subjects
Subjects were 109 psychology students from the University of Toronto who received extra credit toward their
course grades for participating. Data from three subjects
were discarded because they questioned the cover story
(n = 2) or failed to indicate their opinions (« = 1).
Procedure
At a mass testing session held early in the academic
year, all introductory psychology students completed a
large set of questionnaires. One of these questionnaires
solicited students' agreement with a pool of opinion
items that included the target statement, "The University
of Toronto should switch to a trimester system." A list
of students who disagreed with this statement and who
were under 25 years of age was compiled, and experimental subjects were recruited by phone from this list.
Participating in groups of three to nine persons, subjects learned during the laboratory session that the experiment concerned "people's reactions to speeches."
The experimenter stated that subjects would read (listen
to, view) and give their reactions to a "randomly selected" speech on one of a variety of campus issues.
After this introduction, the experimenter announced
the topic (see below) of the speech and distributed written
transcripts of a "background interview" with the speaker
that contained the communicator-likability manipulation (see below). Within each session, subjects were randomly assigned to likability conditions.
Next, the experimenter distributed written transcripts
or played an audiotape or videotape of the speech (persuasive message, see below). Afterwards, subjects completed a questionnaire that assessed their "reactions" to
the speech (dependent variables, see below). Finally, subjects were debriefed and excused.
Topic and Persuasive Message
The position advocated in the persuasive message was
"The University of Toronto should switch to the trimester system of instruction." The message was approximately 700 words long, contained five supportive arguments, and was designed to be relatively easy to understand (see Footnote 1).
Independent Variables
Communicator likability. In the "background interview," the speaker (communicator) was portrayed as a
male University of Toronto (U of T) administrator whose
work involved scholarship coordination and bursaries
and who had recently come to U of T from the University
of British Columbia (UBC) where he'had held a similar
position. Communicator likability was established by the
communicator's response to the interviewer's question,
"How do you like being at U of T compared to UBC?"
The communicator's response to this question appears
below. Phrases common to both likability conditions
appear without parentheses or brackets, and phrases that
appear only in the likable or the unlikable version are
enclosed in parentheses or brackets, respectively.
Well, it's interesting that you asked that question.
. . . To tell you the truth, I really (prefer being here
at U of T and living in Toronto to being out west)
[preferred being out west to being here at U of T and
living in Toronto]. For one thing, the people who I've
met both in my work and other contexts, including
colleagues, students, faculty, and other staff, strike me
as (being much friendlier and nicer than) [not being
really as friendly and nice as] the people I knew at
UBC. . . . Also I think that in terms of the overall
ability of the students and faculty here, that U of T
is a much (higher) [lower] quality institution than
UBC. I feel that U of T students particularly are (especially bright individuals) [really overrated]. . . . I
COMMUNICATION MODALITY AND PERSUASION
245
guess that in general, too, I find life here in Toronto
to be a lot (more) [less] pleasant. (There's so much
more to do here—movies get here sooner, and there
are more and better theater groups and the restaurants
and night life are really much better) [While there are
plenty of things to do here—movies, theaters, restaurants, night life—I really don't think that their overall
quality is very high in Toronto compared to other
places] For me, I guess U of T and Toronto has (been
a really) [not really been a very] good place to be.
point scales, subjects indicated how much effort they had
expended and how pleasant they had found the experience of reading (listening to, viewing) the speech. Subjects also indicated their age and sex and, just prior to
the debriefing, their written interpretations of the experiment. Answers to this open-ended question were
coded for suspicion.
Communication modality. A 35-year-old male colleague of the first author was trained to deliver the persuasive message. The speaker's final performance was
videotaped for use in the videotape conditions, and the
audio portion of the tape was played in the audiotape
conditions. These versions of the message lasted approximately 5 minutes, and subjects in written conditions
were given 5 minutes to read the written transcript of
the message.
The design included three levels of communication modality and two levels each of
communicator likability and subject sex. Because preliminary analyses revealed no significant effects that involved sex on subjects'
opinions, the reported analyses ignored this
variable.
Dependent Variables
Opinions. At the pretest, subjects indicated their
agreement with the statement, "The University of Toronto should switch to a trimester system," by 'marking
a 15-point scale anchored by "agree strongly" and "disagree strongly." In the laboratory, subjects wrote down
the communicator's topic and the overall position he
advocated ("The University of Toronto should switch to
the trimester system of instruction") and then indicated
their agreement with this position by marking the same
15-point scale described above.
Comprehension. Subjects were asked to summarize
each of the communicator's arguments. An argument
was scored as correct if, in the opinion of two independent raters (r = .79), it accurately summarized an argument contained in the message. As noted above, subjects were also asked to write down the message's topic
and overall position.
Cognitive responses. Subjects were given 3 minutes
to list their "thoughts and ideas" about "the speaker and
his speech." Statements were scored by two independent
raters as either message oriented (M) or communicator
oriented (C) and as either positively (+), negatively
(-), or neutrally (0) valenced. Examples of statements
placed in each of the above categories (along with interrater reliability coefficients) are C+ (r = .82), "Speaker
has good voice"; C- (r = .79), "He seems unfriendly";
CO (r = .68), "Seems like a nature lover"; M+ (r = .83),
"The economic advantages of the trimester seem reasonable"; M- (r = .82), "I'm not sure students would
benefit so much from the new system"; MO (r = .73),
"He said that many U.S. schools are on the trimester."
Source perception. Subjects rated the communicator
on 12 bipolar-adjective scales. Positive poles of these 15point scales were likable, knowledgeable, modest, intelligent, approachable, competent, warm, trustworthy,
pleasing, sincere, friendly, and unbiased.
Other measures. On 15-point scales, subjects rated
how distracted they felt from the content of the speech,
how difficult it was to understand the speech, the relative
amount of time they had spent thinking about the communicator's arguments (vs. his personal characteristics),
and the importance of the message topic. Also on 15-
Results
Check on Communicator
Likability Manipulation
A factor analysis of subjects' source ratings
yielded two rotated factors, labeled Attractiveness (likable, friendly, approachable,
pleasing, modest, warm, and unbiased) and
Expertise (knowledgeable, intelligent, and
competent). These factors accounted for
49.2% and 12.3% of the variance, respectively. Analyses of variance (ANOVAS) on subjects' factor scores indicated that the likable
(vs. the unlikable) communicator was perceived as more attractive, F(l, 100) = 41.72,
p < .001, and more expert, F(l, 100) = 5.55,
p < .025. Other source-perception findings
are reported below.
Opinions
An ANOVA revealed no significant effects
on subjects' premessage (i.e., pretest) opinions (all Fs < 1.00). Nevertheless, to control
for extremity of initial opinions and for the
slight differences between experimental conditions with respect to these scores, subjects'
opinion-change scores (postmessage minus
premessage opinions) were treated by an
analysis of covariance, using premessage
opinions as the covariate.3 Mean opinionchange scores (adjusted on the basis of premessage opinions) for the various experimental conditions are shown in Table 1.
As expected, the Modality X Likability in3
An ANOVA on change scores yielded virtually identical findings.
246
SHELLY CHAIKEN AND ALICE H. EAGLY
Table 1
Dependent Variables as a Function of Modality and Communicator Likability: Experiment 1
Communicator likability
Likable
Unlikable
Dependent variable
Written
Audiotape
Videotape
Written
Audiotape
Videotape
Opinion change
Communicator-oriented thoughts
3.66
1.05
4.82
1.72
4.87
2.12
3.43
1.00
1.47
1.41
.48
1.44
Note. Higher numbers indicate greater opinion change and more communicator-oriented thoughts. Cell ns range
from 16 to 19.
teraction proved significant on opinion
change, F(2, 99) = 3.74, p < .03. Planned
comparisons indicated that, compared to
subjects in written conditions, videotape and
audiotape subjects (combined) agreed somewhat more with the likable communicator's
message (p