Content uploaded by Richard E Petty
Author content
All content in this area was uploaded by Richard E Petty on Sep 13, 2020
Content may be subject to copyright.
THE ELABORATION LIKELIHOOD
MODEL
OF
PERSUASION
Richard
E.
Petty
DEPARTMENT
OF
PSYCHOLOGY
UNIVERSITY
OF
MISSOURI
COLUMBIA, MISSOURI
John
T.
Cacioppo
DEPARTMENT
OF
PSYCHOLOGY
UNIVERSITY
OF
IOWA
IOWA CITY, IOWA
I. Introduction..
............
...........................................
124
11. Postulate 1: Seeking Correctness..
..........................................
127
III. Postulate 2: Variations in Elaboration..
..........
A. The Elaboration Continuum..
...............
B.
A. ArgumentIMessage Quality
........
.................................
132
B. Peripheral Cues..
................
.................................
134
C. Affecting Elaboration
.............
.................................
136
V. Postulate
4
Objective Elaboration..
.............................
.......
137
A. Distraction
...............................................
.......
139
B. Repetition
..............................
..........
143
C. Personal Relevance/Involvement
............
..........
144
D. Personal Responsibility
...................
E. Need
for
Cognition.
..
..........
150
A. Personal Relevancehvolvement and the Operation
of
Cues
.....
Developmental Trends in Elaboration..
.......
IV. Postulate
3:
Arguments, Cues,
and
Elaboration
...
VI. Postulate
5:
Elaboration ve
.............................
152
B. Other Moderators
of
Cue Effectiveness
...................................
161
VII. Postulate 6: Biased Elaboration
...
................
162
A.
Prior
Knowledge..
.......................
..................................................
170
ents
...............................................
174
A. Persistence
of
Persuasion
...............................................
176
B. Attitude-Behavior Link
................................
C. Resistance
to
Counterpersuasion
.........................................
I8
1
VIII. Postulate 7: Consequences
of
Elaboration.
...
...
ADVANCES
IN
EXPERIMENTAL
SOCIAL PSYCHOLOGY. VOL.
19
I23
Copyright
0
1986
by
Academic
Press.
Inc.
All
rights
of
reproduction
in
any
form
reserved.
124
RICHARD
E.
PElTY
AND
JOHN
T.
CACIOPPO
IX.
Complicating Factors
.....................................................
182
A. Variables with Multiple Effects
on
Elaboration
.............................
183
B.
Variables That Affect Elaboration
and
Serve as Cues.
.......................
186
X.
Summary
and
Conclusions
.................................................
191
References..
............................................................
192
1.
Introduction
The study of attitudes and persuasion began as the central focus of social
psychology
(Allport,
1935; Ross, 1908). However, after a considerable flourish-
ing of research and theory from the 1930s through the 1960s, interest in the topic
began to wane. Two factors were largely responsible for this. First, the utility of
the attitude construct itself was called into question as researchers wondered if
attitudes were capable of predicting behavior. Because of this concern, some
even concluded that it might be time to abandon the attitude concept (Abelson,
1972; Wicker, 1971). Second,
so
much conflicting research and theory had
developed that it had become clear that “after several decades of research, there
(were) few simple and direct empirical generalizations that (could) be made
concerning how to change attitudes” (Himmelfarb
&
Eagly, 1974, p. 594).
Reviewers of the attitudes literature during the early 1970s lamented this
sorry
state of affairs (e.g., Fishbein
&
Ajzen, 1972). For example, Kiesler and Mun-
son (1975) concluded that “attitude change is not the thriving field it once was
and will be again” (p. 443).
By the late 1970s considerable progress had been made in addressing impor-
tant methodological and theoretical issues regarding the first substantive problem
plaguing the field-the consistency between attitudes and behaviors. Conditions
under which attitudes would and would not predict behavior were specified (e.g.,
Ajzen
&
Fishbein, 1977, 1980; Fazio
&
Zanna, 1981) and researchers began to
explore the processes underlying attitude-behavior correspondence (Sherman
&
Fazio, 1983; Fazio, 1985). The attitude change problem was slower to be ad-
dressed, however. In 1977, Muzifer Sherif asked “What is the yield in the way
of established principles in regard to attitude change?” He answered that there
was a “reigning confusion
in
the area” and a “scanty yield in spite of (a)
tremendously thriving output” (p. 370). In a review that generally heralded the
arrival of a new optimism in the attitudes field, Eagly and Himmelfarb (1978)
noted that “ambiguities and unknowns still abound” (p.
544;
see Cialdini, Petty,
&
Cacioppo, 1981; Cooper
&
Croyle, 1984, for more recent reviews).
As we noted above, the major problem facing persuasion researchers was
that after accumulating a vast quantity of data and an impressive number of
ELABORATION LIKELIHOOD
MODEL
OF
PERSUASION
125
theories-perhaps more data and theory than on any other single topic in the
social sciences (see McGuire, 1985)-there was surprisingly little agreement
concerning if, when, and how the traditional source, message, recipient, and
channel variables (cf. Hovland, Janis,
&
Kelley, 1953; McGuire, 1969; Smith,
Lasswell,
&
Casey, 1946) affected attitude change. Existing literature supported
the view that nearly every independent variable studied increased persuasion in
some situations, had no effect in others, and decreased persuasion in still other
contexts. This diversity of results was apparent even for variables that on the
surface, at least, would appear to be quite simple. For example, although it might
seem reasonable to propose that by associating a message with an expert source
agreement could be increased (e.g., see Aristotle’s
Rheroric),
the accumulated
contemporary research literature suggested that expertise effects were consider-
ably more complicated than this (Eagly
&
Himmelfarb, 1974; Hass, 1981).
Sometimes expert sources had the expected effects (e.g., Kelman
&
Hovland,
1953), sometimes no effects were obtained (e.g., mine
&
Severance, 1970),
and sometimes reverse effects were noted (e.g., Sternthal, Dholakia,
&
Leavitt,
1978). Unfortunately, the conditions under which each of these effects could
be
obtained and the processes involved in producing these effects were not at all
apparent.
Our primary goal in this article is to outline a general theory of attitude
change, called the Elaboration Likelihood Model (ELM; Petty
&
Cacioppo,
1981a), which we believe provides a fairly general framework for organizing,
categorizing, and understanding the basic processes underlying the effectiveness
of persuasive communications. Importantly, the ELM attempts to integrate the
many seemingly conflicting research findings and theoretical orientations under
one conceptual umbrella. The ELM began with our attempts to account for the
differential persistence of communication-induced attitude change. After review-
ing the literature
on
attitude persistence, we concluded that the many different
empirical findings and theories in the field might profitably
be
viewed as empha-
sizing one of just two relatively distinct routes to persuasion (Petty, 1977; Petty
&
Cacioppo, 1978). The
first
type of persuasion was that which likely resulted
from a person’s careful and thoughtful consideration of the true merits of the
information presented in support of an advocacy (central route). The other type
of persuasion, however, was that which more likely occurred as a result of some
simple cue in the persuasion context (e.g., an attractive source) that induced
change without necessitating scrutiny of the true merits of the information pre-
sented (peripheral route). In the accumulated literature, the first kind of persua-
sion appeared to be more enduring than the latter (see Fig.
1;
see Cook
&
Flay,
1978, and Petty, 1977, for reviews).
Following our initial speculation about the two routes to persuasion and the
implications for attitudinal persistence (Petty, 1977; Petty
&
Cacioppo, 1978),
we have developed, researched, and refined a more general theory of persuasion,
126
'
COGNITIVE STRUCTURE
--___
CHANGE:
Are new cognitions adopted and
stored in memory?. are
different responses made
salient than previously?
No
RICHARD
E.
PElTY
AND
JOHN
T.
CACIOPFU
PERSUASIVE COMMUNICATION
-
-
-
-
-
-
-
-
-
-
- -
-
-
-
-
-
-
PERIPHERAL ATTITUDE
1
SHIFT
I
I
Altitude is relatively temporary,
I
susceptible. and unpredictive
I
of
behavior
I
L
--------__________
J
MOTIVATED TO PROCESS?
t
personal relevance: need
Yes
for
cognition: personal
responsibility; etc.
PERIPHERAL CUE
PRESENT?
positiveinegative
affect: attractive!
expelt
sources:
distraction; repetition: number
of
arguments,
ComDrehensibilitv; etc.
prior knowledge: message etc.
J
Fig.
1.
Central and peripheral routes to persuasion.
This
figure
depicts the two anchoring
endpoints on the elaboration likelihood continuum (adapted from Petty,
1977;
Petty
&
Cacioppo,
1978,
1981a).
the
ELM,
which
is
based on these two routes (Petty
&
Cacioppo, 1981a).
In
addition, we have addressed the various applications
of
this model
to
fields such
as psychotherapy and counseling (Cacioppo, Petty,
&
Stoltenberg, 1985; Petty,
Cacioppo,
&
Heesacker,
1984)
and mass media advertising and selling (Caciop-
ELABORATION
LIKELIHOOD MODEL
OF
PERSUASION
127
po
&
Petty, 1985; Petty
&
Cacioppo, 1983a, 1984b; Petty, Cacioppo,
&
Schumann, 1984). In the remainder of this article we will outline the
ELM
as a
series of postulates that make explicit the guiding assumptions and principles of
the model. We will also provide a methodology for testing the major processes
outlined by the
ELM
and we will review research which provides evidence
relevant to the framework.
Before outlining our model of attitude change, however, it is important to
define our use of the term
attitude.
Consistent with the positions of other attitude
theorists (e.g., Thurstone, 1928), we regard attitudes as general evaluations
people hold in regard to themselves, other people, objects, and issues. These
general evaluations can
be
based on a variety of behavioral, affective, and
cognitive experiences, and
are
capable of influencing or guiding behavioral,
affective, and cognitive processes.
Thus,
a person may come to like a new
political candidate because she just donated
$100
to the campaign (behavior-
initiated change), because the theme music in a recently heard commercial in-
duced a general pleasantness (affect-initiated change), or because the person was
impressed with the candidate’s issue positions (cognitive initiated change). Simi-
larly, if a person already likes a political candidate he may agree to donate money
to the campaign (behavioral influence), may feel happiness upon meeting the
candidate (affective influence), and may selectively encode the candidate’s issue
positions (cognitive influence).
11.
Postulate
1:
Seeking Correctness
Our first postulate and an important guiding principle in the
ELM
agrees
with Festinger’s (1950) statement that:
People
are
motivated to
hold
correct attitudes.
Incorrect attitudes are generally maladaptive and can have deleterious behav-
ioral, affective, and cognitive consequences. If a person believes that certain
objects, people, or issues are “good” when they
are
in fact “bad,” a number of
incorrect behavioral decisions and subsequent disappointments may follow.
As
Festinger (1954) noted, the implication of such a drive is that “we would expect
to observe behavior on the part of persons which enables them to ascertain
whether or not their opinions are correct” (p. 118). In his influential theory of
social comparison processes, Festinger
(1
954) focused on how people evaluated
the correctness of their opinions by comparing them to the opinions of others. In
Section IX,B we address how the
ELM
accounts for attitude changes induced by
exposure to the opinions of varying numbers
of
other people. But first we need to
outline our other postulates.
128
RICHARD
E.
PETTY
AND
JOHN
T.
CACIOPPO
111.
Postulate
2:
Variations in Elaboration
Postulate
2
states that:
Although people want to hold correct attitudes, the amount and nature of issue-relevant
elaboration in which people
are
willing
or
able
to
engage
to
evaluate1 a message vary with
individual and situational factors.
By
elaboration
in a persuasion context, we mean the extent to which a person
thinks about the issue-relevant arguments contained in a message. When condi-
tions foster people’s motivation and ability to engage in issue-relevant thinking,
the “elaboration likelihood” is said to be high. This means that people are likely
to attend to the appeal; attempt to access relevant associations, images, and
experiences from memory; scrutinize and elaborate upon
the
externally provided
message arguments in light of the associations available from memory; draw
inferences about the merits of the arguments
for
a recommendation based upon
their analyses; and consequently derive an overall evaluation of,
or
attitude
toward, the recommendation. This conceptualization suggests that when the
elaboration likelihood is high, there should be evidence for the allocation of
considerable cognitive resources to the advocacy. Issue-relevant elaboration will
typically result in the new arguments,
or
one’s personal translations of them, being
integrated into the underlying belief structure (schema) for the attitude object
(Cacioppo
&
Petty, 1984a).
As
we will note shortly, sometimes this issue-relevant
elaboration proceeds in a relatively objective manner and is governed mostly by
the strength of the issue-relevant arguments presented, but at other times this
elaboration is more biased and may be guided mostly by the person’s initial
attitude.
Of course, people are not motivated nor are they able to scrutinize carefully
every message that they receive (cf. McGuire’s, 1969, “lazy organism”), and it
would not be adaptive for them to do
so.
As
Miller, Maruyama, Beaber, and
Valone (1976) noted, “It may be irrational to scrutinize the plethora of coun-
terattitudinal messages received daily.
To
the extent that one possesses only a
limited amount
of
information processing time and capacity, such scrutiny would
disengage the thought processes from the exigencies of daily life” (p. 623).
Current research in cognitive and social psychology provides strong support
for
the view that at times people engage in “controlled,” “deep,” “systematic,”
and/or “effortful” analyses of stimuli, and at other times the analyses are better
characterized as “automatic,
”
“shallow,
”
“heuristic,
”
and/or “mindless”
(for further discussion, see Craik, 1979; Eagly
&
Chaiken, 1984; Kahneman,
Slovic,
&
Tversky, 1982; Langer, 1978; and Schneider
&
Shiffrin, 1977).’
‘See
Petty and Cacioppo
(1986)
for discussion of the relationship between these distinctions and
the central/pe.ripheral distinction of the
ELM.
ELABORATION LIKELIHOOD MODEL
OF
PERSUASION
I29
A. THE ELABORATION CONTINUUM
One can view the extent of elaboration received by a message as a con-
tinuum going from no thought about the issue-relevant information presented to
complete elaboration of every argument, and complete integration of these elab-
orations into the person’s attitude schema. The likelihood of elaboration will
be
determined by a person’s motivation and ability to evaluate the communication
presented
(see
Fig.
1).
In an earlier review
of
the attitude change literature (Petty
&
Cacioppo, 1981a), we suggested that the many theories
of
attitude change
could be roughly placed along this elaboration continuum. At the high end of this
continuum are theoretical orientations such as inoculation theory (McGuire,
1964),
cognitive response theory (Greenwald,
1968;
Petty, Ostrom,
&
Brock,
1981),
information integration theory (Anderson,
1981),
and the theory of rea-
soned action (Ajzen
&
Fishbein,
1980;
Fishbein,
1980),
which all assume that
people typically attempt to carefully evaluate (though not always successfully)
the information presented in a message, and integrate this information into a
coherent position. Researchers within this tradition have emphasized the need to
examine what kinds of arguments are persuasive and how variables affect the
comprehension, elaboration, learning, integration, and retention of issue-rele-
vant information (McGuire,
1985).
Other persuasion theories do not place much credence on the arguments
in
a
message or issue-relevant thinking. Instead, they focus on how simple affective
processes influence attitudes or on how people can employ various rules or
inferences to judge their own attitudes or the acceptability of an attitudinal
position. Although in most laboratory studies of attitude change subjects
will
have some motivation and/or ability to form at least a reasonable opinion either
by scrutinizing arguments or making an inference about the acceptibility
of
the
recommendation based on cues in the context, there are circumstances in which
neither arguments nor acceptance cues are present. For example, when subjects
are exposed to nonsense syllables (Staats
&
Staats,
1957)
or
polygons (Kunst-
Wilson
&
Zajonc,
1980),
no elaboration of arguments is possible because no
arguments are presented, and validity cues may be irrelevant because there is no
explicit “advocacy” to judge. Theories such as classical conditioning (Staats
&
Staats,
1958)
and mere exposure (Zajonc,
1968,
1980),
which describe evalua-
tions of objects changing as a result of rather primitive affective and associational
processes, are especially relevant under these circumstances. Although these
theories have been tested and applied primarily in situations where no explicit
“advocacy” is presented, they also should be applicable to situations in which
an issue position is advocated, but people have virtually no ability and/or moti-
vation to consider it. In these situations, attitudes may still
be
changed if the
attitude object is associated with a relatively strong positive or negative affective
cue, or a weaker cue is continually paired with the attitude object.
130
RICHARD
E.
PEITY AND
JOHN
T.
CACIOPPO
If no strong affective cues are presented, it is still possible for people to
form a “reasonable” attitude without relying on scrutiny of the issue-relevant
arguments presented by relying on various persuasion rules or inferences that
may be either rather simple or relatively complex. For example, according to
self-perception theory (Bem,
1972),
people may come to like or dislike an object
as a result of a simple inference based on their own behavior (e.g., if
I
bought it,
I
must like it). According to the heuristic model of persuasion (Chaiken,
1980;
Eagly
&
Chaiken,
1984),
people may evaluate messages by employing various
rules that they have learned on the basis of past experience (e.g., people agree
with people they like). Social judgment theory (Sherif
&
Sherif,
1967)
proposes
that people evaluate messages mostly on the basis of their perceived position-
messages are contrasted and rejected if they appear too discrepant (fall
in
the
latitude of rejection), but are assimilated and accepted if they appear closer to
one’s initial position (fall in the latitude of acceptance; Pallak, Mueller, Dollar,
&
Pallak,
1972).
In addition to the relatively simple acceptance/rejection rules proposed by
the preceding models, attitude change may be affected by more complex reason-
ing processes, such as those based on balance theory (Heider,
1946;
Insko,
1984)
or certain attributional principles (e.g., Kelley,
1967;
Eagly, Wood,
&
Chaiken,
1978).
Importantly, even reliance on more complex inferences obviates the need
for careful scrutiny of the issue-relevant arguments in a message. In other words,
each of these processes (e.g., self-perception, assimilation, balance) is postu-
lated to be sufficient to account for attitude change without requiring a personal
evaluation of the issue-relevant arguments.2 In sum, we have proposed that when
either motivation or ability to process issue-relevant arguments is low, attitudes
may be changed by associating an issue position with various affective cues, or
people may attempt to form a reasonable opinion position by making an in-
ference about the likely correctness or desirability of a particular attitude position
based on cues such as message discrepancy, one’s own behavior, and the charac-
teristics of the message source.
B. DEVELOPMENTAL TRENDS IN ELABORATION
Interestingly, the attitude change processes that we have just described form
an
elaboration continuum which likely coincides with the manner in which
attitude change processes develop through adulthood. Specifically, the very
young child probably has relatively little motivation to think about the true merits
of people, objects, and issues, and even less ability to do
so.
Thus, attitudes may
2Insko
(1981)
extended balance theory
to
include a person’s consideration of issue-relevant
arguments.
This
more general balance formulation therefore broadens the theory beyond peripheral
processing.
ELABORATION LIKELIHOOD MODEL
OF
PERSUASION
131
be
affected primarily by what feels good or bad.
As
children mature, they
become more motivated to express correct opinions on certain issues, but their
ability to scrutinize issue-relevant arguments may still
be
poor due to lack of
knowledge. Therefore, they may be particularly reliant on certain cognitive rules
based on personal experience such as, “My mother knows what’s right,” or “If
I
play with it,
I
must like it.” Consistent with this reasoning, children have been
shown to be more susceptible
to
appeals based on behavioral cues and self-
perceptions than issue-relevant argumentation (e.g., Miller, Brickman,
&
Bolen,
1975).
Finally, as people move into adulthood, interests become more focused and
the consequences of holding correct opinions on certain issues increase. In addi-
tion, as people’s acquired knowledge and cognitive skills grow, this renders
them more able to critically analyze issue-relevant information on certain topics
and makes them less reliant than children on certain primitive heuristics (cf.
Ross,
1981).
As
we noted earlier, of course, although people may have the
requisite ability and motivation to scrutinize certain attitude issues, they will lack
motivation and ability on others. Thus, simple inferences and affective cues may
still produce attitude change in adults.
In sum, one’s initial evaluations are likely to be largely hedonistic since,
lacking the motivation and/or ability to consider issue-relevant arguments, at-
titudes will be based primarily on positive and negative affective cues associated
with the attitude object. As development proceeds, some attitudes may be
formed on the basis of simple inferences, decision rules, and social attachments.
Finally, the formation and change of some attitudes become very thoughtful
processes in which issue-relevant information is carefully scrutinized and evalu-
ated in terms of existing knowledge. Importantly, our sequence
of
the develop-
mental stages of influence
is
consistent with other developmental models of
judgment. For example, in discussing the development of moral standards,
Kohlberg
(1963)
identifies three developmental levels. At the first level (precon-
ventional), moral evaluations are based primarily
on
the affective consequences
of an act. At level
2
(conventional), evaluations of acts are based primarily on
socially accepted rules and laws. Finally, at level 3 (postconventional), an eval-
uation of an act is based on a person’s idiosyncratic but well-articulated moral
code. The parallels to our stages of influence are obvious.
Although we have argued that there is a continuum of message elaboration
ranging from none to complete, and that different attitude change processes may
operate along the continuum, it is also important to note that these different
theoretical processes can
be
viewed as specifying just two qualitatively distinct
routes to persuasion. The first route, which we have called the “central route,”
occurs when motivation and ability to scrutinize issue-relevant arguments are
relatively high. The second, or “peripheral route,” occurs when motivation
and/or ability are relatively low and attitudes are determined by positive or
negative cues in the persuasion context which either become directly associated
132
RICHARD E. PETTY AND
JOHN
T.
CACIOPPO
with the message position
or
permit a simple inference as to the validity of the
message. In short, even though one can view message elaboration as
a
con-
tinuum, we can distinguish persuasion that is primarily a result of issue-relevant
argumentation from persuasion that is primarily a result of some cue in the
persuasion context that permits attitude change without argument scrutiny. In
fact, we will find
it
useful elsewhere in this article to talk about the elaboration
likelihood continuum by referring to the prototypical processes operative at each
extreme.
IV.
Postulate
3:
Arguments,
Cues,
and Elaboration
Much of
our
discussion
so
far is summarized in the next postulate.
Variables can affect the amount and direction
of
attitude change
by:
(A)
serving as
persuasive arguments,
(B)
serving as peripheral cues, and/or
(C)
affecting the extent
or
direction
of
issue and argument elaboration.
In subsequent sections we discuss how many of the typical source, message,
recipient, channel, and context variables manipulated in the accumulated persua-
sion research can be understood in terms of the three-part categorization above,
but first we need to define and operationalize the constructs.
A. ARGUMENTlMESSAGE QUALITY
One of the least researched and understood questions in the psychology of
persuasion is “What makes an argument persuasive?”
As
we noted earlier,
literally thousands of studies and scores of theories have addressed the question
of how some extramessage factor (e.g., source credibility, repetition) affects the
acceptance
of
a particular argument, but little is known about what makes a
particular argument
(or
message) persuasive in isolation. In fact, the typical
persuasion experiment employs only one message and examines how some extra-
message factor affects acceptance of the message conclusion. Furthermore, stud-
ies that do include more than one message often do
so
for purposes of gener-
alizability across topics, not because the messages are proposed to differ
in
some
theoretically meaningful way (e.g., Hovland
&
Weiss,
1951).
There are, of
course, notable exceptions to
our
generalization.
For
example, a few studies
have manipulated the comprehensibility
or
complexity of a message (e.g., Ea-
gly,
1974;
Eagly
&
Warren,
1976;
Regan
&
Cheng,
1973),
mostly to test
McGuire’s
(1968)
information processing model, but even these studies were not
aimed at uncovering the underlying characteristics of persuasive arguments.
Perhaps the most relevant research to date is that in which subjects are asked to
rate arguments along various dimensions (e.g., validity, novelty) in order to
determine what qualities make an argument persuasive (see Vinokur
&
Burn-
ELABORATION
LIKELIHOOD MODEL
OF
PERSUASION
133
stein, 1974). but this kind of research is rare and in its infancy. After over 40
years of work on persuasion in experimental social psychology, Fishbein and
Ajzen (1981) could accurately state that “the general neglect
of
the information
contained in a message.. .is probably the most serious problem in communication
and persuasion research” (p. 359).3
In the
ELM,
arguments are viewed as bits
of
information contained
in
a
communication that are relevant to a person’s subjective determination of the
true merits of an advocated position. Because people hold attitudes for many
different reasons (Katz, 1960), people will invariably differ in the kinds of
information they feel
are
central to the merits of any position (Snyder
&
DeBono,
1985). Nevertheless, for purposes of testing the
ELM,
it
is
necessary to specify
arguments that the vast majority of a specifiable population finds compelling
rather than specious. In our research on the
ELM,
we have postponed the ques-
tion of what specific qualities make arguments persuasive by defining argument
quality in an empirical manner. In developing arguments for a topic, we begin by
generating a large number of arguments, both intuitively compelling and spe-
cious ones, in favor
of
some issue (e.g., raising tuition). Then, members
of
the
appropriate subject population are given these arguments to rate for per-
suasiveness. Based on these scores we select arguments with high and
low
ratings to comprise at least one “strong” and one “weak” message. Subse-
quently, other subjects are given one of these messages and are told to think
about and evaluate it carefully. Following examination of the message, subjects
complete a “thought-listing measure” (Brock, 1967; Greenwald, 1968),
in
which they are instructed to record the thoughts elicited by the message. These
thoughts
are
then coded as to whether they are favorable, unfavorable, or neutral
toward the position advocated (see Cacioppo
&
Petty, 1981~; Cacioppo,
Har-
kins,
&
Petty, 1981, for further discussion of the thought-listing procedure). We
define a “strong message” as one containing arguments (e.g., we should raise
tuition
so
that more books can be purchased for the library) such that when
subjects are
instructed
to think about the message, the thoughts that they generate
are predominantly favorable. Importantly, for positive attitude change to occur,
the thoughts should
be
more favorable than those available prior to message
exposure. On the other hand, we define a
“weak
message” as one containing
arguments (e.g., we should raise tuition
so
that more trees and shrubs can be
planted on campus) such that when subjects are instructed to think about them,
the thoughts that they generate are predominantly unfavorable. For negative
change (boomerang) to occur, the thoughts should be more unfavorable than
those available prior to message exposure.
3Notably, Fishbein and Ajzen
(1975)
and other expectancy value theorists (e.g.. Rosenberg,
1956)
have examined argument or attribute persuasiveness from a phenomenological perspective.
However, the question
of
why
a particular argument or attribute is seen as more positive or negative
than others
is
still not addressed.
134
RICHARD E. PE’lTY
AND
JOHN
T.
CACIOPPO
Once the messages meet the criterion of eliciting the appropriate profile
of
thoughts, they are checked for other characteristics. First, a panel of subjects
rates the messages for overall believability. Our goal is to develop arguments that
are strong and weak, but that do not strain credulity. (This is not to say that our
arguments are necessarily veridical-just reasonably plausible to our subjects.)
Next, people from the relevant subject pool rate the messages for comprehen-
sibility, complexity, and familiarity. Again, our goal is to develop strong and
weak messages that are roughly equivalent in their novelty and in our subjects’
ability to understand them. The top panel of Fig.
2
depicts the results of a
hypothetical study in which some extramessage “treatment” has no effect
on
persuasion. In this study, only the quality
of
the message arguments determined
the extent of attitude change. We will compare this simple result with the other
possibilities depicted in Fig.
2
in the remainder of this article.
B.
PERIPHERAL
CUES
According to the Elaboration Likelihood Model, one way to influence at-
titudes is by varying the quality of the arguments in a persuasive message.
Another possibility, however, is that a simple cue in the persuasion context
affects attitudes in the absence of argument processing. As we noted earlier,
some cues will
do
this because they trigger relatively primative affective states
that become associated with the attitude object.
Various
reinforcing (e.g., food;
Janis, Kaye,
&
Kirschner,
1965)
and punishing (e.g., electric shock; Zanna,
Kiesler,
&
Pilkonis,
1970)
stimuli have proved effective in this regard. Other
cues work, however, because they invoke guiding rules (e.g., balance; Heider,
1946)
or inferences (e.g., self-perception; Bem,
1972).
Since cues are postulated to affect attitude change without affecting argu-
ment processing, it is possible to test manipulations as potential cues by present-
ing them to subjects with the advocated position only (i.e., without accompany-
ing persuasive arguments), as in prestige suggestion (see Asch,
1948).
If the
manipulation is a potential cue, it should have the ability to affect attitudes in the
absence of any arguments. Alternatively, one could present an incomprehensible
message (e.g., in a foreign language)
on
some topic along with the potential cue
(e.g., speed of speech; Miller
er
af.,
1976).
Subjects could be asked to rate, for
example, how likely it is that the speaker is convincing. Again, if the cue is
operative, it should be capable of affecting judgments even if there are
no
arguments to process. Finally, a simple procedure might involve merely describ-
ing various potential cues to subjects (e.g., a message with 1 vs.
10
arguments; a
message from an attractive vs.
an
unattractive source) and asking them which
would more likely be acceptable and/or persuasive. These procedures would not,
of
course, indicate
why
a cue was effective (e.g., were the judgments due to
affective association
or
the invocation of a simple decision rule?), nor would
they
ELABORATION LKELMOOD MODEL
OF
PERSUASION
135
eliminate the possibility that more thoughtful processes were involved (e.g.,
subjects might attempt
to
generate arguments consistent with the position; cf.,
Burnstein, Vinokur,
&
Trope,
1973).
However, these procedures
would
indicate
whether
or
not a manipulation has the
potential
to
serve as a peripheral cue.
I.
No
Eilect
Message
Main
Effect
Treatment
Main
Effect
(Cue)
Message
Treatment
interaction
X
Treatment
Main Effect
Message
x
Treatment
lnteraction
(Cue)
+
II.
Cue Effect
-
...
-
-tW
Basdine Treatment Baseline Treatment
111.
Objective Processing
Baseline Treatment
IV.
Biased Processing
-
Baseline Trealmenl
I
(b)
negalive
Baseline Treatmenl
Fig.
2.
Impact
of
variables on attitude change according
to
the
ELM.
Under conditions
of
high
elaboration likelihood, attitudes
are
affected mostly by argument quality
(I).
Under conditions of
low
elaboration likelihood, attitudes
are
affected mostly by peripheral cues
(11).
Under conditions
of
moderate elaboration likelihood, varibles may enhance
or
reduce message processing
in
either a
relatively objective
@I)
or
relatively biased
(IV)
manner (adapted
from
Petty
&
Cacioppo, 1984~).
136
RICHARD
E.
PETTY
AND
JOHN
T.
CACIOPPO
Panel
I1
in Fig.
2
presents the results of a hypothetical study in which
strong, weak, and mixed argument messages were presented along with a treat-
ment that served as a peripheral cue. Note that in the pure case of cue processing,
the cue affects all three kinds of messages equally. Since cues are most likely to
operate when subjects are either unmotivated
or
unable to process issue-relevant
arguments (as depicted in Fig.
l),
the data show a strong effect for the cue
treatment, but little effect for argument quality.
In
the
left
half of Panel
2
the cue
is positive, and in the right half the cue is negative.
C.
AFFECTING
ELABORATION
We have now defined two of the key constructs in the Elaboration Like-
lihood Model: argument quality and peripheral cues. The third way in which a
variable can affect persuasion is by determining the extent or direction of mes-
sage processing. Variables can affect argument processing in a relatively objec-
tive or a relatively biased manner (Petty
&
Cacioppo, 1981a). In relatively
objective
processing, some treatment variable either motivates
or
enables sub-
jects to see the strengths of cogent arguments and the flaws in specious ones, or
inhibits them from doing
so.
In relatively
biased
processing some treatment
variable either motivates
or
enables subjects to generate a particular kind
of
thought in response to a message,
or
inhibits a particular kind
of
thought. Rela-
tively objective elaboration has much in common with “bottom-up” processing
since the elaboration is relatively impartial and data driven. Relatively biased
elaboration has more in common with “top-down’’ processing since the elabora-
tion, for example, may be governed by a relevant attitude schema which guides
processing in a manner leading to the maintenance or strengthening of the sche-
ma (cf. Bobrow
&
Norman, 1975; Landman
&
Manis, 1983). Postulate
4
deals
further with the nature of relatively objective processing, and Postulate
6
deals
further with the nature of relatively biased processing.
Of course,
in
order to test the ELM, it is important to assess how much
message processing subjects are engaged in (i.e., how much cognitive activity or
effort is devoted to issue-relevant thinking), and what variables affect elabora-
tion. We have used four different procedures to assess the extent of thinking. The
first procedure is the simplest and involves directly asking people how much
effort they expended in processing the message,
or
how much thinking they were
doing about the advocacy. Although we have found this method to prove sen-
sitive in some studies (e.g., Cacioppo, Petty,
&
Morris, 1983; Petty, Harkins,
&
Williams, 1980),
in
others it has not produced differences even though there
were other indications of differential processing (e.g., Harkins
&
Petty, 1981a,
1982).
The problem, of course, is that although people may sometimes be aware
of how much cognitive effort they are expending, people do not always have
ELABORATION LIKELIHOOD
MODEL
OF
PERSUASION
137
access to their cognitive processes (Nisbett
&
Wilson, 1977).
A second procedure involves using the thought-listing technique developed
by Brock (1967) and Greenwald (1968). In this procedure, subjects list their
thoughts either in anticipation of, during, or after message exposure, and the
thoughts are subsequently categorized into theoretically meaningful units (e.g
.
,
counterarguments; source-related thoughts) by the subjects or independent jud-
ges. The thought-listing technique has proved to be an important supplemental
tool
in
tracking the amount and type of cognitive activity involved in persuasion
and resistance (see Cacioppo
el
al.,
1981; Cacioppo
&
Petty, 1981c; for reviews
of thought-listing methodology and results). Although statistical procedures have
been used to show that cognitive activity mediates attitude effects in some
instances (e.g., Cacioppo
&
Petty, 1979b; Insko, Turnbull,
&
Yandell, 1974;
Petty
&
Cacioppo, 1977), thought listings do not provide definitive evidence for
cognitive mediation because the evidence is basically correlational (cf. Miller
&
Colman, 1981).
A third procedure that we have used to assess the extent and affectivity of
information processing activity involves the use of psychophysiological mea-
sures. For example, we have shown that facial electromyographic (EMG) ac-
tivity is capable of distinguishing positive from negative reactions to stimuli
(e.g., Cacioppo
&
Petty, 1979a) and that perioral (e.g., lip) EMG activity is
capable of distinguishing cognitively effortful from less taxing mental work
(e.g., Cacioppo
&
Petty, 1981b). The physiological procedures have several
potential advantages over self-reports of cognitive activity and thought listings.
For example, these measures can track psychological processes over time, and
may be less susceptible to artifacts (e.g., demand characteristics) and subjects’
inability to recall the process or content of their thoughts. Although work
on
psychophysiological assessments of attitudina processes is in its early stages,
these measures hold considerable promise for tracking and marking the underly-
ing mediation of persuasion and resistance (see Cacioppo
&
Petty, 1981a, 1986;
Petty
&
Cacioppo, 1983; for reviews).
The fourth procedure for assessing the extent
of
cognitive processing, and
the one highlighted in this article, is based on our manipulation of message
argument quality. This procedure is discussed in the next section.
V.
Postulate
4:
Objective Elaboration
Postulate
3
noted that variables could serve as arguments, cues, or affect
processing. We further noted that processing could proceed in a relatively objec-
tive or biased manner. Postulate
4
deals with objective processing. Specifically:
138
RICHARD
E.
PEITY
AND
JOHN
T.
CACIOPPO
Affecting motivation and/or ability to process a message in a relatively objective
manner
can
do
so
by
either enhancing
or
reducing argument mutiny.
As
we hinted above,
our
empirical method of defining argument quality allows
us to assess the extent to which a variable affects argument processing and the
extent to which this processing is relatively objective or biased. We shall consid-
er first the expected consequences of variables affecting relatively objective
processing.
Assume for the moment that we have created a control condition in which
motivation or ability to process issue-relevant arguments is rather low. Subjects
should show relatively little differentiation of strong from weak arguments in this
condition. However, if a manipulation enhances argument processing in a rela-
tively objective manner, then subjects should show greater differentiation of
strong from weak arguments. More specifically, a message with strong argu-
ments should produce more agreement when it is scrutinized carefully than when
scrutiny is low, but a message with weak arguments should produce less overall
agreement when scrutiny is high rather than low. This pattern of results is
depicted in the left half of Panel
III
in Fig.
2.
In
a similar fashion, we can assess
the extent to which a variable disrupts processing in a relatively objective man-
ner. Consider a situation in which subjects
are
processing the message arguments
quite diligently. These subjects should show considerable differentiation of
strong from weak arguments. However,
if
argument processing is disrupted, due
either to reduced motivation or ability, argument quality should be a less impor-
tant determinant of persuasion. More specifically, a strong message should in-
duce less agreement when processing is disrupted than when it is not, but a weak
message should produce more agreement when processing is disrupted than
when it is not. The right half of Panel
III
in Fig.
2
depicts this pattern.
In
addition
to subjects’ attitudes being more differentiated to weak and strong messages
when argument processing is high rather than low, the profile of subjects’
thoughts also should show greater differentiation of arguments when processing
is high rather than low.
In sum, by manipulating argument quality along with some other variable, it
is possible to tell whether that variable enhances or reduces argument processing
in a relatively objective manner. If the variable enhances argument processing,
subjects’ thoughts and attitudes should be more polarized when the variable is
present rather than absent, but
if
the variable reduces argument processing,
subjects’ thoughts and attitudes should
be
less polarized when the variable is
present rather than absent. Before moving
on
to our postulates concerning
pe-
ripheral cues and biased processing, we review some evidence that variables can
affect persuasion by affecting the extent
of
argument processing in a relatively
objective manner.
ELABORATION LIKELIHOOD MODEL
OF
PERSUASION
139
A.
DISTRACTION
Research on distraction’s effect on persuasion can be traced to
an
intriguing
study by Allyn and Festinger (1961), in which high school students were present-
ed with a speech which argued that teenage drivers are dangerous. The students
were either forewarned of the opinion topic and told that their opinions would be
assessed (opinion orientation)
or
were told simply that they were to assess the
personality of the speaker (personality orientation). Although these two condi-
tions did not differ in the average opinion change they induced, when analyses
were conducted on the most involved subjects (those with extreme opinions
or
those who said the issue was important), a significant difference was found such
that there was more persuasion in the personality than in the opinion orientation
condition. Two possible explanations for this effect were offered. The initial
explanation favored by Allyn and Festinger was that the “forewarning” in the
opinion orientation condition stimulated the involved students to counterargue
and/or derrogate the source (see
also
Freedman
&
Sears, 1965). A second
explanation, proposed initially by Festinger and Maccoby (1964), was that the
involved subjects in the personality orientation condition were distracted from
the counterarguing and/or source derrogating in which they normally would have
engaged.
In the years since the Allyn and Festinger experiment, a considerable
number of studies have accumulated on both “forewarning” and “distraction,”
and it is now clear that both effects are viable. In this section we apply the ELM
framework to “distraction” and discuss how this variable works by affecting
information processing in a relatively
objective
manner. In section VII,B we
apply the
ELM
to “forewarning” and address how this variable also works by
affecting information processing, but in a relatively
biased
manner.
In 1973, Baron, Baron,
&
Miller reviewed the accumulated research on
“distraction” and concluded that although many individual studies were suscep-
tible to a wide variety of mediational interpretations, there were just two the-
oretical explanations that could account for the existing data parsimoniously.
One explanation was the disruption of counterarguing interpretation favored by
Festinger and Maccoby. Another interpretation offered by Baron
et
al.,
however,
was based, ironically, on Festinger’s (1957) theory of cognitive dissonance.
Baron
et
af.
argued that distraction manipulations require subjects to exert more
effort than usual in order to understand the message. Furthermore, “since choos-
ing to hear a counterattitudinal message can be viewed as attitude-discrepant
behavior, the effort required to comprehend a counterattitudinal message will
directly determine the amount of dissonance created by the choice” (p. 317).
One way for subjects to reduce this dissonance, of course, is for them to justify
their effort by overvaluing the communication.
140
RICHARD
E.
PETTY
AND
JOHN
T.
CACIOPPO
At the time of the review by Baron
et
al.,
the available experiments did not
allow a distinction between the two alternative theories because evidence that
appeared to support either the counterargument
or
the dissonance position also
could be seen as consistent with the other account. Importantly, even research
using the thought-listing technique, which showed that with increasing distrac-
tion the number of counterarguments listed decreased (Keating
&
Brock, 1974;
Osterhouse
&
Brock, 1970), was open to multiple interpretations. Was a reduc-
tion in negative thoughts obtained with distraction because distraction disrupted
counterarguing,
or
was it because distraction induced attitude change via disso-
nance (or some other process) which was subsequently justified in the thought
listings (Miller
&
Baron, 1973)?
Our initial use of the manipulation of strong and weak arguments (see
section IV,A) came in an experiment that attempted to distinguish the dissonance
from the counterargument disruption interpretations of distraction (Petty, Wells,
&
Brock, 1976, Experiment 1). A second aim of our experiment was to test a
more general distraction formulation than “counterargument disruption.
”
Spe-
cifically, we reasoned that
if
the predominant thoughts
to
a message without
distraction were unfavorable, then distraction should disrupt these unfavorable
thoughts and lead to increased agreement. However,
if
the predominant thoughts
to a message without distraction were favorable, then distraction should disrupt
these favorable thoughts resulting in decreased agreement. Our manipulation of
argument quality provides a means of assessing this general “thought disrup-
tion” hypothesis as well as testing it against the predicted results from disso-
nance theory.
The thought dipption interpretation holds that distraction should enhance
persuasion for a message containing weak arguments (since unfavorable thoughts
should dominate under no distraction and would therefore be disrupted), but that
distraction should
reduce
persuasion for a message containing strong arguments
(since favorable thoughts should dominate under no distraction and would there-
fore be disrupted). The predictions from dissonance theory are quite different,
however. Research on selective exposure and attention indicates that people
prefer to hear weak rather than strong arguments against their own position
(Kleinhesselink
&
Edwards, 1975; Lowin, 1967), suggesting that exerting effort
to hear strong counterattitudinal arguments would induce more dissonance than
exerting effort to hear weak ones. Because of this, dissonance theory predicts
that for counterattitudinal messages, distraction should enhance persuasion more
for
strong arguments than for weak ones.
Two discrepant messages were prepared for our study. Both messages ar-
gued that tuition at the students’ university should be increased by
20%,
but the
messages differed
in
the presentation of five key arguments. As explained pre-
viously, the strong arguments were selected
so
that they elicited primarily favor-
able thoughts when subjects were instructed to think about them, and the weak
ELABORATION LIKELIHOOD MODEL
OF
PERSUASION
141
arguments were selected
so
that they elicited primarily negative thoughts. The
distraction task required subjects to record on a monitoring sheet the quadrant
in
which
Xs
flashed on a screen in front of them. Subjects were either told that “no
Xs
would flash for now” (no distraction), or the
Xs
appeared on the screen at
15-
(low distraction),
5-
(medium distraction),
or 3-
(high distraction) sec intervals
during the message. After hearing one of the messages over headphones, sub-
jects completed attitude measures, were given
2.5
min
to
list their thoughts, and
responded to ancillary questions. The attitude results are presented
in
Fig.
3,
Box
1.
Consistent with the general thought disruption hypothesis, a significant mes-
sage quality
X
distraction interaction was obtained: increasing distraction was
associated with more favorable attitudes when the message was weak, but in-
creasing distraction was associated with less favorable attitudes when the mes-
sage was strong. Analyses of the postmessage thoughts listed indicated that
overall the messages differed in the number of counterarguments they elicited. In
addition, high distraction reduced counterargument production for the weak, but
not the strong message. Finally, high distraction tended to reduce the number of
favorable thoughts elicited by the strong, but not the weak me~sage.~
Several conceptual replications of our results have been reported. In one
study, we exposed subjects to a strong or weak proattitudinal message under
conditions of either low or medium distraction (Petty
el
al.,
1976,
Experiment
2).
As in our initial study,’a significant message quality
X
distraction interaction
was obtained: distraction was associated with increased agreement when the
message was weak, but with decreased agreement when the message was strong
(see Box
2,
Fig.
3).
In another study, Tsal
(1984)
prepared print ads containing
strong
or
weak arguments for a variety of consumer products. As subjects were
exposed to the ads via slides, they were either not distracted
or
were required to
count the number
of
random “clicks” presented on tape. Again, distraction was
associated with more favorable attitudes toward the products when the arguments
were weak, but with less favorable attitudes when the arguments were strong (see
also, Lammers
&
Becker,
1980).
In sum, the accumulated literature is very consistent with the view that
distraction is one variable that affects a person’s ability to process a message
in
a
relatively objective manner. Specifically, distraction disrupts the thoughts that
would normally be elicited by a message. Distraction should be especially impor-
tant as a thought disrupter when people
are
highly motivated and able to pro-
cess the message. If motivation and/or ability to process the message are low,
distraction should have little effect (see Petty
&
Brock,
1981,
for further discus-
sion).
4Since the thought-listing data parallel the attitude data in nearly all of the studies that we report
here, detailed results
on
this measure will not
be
described
for
the remaining studies that we review.
Readers
are
referred to the original reports.
142
RICHARD
E.
PE'ITY
AND
JOHN
T.
CACIOPPO
10
o,
Weak
-5
II
NO
Low
Med
Hrgh
Distraction
52
59
i
'\
38
Repetition
Involvement
11
o
Weak
5
One Ten
Number
of
Evaluators
,O
Weak
,,_='
,
I
LOW
Medium
Distraction
Strong
:ji
o<
70
\\.
Weak
1
3
5
Repetition
Strong
70
1
3
5
Repetition
-0
Weak
'.
LOW
High
Involvement
8
75
Strong
55
45
b
-0
Weak
LOW
High
Need for Cognition
Fig.
3.
Variables that may enhance
or
reduce elaboration in a relatively objective manner.
(1)
Effects of distraction on attitudes following strong and weak counterattitudinal messages (data from
Petty, Wells,
&
Brock,
1976; Experiment 1).
(2)
Effects of distraction on attitudes following strong
and weak proattitudinal messages (data from Petty, Wells,
&
Brock,
1976; Experiment
2).
(3)
Effects
of message repetition on initial attitudes following strong and weak messages (data from Cacioppo
&
Petty, 1985).
(4)
Effects of message repetition on delayed attitudes following strong and weak
messages (data from Cacioppo
&
Petty, 1980a, Experiment
2).
(5)
Effects of personal relevance on
ELABORATION LIKELIHOOD MODEL
OF
PERSUASION
143
B.
REPETITION
Repetition of stimuli has been shown to increase liking (e.g., Zajonc,
1968), decrease liking (e.g., Cantor, 1968) and have no effect on attitudes (e.g.,
Belch, 1982). The most common finding in the persuasion literature, however, is
that repeating a persuasive communication tends to first increase and then de-
crease agreement (e.g., Cacioppo
&
Petty, 1979b; Calder
&
Sternthal, 1980;
Gorn
&
Goldberg, 1980).
A
variety of theoretical accounts has been proposed for
the effects of repeated exposure, including message learning, response competi-
tion, and others (see reviews by Harrison, 1977; Sawyer, 1981).
Based on the accumulated research, we proposed that message repetition
guides a sequence
of
psychological reactions
to
a persuasive communication best
conceptualized as a two-stage attitude-modification process (Cacioppo
&
Petty,
1979b). In the first stage, repeated presentations of a message provide recipients
with a greater opportunity
to
consider the implications of the content of the
message in a relatively objective manner. Thus, just as distraction can disrupt
information processing, repetition can enhance a person’s ability to process the
message arguments. The benefit of repetition should be most apparent when
additional opportunities are needed to process a message, such as when ability to
process the full implications
of
the message with only one exposure is low (e.g.,
the message is complex), or when motivation to process with one exposure is
low. Once a person has considered the implications of the message, however, the
second stage of information processing commences. In this second stage, the
relatively objective processing of the frrst stage ceases as tedium and/or reac-
tance are elicited by the excessive exposures. Both tedium and reactance will
tend to result in decreased message acceptance either by serving as simple
negative affective cues or by biasing the nature of information processing in a
negative direction (see Section VI1,C). In this section we explore the conse-
quences of the first (objective) stage of information processing.
In order to provide a test of our view that moderate repetition can affect
persuasion by increasing the opportunity to scrutinize arguments
in
a relatively
objective manner, we conducted a study in which students were exposed to a
message advocating that seniors at their university
be
required to take a compre-
hensive exam in their major area as a requirement for graduation (see Cacioppo
attitudes following pro- (strong) and counterattitudmal (weak) messages (data from Petty
&
Caciop-
po,
1979b; Experiment
1).
(6)
Effects of personal relevance on attitudes following strong and weak
counterattitudinal messages (data from Petty
&
Cacioppo, 1979b; Experiment
2).
(7) Effects of
personal responsibility on attitudes following strong, weak, and mixed messages (data from Petty,
Harkins,
&
Williams, 1980; Experiment
2).
(8)
Effects of need for cognition on attitudes following
strong and weak messages (data from Cacioppo, Petty,
&
Moms,
1983; Experiment
2).
144
RICHARD
E.
PETTY AND JOHN
T.
CACIOPPO
&
Petty,
1985,
for details). As in
our
work on distraction, half of the subjects
heard a message containing strong arguments and half heard a message contain-
ing weak arguments. In addition, half of the subjects heard the message once,
and half heard the message three times in succession. An analysis of subjects’
postmessage attitudes toward the senior comprehensive exam issue revealed a
message quality
X
repetition interaction (see Box
3,
Fig.
3).
Subjects showed
greater attitudinal differentiation of strong from weak arguments when the mes-
sage was presented three times rather than just once.
In another study (Cacioppo
&
Petty, 1980a, Experiment
2),
we provided a
conceptual replication and in addition examined the delayed impact of message
repetition. In this study, students were exposed to a strong or weak message in
favor of raising the price of their local newspaper. The message was presented to
subjects as an audiotape of a telephone interview with a local resident. The strong
message emphasized the benefits subscribers would receive from the price in-
crease, whereas the weak message emphasized the benefits to management.
Subjects were instructed to evaluate the sound quality of the tapes, and the
message was played either one, three, or five times in succession. Immediately
following exposure, subjects listed their thoughts about the tapes and rated the
sound quality. From
8
to
14
days later, individuals were contacted by an inter-
viewer who appeared unrelated to the initial experimenter. The second experi-
menter, who was blind to the respondents’ initial experimental conditions, inquir-
ed about a number of community issues including attitudes toward increasing the
price of the local paper. Consistent with the previous study, a message quality
X
repetition interaction was obtained
(see
Box
4,
Fig.
3).
Again, subjects showed
greater attitudinal differentiation of strong from weak arguments as repetition
in-
creased.
C.
PERSONAL RELEVANCE/INVOLVEMENT
We have now discussed two
of
the major variables that can affect a person’s
ability
to scrutinize issue-relevant arguments in a relatively objective manner.
Motivational variables are also important in affecting the likelihood of message
elaboration. Perhaps the most important variable in this regard is the personal
relevance of the message. Previous social psychological analyses of personal
5F0r
exploratory purposes, a third group of subjects received a message containing novel
arguments that were weak but “subtly contradictory.” Subjects exposed
to
this message showed
an
inverted-U attitude pattern with repetition. It is
also
important
to
note in considering the effects of
repetition that the number of repetitions required to enhance argument processing but not induce
tedium
or
reactance will depend
on
a number
of
factors.
For
example, the more complex, the more
lengthy,
or
the more rapidly presented is the message, the more repetitions that may
be
necessary for
the
full
implications
of
the arguments to be realized.
Thus,
what is “moderate” and what is
“excessive” repetition will depend
on
a number
of
factors
(see
Cacioppo
&
Petty,
1985).
ELABORATION LIKELIHOOD MODEL
OF
PERSUASION
145
relevance have labeled this construct (or variations of it) “ego-involvement”
(Rhine
&
Severance, 1970; Sherif, Sherif,
&
Nebergall, 1965), “issue involve-
ment” (Kiesler, Collins,
&
Miller, 1969), “personal involvement” (e.g., Ap-
sler
&
Sears, 1968; Sherif, Kelly, Rodgers, Sarup,
&
Tittler, 1973), “vested
interest” (Sivacek
&
Crano, 1982), and others. In brief, consistent with prevail-
ing definitions, we regard personal relevance as the extent to which an advocacy
has “intrinsic importance” (Sherif
&
Hovland, 1961) or “personal meaning”
(Sherif
et
al.,
1973). Personal relevance occurs when people expect the issue “to
have significant consequences for their own lives” (Apsler
&
Sears, 1968)
.
Of
course, relevance can be judged in terms of a variety of dimensions, such as the
number of personal consequences of an issue, the magnitude of the conse-
quences, and the duration of the consequences. For example, some advocacies
may remain high in personal relevance for many people over a long period of
time (e.g., changing the United States income tax structure), other advocacies
may have personal relevance for a more circumscribed period and/or audience
(e.g., raising college tuition), and
still
other advocacies may have personal
relevance only under certain very transient conditions (e.g., refrigerator ads have
higher relevance when a person is in the market for this appliance).6
Most of the early research on the personal relevance of an issue indicated
that increasing personal involvement was associated with resistance to persua-
sion (Miller, 1965; Sherif
&
Hovland, 1961), and the most prominently men-
tioned explanation for this finding was derived from social judgment theory
(Sherif
et
al.,
1965). Involvement was believed to be associated with a greater
probability of message rejection because people were postulated to hold ex-
panded
‘
‘latitudes of rejection” as personal involvement increased, and incom-
ing messages would therefore be more likely to fall within the unacceptable
range of a person’s implicit attitude continuum (Eagly
&
Manis, 1966). To
account for the fact that increasing relevance was associated with increased
resistance mostly for counterattitudinal and not proattitudinal issues (e.g., Eagly,
1967), Pallak
et
al.
(1972) proposed that increasing involvement (or commit-
ment) increased the probability of rejecting counterattitudinal messages because
these messages were
contrasted
(seen as further away from one’s own position
This kind of “issue relevance” can
be
contrasted with another kind of self-relevance referred
to
as “response involvement” (Zimbardo,
1960)
or
“task involvement” (Sherif
&
Hovland,
1961).
In
this second kind of involvement, the attitudinal issue per
se
is not particularly important
or
relevant
to the
person,
but adopting a position that will maximize the immediate situational rewards is (cf.
Zanna
&
Pack,
1975).
For
example, the issue of raising taxes in the United States has personal
implications for most United States taxpayers (high issue involvementj whereas the issue of raising
taxes in England does not. However, one’s expressed attitude
on
the latter topic may become
important while entertaining one’s British
boss
for dinner (high response involvement).
In
some
cases, response involvement should lead to increased influence (Zimbardo,
1960)
and in other cases
to decreased influence (e.g., Freedman,
1964),
depending upon which enhances self-presentation.
146
RICHARD
E.
PEITY
AND
JOHN
T.
CACIOPPO
than they really were and therefore more objectionable), but proattitudinal mes-
sages were
ussimilured
(seen as closer to one’s own position and therefore more
acceptable).
Importantly, explanations of involvement based on social judgment theory
did not consider the nature of the issue-relevant arguments presented in the
communication. Instead, as involvement increased, a message was thought to
induce increased assimilation (and acceptance) or increased contrast (and rejec-
tion) based on the particular position that it was judged to espouse. The ELM
suggests an alternative analysis of the effects of personal involvement or rele-
vance (Petty
&
Cacioppo, 1979b). Specifically, we suggested that as personal
relevance increases, people become more motivated to process the issue-relevant
arguments presented. As the personal consequences of an advocacy increase, it
becomes more important for people to form a veridical opinion because the
consequences
of
being incorrect are greater. Because of the greater personal
implications people should
be
more motivated to engage in the cognitive work
necessary to evaluate the
true
merits of the proposal.
Much of the early work
on
issue involvement was conducted by finding
existing groups that differed in the extent to which an issue was important (as
assessed by membership in issue-relevant groups), and thus was correlational in
nature (e.g., Hovland, Harvey,
&
Sherif, 1957). More recent investigators have
chosen
to
study issue relevance by varying the issue and message between
subjects (e.g., Lastovicka
&
Gardner, 1979). For example, some undergraduate
students would receive a message on a highly involving issue (e.g., increasing
tuition), whereas others would receive a message on an issue of low relevance
(e.g., increasing park acreage in a distant city; Wine
&
Severance, 1970).
Although this research is interesting in that these involvement classifications
probably capture the personal relevance concept as it often occurs in the “real
world,” several interpretive problems are introduced. Specifically, distinctions
based on different kinds of people or different issues may confound personal
relevance with other factors
(see
discussion by Kiesler
erul.,
1969). One particu-
larly likely confound is that people in the high relevance groups or who receive
the high relevance issues may be more familiar with the issue and may have more
topic-relevant knowledge. Thus, in addition to possessing greater motivation to
process the messages, it is likely that these subjects also have greater ability to do
so.
Thus, when a message contains information that is inconsistent with subjects’
initial opinions, high relevance subjects should be more motivated and generally
more able to generate counterarguments to the arguments presented. However,
when a message contains information that is consistent with the subjects’ initial
attitudes, high relevance subjects should be more motivated and generally more
able to elaborate the strengths of the arguments. In sum, it is possible that
differences in message-relevant elaboration between high and low relevance
subjects (rather than assimilation/contrast effects) may account for the different
ELABORATION
LIKELIHOOD MODEL
OF
PERSUASION
147
effects obtained for pro- and counterattitudinal issues in previous research on
personal involvement.
In order to test our formulation, we first sought to replicate previous re-
search using a manipulation of personal relevance that did not include differences
in familiarity with the issue and arguments as a component. Employing a pro-
cedure introduced by Apsler and Sears
(1968).
we had subjects in both high and
low relevance groups receive the same message on the same topic, but high
involvement subjects were led to believe that the advocacy would affect them
personally, whereas low involvement subjects were led to believe that the ad-
vocacy would have no personally relevant implications.
In our initial experiment (Petty
&
Cacioppo, 1979b, Experiment l), under-
graduate students received either a proattitudinal message extolling the virtues
of
more lenient coed visitation hours on college campuses, or a counterattitudinal
message contending that colleges should be more strict in their coed visitation
policies. The message arguments were pretested
so
that the counterattitudinal
message arguments were weak and elicited predominantly unfavorable thoughts,
and the proattitudinal message arguments were strong and elicited predominantly
favorable thoughts when subjects were instructed
to
think about them.
To
manip-
ulate personal relevance, half of the subjects was told that the speaker was
advocating that the change in visitation hours be implemented at their own
university (Notre Dame), whereas the other half was told that the speaker advo-
cated the change for a distant college (Juanita Junior College). As depicted in
Box
5
of Fig.
3,
a message direction/quality
X
relevance interaction was ob-
tained on the measure of subjects’ attitudes toward the change in visitation
policy. When the message was counterattitudinal (and weak), increased rele-
vance was associated with decreased acceptance, but when the message was
proattitudinal (and strong), increased relevance was associated with greater
acceptance.
Although this study provides evidence consistent with our view that increas-
ing personal relevance enhances motivation to scrutinize message content, it is
still possible that attitude change was mediated by assimilation/contrast effects
since the strong arguments advocated a proattitudinal position and the weak
arguments advocated a counterattitudinal one. To provide a stricter test of the
Elaboration Likelihood Model, we conducted a second experiment (Petty
&
Cacioppo, 1979b, Experiment
2)
in which all subjects were exposed to a coun-
terattitudinal message advocating that college seniors should
be
required to pass
a comprehensive exam in their major area as a requirement for graduation. For
half of the subjects, the arguments in the message were strong and compelling,
and for the other half, the arguments were weak and specious. Finally,
for
half
of
the subjects the speaker advocated that the exam policy be instituted at their own
university (University of Missouri), and for half the speaker advocated imple-
mentation at a distant school (North Carolina State). The results were identical
to
148
RICHARD
E.
PE’lTY
AND JOHN
T.
CACIOPPO
those in the preceding study (see Box
6,
Fig.
3).
A message quality
X
relevance
interaction indicated that as relevance increased, subjects’ attitudes and thoughts
showed greater discrimination of strong from weak arguments. More specifical-
ly, when the message was strong, increasing relevance produced a significant
increase in attitudes, but when the message was weak, increasing relevance
produced a significant decrease in attitudes.
In the context of examining the effects of other variables, we have repli-
cated the interaction of personal relevance and argument quality several times
(e.g., Petty, Cacioppo,
&
Heesacker, 1981; Petty
&
Cacioppo, 1984a). Subse-
quent studies have also supported the view that as personal relevance increases,
information processing increases in intensity and/or complexity (e.g., Harkness,
DeBono,
&
Borgida, 1985; see Burnkrant
&
Sawyer, 1983). Although this
research is consistent with the idea that people become more likely to undertake
the cognitive work of evaluating issue-relevant arguments as personal relevance
increases, several caveats are in order concerning possible limitations on this
effect. First, we suspect that there are some circumstances where personal in-
terests are
so
intense, as when an issue is intimately associated with central
values (e.g., Ostrom
&
Brock, 1968), that processing will either terminate in the
interest of self-protection or will become biased in the service of one’s own ego
(e.g., Greenwald, 1980, 1981).
A
second factor to consider, however, is that, as we noted above, in the
“real world” there is likely to be a natural confounding between the personal
relevance of an issue and the amount of prior thinking a person has done about
the pool of issue-relevant arguments. There are at least two potentially important
consequences of this prior thinking. First, because of the prior consideration,
people may have a greater ability or may be more practiced in defending their
beliefs. This would reduce susceptibility to counterattitudinal appeals. Second,
if
a person has considered an issue many times in the past, it may be more difficult
to motivate the person to think about another message on the same topic because
the person may feel that all arguments have been evaluated (and rejected) al-
ready. This would make it less likely that new compelling arguments would be
processed.
A
final factor to consider is the empirically derived nature of the strong and
weak arguments used in our research. This empirical derivation is an important
methodological tool in that it allows
us
to test the extent of argument processing
induced by different variables. However, in the “real world,” where persuaders
are often confined to posing arguments that
are
veridical (rather than plausible),
it may generally be difficult to generate arguments on some issues that elicit
primarily favorable thoughts when people scrutinize them. Importantly, even
if
all of these factors combine to make it generally more difficult to obtain
in-
creased persuasion with increased personal relevance in the real world, the ELM
accounts for this resistance by tracking the extent to which enhancing relevance
affects the elaboration of the issue-relevant arguments presented.
ELABORATION
LIKELIHOOD
MODEL
OF
PERSUASION
149
D.
PERSONAL
RESPONSIBILITY
We have argued and provided evidence for the view that personal relevance
enhances motivation to process issue-relevant arguments. There is also reason to
believe that personal responsibility produces similar effects. Ever since Ring-
elmann, a German researcher, found that group productivity on a rope-pulling
task failed to reach the levels predicted based on individual performance (see
Steiner, 1972), several contemporary social psychologists have replicated this
effect and pursued its underlying cause. Recent research has documented that at
least part
of
the
reduced performance in groups (called “social loafing” by
LatanC, Williams,
&
Harkins, 1979) results from
loss
of motivation rather than
ability (Ingham, Levinger, Graves,
&
Peckham, 1974; LatanC
et al.,
1979).
Although most of the research following Ringelmann has focused on tasks
requiring physical exertion (e.g., Harkins, LatanC,
&
Williams, 1980; Kerr
&
Bruun, 198
l),
in an exploratoly study we examined the possibility that people who
shared responsibility for a
cognitive
task would exert less
mental
effort than people
who were individually responsible. In this study (Petty, Harkins, Williams,
&
LatanC, 1977) we asked undergraduates to judge a poem and an editorial ostensi-
bly written by fellow students. Our subjects were led to believe that they were the
only one, 1 of 4, or 1 of 16 evaluators. All
of
them actually read the same two
communications, and after exposure to each stimulus they were asked three
questions designed to measure their perceived cognitive involvement in the task
(e.g., to what extent were you trying hard to evaluate the communication?).
Students who were solely responsible for the evaluation reported putting more
effort into their evaluations than those who shared responsibility. Although no
measures
of
actual cognitive effort or work were obtained in our initial study,
subsequent research has obtained relevant evidence. For example, Harkins and
Petty (1982) employed a brainstorming task in which students were asked to
generate uses for objects. The students were either told that “you alone are
responsible for listing uses” or that “you share the responsibility for listing uses
for this object with nine other persons whose uses will
be
combined with yours.”
When confronted with objects for which it was relatively easy to generate uses
(i.e., knife, box), solely responsible subjects generated significantly more uses
than subjects who shared the responsibility (when the task was more difficult and
challenging, no loafing was obtained).
In
three studies, Brickner, Harkins, and Ostrom (1985) asked subjects to list
their thoughts about the implementation of senior comprehensive exams (no
messages were presented). Subjects were either told that they were the only
person listing thoughts or that they shared the responsibility with a partner. In
addition, the. personal relevance
of
the exam proposal was varied by telling
subjects either that the exam proposal was being considered for next year at their
own university or that it was being considered either for a future date or for
another university. When the issue was low in personal relevance, subjects who
150
RICHARD
E.
PE’ITY
AND
JOHN
T.
CACIOPPO
shared responsibility generated significantly fewer thoughts than those who were
individually responsible. As might be expected if personal relevance motivates
issue-relevant thinking (Petty
&
Cacioppo, 1979b), less loafing occurred in
groups when the issue had high personal relevance.
The implications of this research for persuasion are straightforward: the
greater the personal responsibility for evaluating an issue, the more people
should be willing to exert the cognitive effort necessary to evaluate the issue-
relevant arguments presented.
To
test this hypothesis, we asked undergraduates
to provide peer feedback on editorial messages ostensibly written by journalism
students (Petty, Harkins,
&
Williams, 1980, Experiment 2). Subjects were led to
believe that they were either the only person responsible for evaluating an edi-
torial or
1
of 10 people who shared the responsibility. Subjects received one of
three versions of a message arguing that seniors should be required to pass a
comprehensive exam in their major as a requirement for graduation. One mes-
sage contained strong arguments, another contained weak arguments, and a third
contained a mixture of arguments (and elicited a mixture of favorable and un-
favorable thoughts). After reading the appropriate message, subjects provided an
evaluation and listed their thoughts. The attitude results, graphed in Box
7
of
Fig.
3,
revealed a message quality
X
responsibility interaction. As personal
responsibility for evaluation decreased, the quality of the arguments in the mes-
sage became a less important determinant of the evaluations. More specifically,
group evaluators were significantly more favorable toward the weak message,
but were significantly less favorable toward the strong message than individual
evaluators.
As
expected, evaluations of the mixed message were unaffected by
the extent of responsibility.
E. NEED
FOR
COGNITION
Just as there are situational factors that influence the likelihood that indi-
viduals will think about and elaborate upon the arguments provided in a message,
so
too must there be individual factors governing message processing, and,
indirectly, persuasion. Cohen, Stotland, and Wolfe (1955) introduced
an
indi-
vidual difference called the “need for cognition,” which they described as “a
need to structure relevant situations in meaningful, integrated ways. It is a need
to understand and make reasonable the experiential world”
(p.
291). Early
research on this construct suggested that people high in need for cognition made
more discriminating judgments and were more motivated to think about per-
suasive communications (e.g., Cohen, 1957). Unfortunately, the objective tests
used to gauge individual differences in need for cognition were never described
in detail or published, and are apparently no longer available. Because of the
great relevance of individual differences in motivation to think to the ELM and to
ELABORATION LIKELIHOOD MODEL
OF
PERSUASION
151
cognitive social psychology more generally, we developed and validated a new
assessment instrument (Cacioppo
&
Petty,
1982;
Cacioppo, Petty,
&
Kao,
1984).
Specifically, in an initial study, we generated a pool of statements con-
cerning a person’s reactions to engaging in effortful thinking in a variety of
situations (e.g.,
“I
really enjoy a task that involves coming up with new solu-
tions to problems”) and tested them on two groups of people presumed to differ
substantially in their tendencies to engage in and enjoy effortful cognitive en-
deavors (i.e., university faculty vs. assembly line workers). Thus, the need for
cognition scale
(NCS)
was designed to distinguish individuals who disposi-
tionally tend to engage in and enjoy effortful analytic activity from those who do
not (see Cacioppo
&
Petty,
1982, I984b,
for further information about scale
construction and validation).
The results of several studies indicate that individuals high in need for
cognition do indeed enjoy relatively effortful cognitive tasks, even in the absence
of feedback about performance. For example, in one study (Cacioppo
&
Petty,
1982;
Experiment
4),
subjects were given either simple or complex rules to use
in performing
a
boring number circling task. Afterward, subjects were asked to
express their attitudes about the task. Results revealed that subjects generally
disliked the task, but a significant interaction revealed that individuals high in
need for cognition tended to prefer the complex to the simple task whereas
individuals low
in
need for cognition tended to prefer the simple to the complex
task.
In
another study, subjects who were low in need for cognition “loafed” on
a brainstorming task when they were part of a group that was responsible for
generating uses for an object, but subjects who were high in need for cognition
did not loaf on this cognitive task (i.e., they generated the same high number of
uses whether they were solely or jointly responsible; Petty, Cacioppo,
&
Kas-
mer,
1985).
Again, the implications for responses to persuasive communications are
straightforward. If people high in need for cognition tend to engage in and enjoy
effortful cognitive activity, they should be particularly likely to evaluate a mes-
sage by scrutinizing and elaborating the issue-relevant arguments presented.
In
order to test this hypothesis, we exposed high and low need for cognition sub-
jects to a set of strong or weak arguments for a counterattitudinal position
(raising tuition at their university; Cacioppo
el
al.,
1983,
Experiment
2).
After
message exposure, subjects were asked to provide an overall evaluation of the
message arguments and their personal opinion about the issue. Both measures
indicated that subjects high in need for cognition scrutinized the message more
carefully than subjects low in need for cognition. Specifically, the strong and
weak messages induced more polarized evaluations and attitudes for high than
low need for cognition subjects (attitude results are graphed in Box
8,
Fig.
3). In
addition, we reasoned that if subjects high in need for cognition were more likely
to derive their attitudes through a considered evaluation of the arguments central
152
RICHARD
E.
PETTY
AND
JOHN
T.
CACIOPPO
to the recommendation, then there should
be
a stronger association between
message evaluations and attitudes for subjects high than low in need for cogni-
tion. Separate correlations within each group provided support for this hypoth-
esis. As expected, the correlation between argument evaluation and personal
opinion was significantly larger in the high
(r
=
.70)
than the low
(r
=
.22)
need
for cognition group.
VI.
Postulate
5:
Elaboration versus Cues
It is now clear that a wide variety of variables can affect a person’s moti-
vation and/or ability to consider issue-relevant arguments in
a
relatively objec-
tive manner. The implications of this are that when the arguments in a message
are “strong,” persuasion can be increased by enhancing message scrutiny but
reduced by inhibiting scrutiny. However, when the arguments are weak, persua-
sion can be increased by reducing scrutiny, but can be decreased by enhancing
scrutiny. In detailing these processes (depicted in Panel
111,
Fig.
2),
Postulate
4
brings under one conceptual umbrella the operation
of
a seemingly diverse list of
variables such as distraction, repetition, personal relevance, and others, whose
effects had been explained previously with a variety of different theories (e.g.,
dissonance, social judgment). In Section
IX
we discuss additional variables that
affect objective processing.
Although
it
is
now apparent that argument quality will
be
an important
determinant of persuasion when motivation and ability to process message argu-
ments are high, what happens when motivation and/or ability are low? Postulate
5
addresses this issue:
As
motivation and/or ability to process arguments is decreased, peripheral cues
be-
come relatively more important determinants
of
persuasion. Conversely, as argument SCN-
tiny is increased, peripheral cues become relatively less important determinants
of
persuasion.
In the remainder of this section we examine this postulate in regard to
variables affecting processing in a relatively objective manner (e.g., personal
relevance). In Section
VII,
we apply this same postulate to variables affecting
processing in a relatively biased manner.
A.
PERSONAL RELEVANCE/INVOLVEMENT AND THE
OPERATION
OF
CUES
Testing Postulate
5
requires establishing two kinds of persuasion contexts:
one in which the likelihood of message-relevant elaboration is high, and one in
153
ELABORATION LIKELIHOOD MODEL
OF
PERSUASION
which the elaboration likelihood is low. In discussing Postulate
4
we noted
several candidates for varying the elaboration likelihood (e.g., distraction, repe-
tition), but most research pertaining to this postulate has varied the personal
relevance of the communication. In this section we discuss our own work and
other studies in which peripheral cues were tested under different personal rele-
vance conditions. We focus first on source cues, and then on message cues.
1.
Source
Cue
Studies
In our initial investigation of source cues, we asked college students to
listen to a message over headphones that advocated that seniors be required to
pass a comprehensive exam in their major as a requirement for graduation (Petty,
Cacioppo,
&
Goldman, 1981). Three variables were manipulated in the study:
personal relevance, argument quality, and source expertise. In the high relevance
conditions, the speaker advocated that the exam policy be instituted at the stu-
dents’ own university next year, thereby affecting all current students. In the low
relevance conditions, the speaker advocated that the policy begin in
10
years,
thereby affecting no current students. Half of the students heard eight cogent
arguments in favor of the recommendation and half heard eight weak arguments.
Finally, half of the students were told that the tape they would hear was based on
a report prepared by a local high school class, and half were told that the tape was
based on a report prepared by the Carnegie Commission on Higher Education,
which was chaired by a Princeton University Professor. The expertise
of
the
message source, of course, provides a peripheral cue that permits an assessment
of the advocacy without any need to think about the issue-relevant arguments.
Following message exposure, subjects rated their attitudes concerning com-
prehensive exams. In addition to significant main effects for source and argu-
ments (more favorable evaluations with strong than weak arguments, and expert
than inexpert source), two significant interactions provided support for Postulate
5.
First, a relevance
X
message quality interaction replicated our previous find-
ing that argument quality was a more important determinant of persuasion for
high than low relevance subjects (Petty
&
Cacioppo, 1979b). In addition, how-
ever, a relevance
X
source expertise interaction indicated that the source cue was
a more important determinant of attitudes for low than high relevance subjects.
The results for all cells of this study are graphed in the left half of Fig.
4.
In the
top panel it can be seen that under low relevance conditions, increasing source
expertise enhanced attitudes regardless of message quality (a cue effect as de-
picted in the left side of Panel
I1
in Fig.
2).
However,
in
the bottom left panel of
Fig.
4,
it can be seen that under high relevance conditions, source expertise had
no impact on attitudes; only argument quality was important.
In a conceptual replication of this study we employed a different manipula-
tion of relevance, a different issue and arguments, a different cue, and a different
154
RICHARD
E.
PEITY
AND
JOHN
T.
CACIOPPO
.6
.4
.2
E
E
*-
4
-.2
-.4
-
LQW
INVOl
VFMENT
Strong
arguments
,a
Weak
arguments
/
-
-
/
/
-
/
,
0
-
0
0
I
I
I
Low
expertise High expertise
source
source
HIGH
INVOLVEMFNT
.-0
Strong
arguments
Low
expertise High expertise
SOUTCB
source
Fig.
4.
Source factors under high and
low
relevance.
(Left)
Source expertise serves as a
peripheral cue under
low
relevance conditions (top), but
only
argument quality affects attitudes under
high relevance (bottom) (data from Petty, Cacioppo,
&
Goldman,
1981).
(Right) Famous product
endorsers serve
as
a peripheral cue under
low
relevance conditions (top), but
only
product quality
information affects attitudes under high relevance (bottom) (data from Petty, Cacioppo,
&
Schu-
mann,
1983).
method
of
message presentation. In this study (Petty, Cacioppo,
&
Schumann,
1983),
undergraduates were asked to examine a booklet containing
12
magazine
advertisements. Each of the ads was preceded by a brief description of the
purpose of
the
ad.
A
variety of both familiar and unfamiliar ads appeared in the
booklet, but the crucial ad was for a fictitious new product, “Edge disposable
razors.” Two things were done to either enhance or reduce the personal rele-
vance
of
the ad for this product. In the high relevance groups, the ad was
preceded by a description indicating that the product would be test marketed soon
in the subjects’ community.
In
the low relevance groups, the crucial ad was
preceded by a description indicating that the product would
be
test marketed soon
in several distant cities. In addition, all subjects were told before examining any
ELABORATION
LIKELIHOOD
MODEL
OF
PERSUASION
2.0
1.5
j
":I
P
O-
-.5
-1.0
155
-
-
-
-
I
OW
IN
VOI
VEMEK
T
-1.0
-
e--
----------
-0
Weak arguments
I
I
I
I
Citizen
Famous
endorsers
endorsers
HIGH
INVOl
VEtvlJX
T
Strong arguments
1
.o
endorsers
endorsers
Fig.
4.
(continued).
ads that at the end of the experiment they would
be
given a free gift for their
participation. In the high relevance groups, they were told that they would be
allowed to choose among several brands
of
disposable razors. In the low rele-
vance groups, they were told that they would be selecting among brands of
toothpaste (an ad for toothpaste appeared in the ad booklet). In sum, the high
relevance subjects were not only led to believe that the crucial product would be
available in their local area soon, but they also believed that they would make a
decision about the product class. In contrast, the low relevance subjects believed
that the product would not
be
available in their local area in the forseeable future
and did not expect to make a decision about that product class.
Four different versions of the razor ad were constructed. Two featured
photographs of two well-known and -liked sports celebrities, and two featured
middle-aged citizens described as Californians. The product endorsers served as
the manipulation
of
the peripheral cue. Finally, two of the ads contained six
156
RICHARD E. PETTY AND
JOHN
T.
CACIOPPO
persuasive statements about the product (e.g., handle is tapered and ribbed to
prevent slipping) and two ads contained
six
specious or vague statements (e.g.,
designed with
the
bathroom in mind).
Following examination of the ad booklet, subjects indicated their attitudes
about the products depicted, including of course, Edge razors. In addition to
main effects for argument quality and relevance (more favorable attitudes with
strong than weak arguments and low than high relevance), two significant in-
teractions paralleled the results of our previous study (Petty, Cacioppo,
&
Gold-
man, 1981). A relevance
X
message quality interaction revealed that the argu-
ments in the ad were a more important determinant of product attitudes for high
than low relevance subjects, but a relevance
X
endorser interaction revealed that
the status of the product endorsers was a more important determinant of attitudes
for low than high relevance subjects. The results of this study are graphed in the
right half
of
Fig.
4.
In the top panel
it
can be seen that the endorsers served
as
a
simple cue under low relevance conditions (enhancing the effectiveness
of
both
messages). The bottom panel indicates that only argument quality affected at-
titudes in the high relevance conditions.
Other studies have also provided support for Postulate 5 by showing that
simple source cues are more important determinants of persuasion when personal
relevance is low rather than high. For example, in one of the earliest experimen-
tal studies on source expertise, Hovland and Weiss (1951) had subjects read a
message and then told them about the source. The source was either highly
credible or lacked credibility. Four different topics (with appropriate sources)
were used in the experiment. Although Hovland and Weiss in collapsing their
data across the four topics concluded that the high credibility sources produced
more change than the sources of low credibility, an analysis of the credibility
effect for individual topics indicates that the credibility effect was reasonably
strong for the two topics with the lowest direct relevance and prior knowledge
(e.g., “Can a practical atomic powered submarine be built ,in the present
time?”), but was weak and insignificant for the
two
most relevant topics (e.g.,
“As
a result of
TV,
will there be a decrease in the number of movie theaters in
operation by 1955?”).
In a more recent study, Chaiken
(1980;
Experiment
2)
manipulated the
personal relevance of an issue by telling students that their university was consid-
ering switching from a semester to a trimester system either next year or after
they graduated. Subjects either read a message from a likable source who pre-
sented one strong argument or from a dislikable source who presented five strong
arguments. When the issue was of little relevance, the likable source was signifi-
cantly more persuasive than the dislikable source (i.e., the source cue was
effective). When the issue was of high relevance, however, subjects tended to
be
more persuaded by the message with five strong arguments than one even though
the source was dislikable (see also Rhine
&
Severance, 1970).
ELABORATION
LIKELIHOOD MODEL
OF
PERSUASION
157
2.
Message
Cue
Studies
Distinctions between attitude changes based on source factors versus changes
based on message factors have a long history in social psychology (e.g., Kelman
&
Hovland, 1953). In fact, the studies of source cues just described may appear to
provide evidence consistent with the distinctions others have made between source
and message orientations (e.g., Kelman
&
Eagly, 1965; McDavid, 1959; Harvey,
Hunt,
&
Schroder, 1961). However, the centdperipheral distinction of the ELM
is not equivalent to a sourcehessage dichotomy. Importantly, the ELM holds that
both source and message factors may serve as peripheral cues (and both source and
message factors may affect information processing; see Section
IX,B).
Consider a
person who is not motivated or able to think about the actual merits of the
arguments in a message. For this person, it might be reasonable to assume that the
more arguments contained in the message, themore meritorious it is. Although the
literature on persuasion clearly indicates that increasing the number of arguments
in a message is often an effective way to increase persuasion (e.g., Eagly
&
Warren, 1976; Insko, Lind,
&
LaTour, 1976; Maddux
&
Rogers, 1980), most
have argued that this is because with more arguments, people generate and/or
integrate more favorable issue-relevant beliefs (e.g., Calder, Insko,
&
Yandell,
1974; Chaiken, 1980). According to the ELM, it would
be
possible for the number
of arguments in a message to affect issue-relevant thinking in some circumstances,
but to affect persuasion by serving as a simple cue in other situations.
To
test this hypothesis we conducted two studies (Petty
&
Cacioppo,
1984a). In one experiment, undergraduates received a written message on the
topic
of
instituting senior comprehensive exams. For some subjects, the message
had high personal relevance (it advocated that the exam policy begin at their
university next year), and for others the relevance was very low (it advocated that
the exam policy be instituted
in
10
years). Subjects received one
of
four
mes-
sages in favor of the exam proposal. One message contained nine strong argu-
ments, one contained three strong arguments (randomly selected from the nine),
one contained nine weak arguments, and one contained three weak ones (ran-
domly selected from the nine). Following exposure, subjects gave their attitudes
on the exam proposal. A main effect for message quality was obtained as were
two significant interactions. A relevance
X
message length interaction revealed
that the number of arguments in the message was a more important determinant
of persuasion under low than high relevance. However, a relevance
X
message
quality interaction revealed that the cogency of the arguments presented was a
more important determinant
of
persuasion under high than low relevance condi-
tions. The top half
of
Fig.
5
graphs the results. In the left panel it can be seen that
under low relevance, the number of arguments serves as a simple cue, increasing
agreement regardless of argument quality. In the right panel, it can
be
seen that
under high relevance, the number of arguments acts to enhance argument pro-
6
E
siuawn6~e
-1
-z
-&
v
9
-9
-L
I
I
leJM\
1
L
-z
-0
-&
9
-P
f
-5
2
-9
;.
-
L
e.
-8
-6
3
-
01
8
-
11
-
21
ELABORATION
LIKELIHOOD
MODEL
OF
PERSUASION
159
cessing: when the arguments are strong increasing their number enhances persua-
sion, but when their quality is weak, increasing their number reduces persuasion.
In our second study, undergraduates were asked to read one of three mes-
sages. All of the messages concerned a faculty proposal to increase tuition, but
in
the high relevance conditions the proposal was for the students’ own university,
whereas in the low relevance conditions the proposal was for a distant but
comparable university. The message that subjects read contained either three
cogent arguments, three weak arguments or six arguments (three strong and three
weak). After reading the assigned message, subjects indicated their attitudes
toward the idea of raising tuition. Statistical comparison of the messages pro-
cessed under high and low relevance conditions revealed the following (see
bottom half of Fig. 5). When the issue was of low relevance, three strong
arguments did not elicit more agreement than three weak arguments, but the
message with
six
arguments (three strong and weak) elicited more agreement
than either of the three-argument messages. When the message was highly rele-
vant, however, three strong arguments did elicit more agreement than three weak
arguments, but the six-argument message did not enhance persuasion over pre-
senting three strong arguments. Again, argument quantity served as a cue under
low relevance, but argument quality was more important under high relevance.
Langer, Blank, and Chanowitz (1978) explored the importance of the per-
ception of arguments in a field study of compliance. All subjects
in
this study
were standing in line to make copies when a confederate approached them with a
request to make either 5 (low personal consequences) or
20
(high consequences)
copies. The request was accompanied by either a valid reason (“I’m in a rush”),
a “placebic” reason (“I have to make copies”), or no reason. Both kinds of
reasons were more successful than no reason when the personal consequences
were low (i.e., argument quality was unimportant), but the valid reason was
significantly more potent than the placebic reason when the personal conse-
quences were high. Folkes (1985) provided a partial replication of this effect.
In
two field studies using the inconsequential request (making five copies), re-
spondents were equally willing to comply whether the request contained the valid
or the placebic reason. In a third study, however, subjects were asked to guess
how they would respond to the requests and to “think carefully before answer-
ing.” When instructed
to
think before responding, the valid reason produced
significantly more anticipation of compliance than the placebic reason.’ In sum,
’Although providing a partial replication of Langer
er
al.
(1978),
Folkes takes issue with
Langer’s assertion that the placebic information is processed “mindlessly.” Folkes argues that if the
reasons
are
processed automatically under low consequences conditions, then a
poor
reason should
be
as effective as a valid one. However, she found that a
poor
reason (e.g., “because
I
don’t want to
wait”) was significantly less effective than a valid
or
placebic one under low consequences condi-
tions. The
ELM
would predict that the validity
of
a reason would become even
more
important when
the personal consequences
are
high. This was untested in Folkes’ study.
160
RICHARD
E.
PETTY
AND
JOHN
T.
CACIOPPO
as personal relevance or thoughtfulness increases, the quality of issue-relevant
arguments becomes more important than the quantity of arguments provided.
3.
Additional
Cue
Studies
In addition to the research on source and message cues noted above, other
studies have provided support for Postulate 5 by showing that simple cues are
more important determinants of evaluations when personal relevance is low
rather than high. For example, in one study,
Gorn
(1
982) manipulated the per-
sonal relevance of a product (pen) and exposed subjects to ads for two different
brands. One ad was attribute oriented and provided product-relevant information
(e.g., “never smudges”), whereas
the
other ad featured pleasant music rather
than information. Of the subjects in the high relevance condition, 71% chose the
pen advertised with information, but in the low relevance condition,
63%
chose
the pen advertised with the pleasant music
(p
<
.001; see Batra
&
Ray, 1984,
1985, for further discussion on how affectively oriented ads have greater impact
under conditions of low than high involvement).
In two pertinent studies, Borgida and Howard-Pitney (1983) varied the
visual prominence of discussants in a videotaped two-person conversation along
with the personal relevance of the discussion topic. Previous research had shown
that observers’ evaluative judgments and attributions of causality tended to be
more extreme for visually salient than nonsalient actors, a phenomenon called
“top of the head” processing by Taylor and Fiske (1978). Based on the research
we reviewed previously showing that personal relevance enhances message pro-
cessing and reduces
cue
potency, Borgida and Howard-Pitney reasoned that
perceivers’ judgments of the discussion should become less influenced by the
seemingly trivial visual salience cue (and presumably more by the content of the
discussion) as the topic increased in personal importance. Their results supported
this reasoning.
In sum, the accumulated research
on
personal relevance has provided strong
support for Postulate
5
(see also, Taylor, 1975). Some studies have shown that
various simple cues in the situation (i.e., source credibility/likability, mere
number
of
arguments, pleasant music, visual salience) exert a more powerful
effect on judgments when personal relevance is low rather than high. Other
studies have shown that the quality of issue-relevant arguments exerts a more
powerful effect on judgments when personal relevance is high rather than low.
Still other studies have demonstrated both of these effects within the same
experiment (e.g., Petty, Cacioppo,
&
Goldman, 1981).8
Vhaiken
(1980)
argued that just as issue relevance can determine the route
to
persuasion (Petty
&
Cacioppo,
1979b).
so
too can manipulations
of
response involvement, such as varying whether or
not
a
person
expects
to
be
interviewed
on
an issue (see footnote
5).
We suspect that this
is
true mostly
when issue relevance is also reasonably high
(as
it
was in Chaiken’s study; Experiment
I).
If issue
ELABORATION LIKELIHOOD MODEL
OF
PERSUASION
161
B.
OTHER MODERATORS
OF
CUE EFFECTIVENESS
The research that we have just reviewed clearly indicates that the personal
relevance of a message is an important determinant of the route to persuasion.
According to the Elaboration Likelihood Model, however, other variables should
also determine the route to persuasion by affecting a person’s motivation and/or
ability to process the arguments in a message. In discussing Postulate 4, we
identified five variables that affect motivation and/or ability to process a message
in a relatively objective manner. Each of these variables should be capable of
moderating the route to persuasion.
For example, in an early study we showed how distraction disrupted argu-
ment processing resulting in more agreement when the arguments were weak but
less agreement when the arguments were strong (Petty
et
al.,
1976). Just as
arguments become less important determinants of persuasion as distraction is
increased, simple cues should become
more
important determinants of persua-
sion as distraction is increased. Although this hypothesis has not been tested
directly, available research is consistent with this idea. In one study, Kiesler and
Mathog (1968) exposed undergraduates to a variety of relatively involving mes-
sages (e.g., requiring dormitory bed checks).under conditions of either distrac-
tion (copying lists of two-digit numbers) or no distraction. In addition, the
credibility of the message advocacy was manipulated. The study resulted in a
distraction
X
credibility interaction showing that distraction enhanced persuasion
only when the source was highly credible. Consistent with previous theories of
distraction (see Section V,A), this interaction has been accounted for by arguing
that distraction enhances persuasion only when the source is credible because
more credible sources induce more dissonance, or because more credible sources
induce more counterarguing (Baron
et
al.,
1973; Kiesler
&
Mathog, 1968; Petty
&
Brock, 1981). The ELM provides a different yet equally plausible account for
this effect. Rather than emphasizing the finding that distraction enhances persua-
sion when source credibility is high, the ELM views the interaction as showing
that credibility enhances persuasion when distraction is high (Petty
&
Cacioppo,
1984~). In other words, when people are disrupted from processing the issue-
relevant arguments by distraction, simple cues in the persuasion context become
more powerful determinants of influence.
relevance is low, but response involvement is high, impression management motives (rather than
concerns about adopting a veridical position based on examination of issue-relevant arguments) may
determine the attitude expressed (see Cialdini, Levy, Herman, Kozlowski,
&
Petty,
1976).
Although
it is possible for impression management concerns to lead to extensive issue-relevant cognitive
activity in some situations (e.g., a student assigned
to
argue in a public debate may carefully research
the position in order to make a favorable impression), more typically, impression management
concerns may not necessitate a careful evaluation
of
issue-relevant arguments (Cialdini
&
Petty,
1981;
Moscovici,
1980).
162
RICHARD
E.
PE7TY
AND
JOHN
T.
CACIOPPO
In addition to personal relevance and distraction, the other variables dis-
cussed under Postulate 4 should also be moderators of the route to persuasion.
For example, we have already noted that argument quality becomes a more
important determinant of persuasion as people feel more personal responsibility
for message evaluation (Petty
et
al.,
1980), and for individuals high rather than
low in need for cognition (Cacioppo
el
al.,
1983). Although it has not yet been
tested, the ELM expects that peripheral cues in the persuasion context should
generally be more important for group than individually responsible message
evaluators, and for individuals low rather than high in need for cognition. Before
concluding this section, we note two additional variables that appear to moderate
the route to persuasion.
One previously unmentioned variable that appears to affect the extent of
issue-relevant thinking is the modality of message presentation. In general, audio
and video presentations compared
to
print give people less opportunity to process
issue-relevant arguments because exposure is forced rather than self-paced.
Thus, presenting messages in written form should be especially important when
the
arguments are complex and difficult to process rapidly (Chaiken
&
Eagly,
1976). On the other hand, if it is generally more difficult to process issue-
relevant arguments when exposure is forced rather than self-paced, simple cues
in the persuasion context should
be
more powerful determinants of persuasion in
the former than in the latter modality. Studies which have manipulated medium
of presentation and source cues have supported this proposition. Thus, both
source credibility (Andreoli
&
Worchel, 1978) and likability (Chaiken
&
Eagly,
1983) have had a greater impact on attitudes when a message
was
presented on
video or audio
tape
rather than in written form.
Interestingly, the nature of the message itself has also been implicated as a
determinant
of
whether a person processes mostly issue-relevant arguments, or
searches for simple cues to determine message acceptability. For example, re-
search suggests that messages that
are
either overly vague (Pallaker
al.
,11983),or
overly quantified (Yalch
&
Elmore-Yalch, 1984), may induce reliance on
pe-
ripheral cues. The ELM would expect this to occur to the extent that these
messages reduce either subjects' ability (vague message) or motivation (overly
quantified message) to process issue-relevant arguments (Witt, 1976).
VII.
Postulate
6:
Biased
Elaboration
We have now seen that a wide variety of variables can moderate the route to
persuasion by increasing or decreasing the extent to which a person is motivated
or able to process the issue-relevant arguments in a relatively objective manner.
As
we noted in discussing Postulate 3, however, variables can also affect persua-
ELABORATION LIKELIHOOD MODEL
OF
PERSUASION
163
sion by affecting motivation and/or ability to process message arguments in a
more biased fashion. Specifically, Postulate
6
states:
Variables affecting message processing in a relatively biased manner can produce
either a positive (favorable)
or
negative (unfavorable) motivational and/or ability bias to the
issue-relevant thoughts attempted.
As
we will see, there are a number of ways to induce biased processing, but
often the bias results from a person’s initial attitude becoming a more important
schema in guiding processing (e.g., Tesser, 1978). Panel
IV
in Fig.
2
graphs the
expected results for a variable that biases information processing activity. In the
left half of the panel, the effects of a variable that produces a positive cognitive
bias (enhancing favorable thoughts and/or reducing negative thoughts)
is
de-
picted. It is instructive to compare this pattern with the pattern
of
data
in
the two
panels above it. First note that
unlike
a variable operating as a simple positive cue
(left half of panel 11), a variable producing a positive processing bias
is
not
expected to affect all messages equally. Since the pure cue processor is not
elaborating message arguments at all, the effectiveness of the cue is not con-
strained by the arguments presented. The biased processor, however, is attempt-
ing to process the arguments and in this regard is similar to the objective pro-
cessor. Nevertheless, an important difference between objective and biased
processing exists. The objective processor is motivated or is able to discover the
“true validity” of the message, and thus strong arguments induce more persua-
sion and weak arguments induce less persuasion with more processing.
In
stark
contrast, the biased processor is either particularly motivated or able to generate
a particular kind of thought, often in defense of an initial attitude. However, even
though the person is biased in processing a communication, the arguments in the
message pose some limitation on this bias. For example, consider a person who
is truly motivated to
counterargue
(and not simply discount) an advocacy. This
person’s task is simpler to the extent that the message provides weak rather than
strong arguments in support of its position (see right half of Panel
IV,
Fig.
2).
Figure
2
summarizes the ways in which a treatment can affect attitude
change according to the
ELM,
and it shows how these different processes can be
tested by varying argument quality. First, a treatment can have no effect on
persuasion for either strong or weak arguments (such as a peripheral cue under
conditions
of
high elaboration likelihood; Panel
I).
Second, a treatment may
produce only a main effect (Panel
11).
If
so,
it suggests that the treatment is
operating as a simple positive or negative cue (low elaboration likelihood condi-
tions). However, if a treatment interacts with message quality, it suggests that
the treatment
is
affecting the elaboration likelihood. If the interaction follows the
form depicted in Panel 111 of Fig.
2,
it suggests that the processing is relatively
objective. If a treatment main effect and an interaction as depicted in Panel
IV
of
Fig.
2
is obtained, it suggests that the treatment is biasing information pro-
cessing.
164
RICHARD
E.
PEITY AND
JOHN
T.
CACIOPPO
As
Panel
IV
indicates, a treatment which biases thinking
in
a positive
direction should generally have a greater impact on a strong than a weak message
because it will be more difficult for a person to generate favorable thoughts to
weak than strong message arguments. On the other hand, a variable which biasis
thinking in a negative direction should generally have a greater impact on a weak
than a strong message because it will be more difficult for a person to generate
counterarguments to strong than weak arguments.
Importantly, these predictions (and the depictions in Panel
IV
of Fig.
2)
assume that in the baseline (control) condition, relatively little issue-relevant
thinking is occurring. However, consider a control (comparison) ,condition
in
which subjects are maximally processing strong and weak arguments. If the
experimental treatment includes a variable that biases thinking
in
a positive
direction,
it
will be difficult to observe more favorable attitudes to the strong
arguments in the experimental than the control condition since the arguments are
already being processed maximally in the control condition (i.e., a ceiling effect
is operating). However, the positive bias may result
in
more favorable attitudes
toward the weak message than observed in the control condition (since no ceiling
effect is operating). Thus, it may appear that the positive bias is working better
for the weak than the strong message.
Similarly, if the experimental treatment includes a variable that biases
thinking
in
a negative direction, it will be difficult to observe more negative
attitudes toward the weak arguments than
in
the control condition if control
subjects are highly motivated and able to process the message objectively (with-
out bias). Thus, it may appear that the negative bias is working better for the
strong than the weak message (because of a
floor
effect for the weak arguments).
The caveat here is to include an appropriate control
or
baseline condition
so
that
ceiling and
floor
effects are not problems. In general, when testing variables
hypothesized to enhance processing, it is better to include control conditions
in
which processing is minimal. When testing variables hypothesized to reduce
processing, the opposite holds.
In the remainder of this section we review evidence consistent with the view
that some variables affect information processing in a relatively biased rather
than a relatively objective manner. Importantly, a consideration of Postulates
5
and
6
together indicates that just as there is a tradeoff between a person’s
motivation and ability to process a message in a relatively objective manner and
the effectiveness of peripheral cues,
so
too is there a tradeoff between biased
processing and the operation of cues.
As
argument scrutiny is reduced, whether
objective
or
biased, peripheral cues become more important determinants of
persuasion.
As
argument scrutiny is increased, whether objective
or
biased,
peripheral cues become less important. We now
turn
to some of the major
variables affecting information processing
in
a relatively biased manner, and
consider both message processing effects and the operation of peripheral cues.
165
ELABORATION LIKELIHOOD MODEL
OF
PERSUASION
A.
PRIOR
KNOWLEDGE
One
of
the most important variables affecting information processing ac-
tivity is the extent to which a person has an organized structure
of
knowledge
(schema) concerning an issue (Britton
&
Tesser, 1982; Higgins, Herman,
&
Zanna, 1981; Wyer
&
Srull, 1984). Although it is possible for prior knowledge
to enable more objective information processing in some instances (Bobrow
&
Norman, 1975), since stored knowledge tends to be biased in favor of an initial
opinion, more often than not this prior knowledge will enable biased scrutiny of
externally provided communications (Craik, 1979; Taylor
&
Fiske, 1984). Spe-
cifically, schema-driven processing tends to be biased such that external infor-
mation is processed
in
a manner that contributes to the perseverance of the
guiding schema (e.g.,
Ross,
Lepper,
&
Hubbard, 1975). Thus, the more issue-
relevant knowledge people have, the more they tend
to
be able to counterargue
communications opposing their initial positions and to cognitively bolster (pro-
argue) congruent messages (e.g., Lord,
Ross,
&
Lepper, 1979).
I.
Message Processing Efects
The impact of knowledge structures
on
attitude-relevant processing is
shown clearly in Tesser’s program
of
research on the effects of “mere thought”
(e.g., Sadler
&
Tesser, 1973; Tesser
&
Conlee, 1975; Tesser, 1976).
In
a series
of studies, Tesser has shown that when instructed to think about an issue or
object, attitudes tend to become more polarized in the direction of their initial
tendency (i.e, they become more schema consistent; see Tesser, 1978, for a
review). Importantly, this polarization effect requires that subjects have an orga-
nized store of issue-relevant information to guide processing, and that they are
motivated to employ this issue-relevant knowledge in defense
of
their initial
opinions.
In
the absence of these conditions, mere thought may not lead to
polarization (Linville, 1982; Millar
&
Tesser, 1984; Tesser
&
Leone, 1977).
Although Tesser’s research has focused
on
situations in which
no
message
is provided to subjects, similar schema-driven processing can be observed when
people evaluate persuasive messages. For example,
in
one study we exposed
subjects to a proattitudinal message that was either relevant or irrelevant to a self-
schema (Cacioppo, Petty,
&
Sidera, 1982). Our hypothesis was that schema-
relevant messages would be more likely to invoke schematic processing than
irrelevant messages (e.g., Cantor
&
Mischel, 1979), and that schema activatiqn
would enhance a person’s ability to cognitively bolster the congruent message.
Employing a procedure adapted from Markus (1977), we identified two groups
of students who were attending a major Catholic university. Some of the students
were categorized as possessing a “religious” self-schema whereas others were
categorized as possesing a “legalistic” schema. Subjects received a generally
166
RICHARD
E.
PE’ITY
AND
JOHN
T.
CACIOPPO
weak message that supported their own opinions
on
an issue (e.g., against
government support of abortion), and the message either employed a religious or
a legalistic perspective in the arguments presented. An analysis of subjects’
ratings of message persuasiveness revealed a schema type
X
message type
in-
teraction: the legalistic message was seen as more persuasive by the legalistic
than the religious subjects, and the religious message was seen as more per-
suasive by the religious than the legalistic subjects. In addition, recipients gener-
ated more topic-relevant thoughts when the message was reflective than when
it
was unreflective of their self-schema. Further analyses revealed that this effect
was accounted for mostly by the increased generation of favorable thoughts to
schema-reflective messages.
If a message is inconsistent with a person’s initial opinion, however,
it
would be expected that prior knowledge would enhance the person’s ability to
counterargue the message. In a test of this hypothesis, Wood
(1982;
Experiment
1)
assessed the prior knowledge and experience people had on the issue
of
environmental preservation by asking them to list their beliefs and previous
behaviors concerning environmental preservation. Subjects were divided into
high and low belief and behavior retrieval
groups
based on a median split on the
number of beliefs and behaviors listed. Consistent with the view that this assess-
ment technique taps prior knowledge, subjects who generated more behaviors
indicated that they had thought more about preservation, knew more about the
topic, and were more involved than subjects who generated fewer behaviors (no
effects were found for belief retrieval, however). One to two weeks later, sub-
jects returned and read a counterattitudinal message providing four arguments
against environmental preservation. After message exposure, subjects reported
their attitudes and gave their thoughts. Subjects who had high prior knowledge
changed less in the direction of the message than subjects with low prior knowl-
edge. In addition, subjects with high prior knowledge (as assessed by behavior
retrieval) generated more counterarguments and fewer favorable thoughts
in
response to the message. In sum, the available research is generally consistent
with the view that prior knowledge enables counterarguing of incongruent mes-
sages (Wood,
1982)
and bolstering of proattitudinal ones (Cacioppo
et
al.,
1982).
2.
Cue
Effects
Research is also generally consistent with the view that simple cues or
decision rules are more likely to affect susceptibility to influence when prior
knowledge is low rather than high. One
cue
that has been studied in the context
of previous issue-relevant knowledge is gender. Previous studies of sex dif-
ferences in persuasion have provided some support for the view that females are
more susceptible to influence than males in some contexts (see reviews by
ELABORATION LIKELIHOOD MODEL
OF
PERSUASION
167
Cooper, 1979; Eagly
&
Carli, 1983), and one explanation for this effect is based
on the idea that females may have been socialized to be more agreeable (i.e.,
concerned with social harmony) than males (e.g., Eagly, 1978).
To
the extent
that females have learned to be more agreeable and less dominant than males, the
invocation of this socialized female gender role
or
category (cf. Deaux, 1984)
could lead to a sex difference in influenceability. However, according to the
ELM, attitude expression based on the female gender role should
be
more likely
when women have little ability to process the issue-relevant information present-
ed than when ability is high.
In a test
of
this hypothesis, we exposed male and female undergraduates to
photographs relevant to domains for which men and women had rated their
knowledge differently (Cacioppo
&
Petty, 1980b). Half of the photos depicted
football tackles (high male knowledge) and half depicted current fashions (high
female knowledge). Each photograph was accompanied by a set of comments
attributed to another subject. The comments were either completely factual and
descriptive (e.g., the dress is blue, the runner’s feet are off the ground) or
included an evaluation (e.g., that’s a great tackle) that was either accurate or
inaccurate. Subjects were asked to rate the extent
to
which they agreed with the
comments made by the other subject. The ELM suggests that to the extent that
gender roles provide simple rules as to how one should behave (e.g.,
“As
a
woman,
I
should maintain harmony”), such rules should operate mostly when
ability (and/or motivation) to evaluate the stimuli are low. In our study, when the
comments were completely factual and easily verifiable, both males and females
should be equally able to evaluate the comments whether they concerned fash-
ions or football; thus, there should be no sex differences in extent of agreement.
When the comments were evaluative rather than descriptive, however, knowl-
edge
is
required
to
confidently evaluate the statements.
For
football tackles,
then, the invocation of the female gender role should lead to women showing
more agreement than men whether the evaluations were accurate
or
inaccurate.
Actual accuracy should make little difference because in both cases women
would have little confidence (due to low knowledge) in their judgments. When
the judgments concerned fashions, however, women do have the requisite
knowledge and confidence to make judgments. Thus, they should be more ac-
cepting of the accurate evaluations, but less accepting of the inaccurate evalua-
tions than men. The data from our study generally conformed to this pattern.
Other research has also supported the view that prior knowledge
is
an important
determinant of sex differences in influenceability (e.g., Karabenick, 1983; Sis-
trunk
&
McDavid, 1971).
Another simple rule that people sometimes use
is
based on observation of
their own behavior and the situational constraints imposed upon it (i.e., the self-
perception principle; Bem, 1967, 1972).
For
example, if an initially agreeable
behavior is overjustified, people may reason that their behavior
is
governed more
168
RICHARD
E.
PEITY
AND
JOHN
T.
CACIOPF'O
by the reward than their attitude and come to evaluate the behavior less positively
(e.g., Lepper, Greene,
&
Nisbett, 1973; see Deci
&
Ryan, 1980). Wood (1982;
Experiment
2)
reasoned that this relatively simple inference process based
on
a
behavioral cue should be a more potent determinant of attitudes for people who
have relatively little knowledge on a topic. In a test of this hypothesis, she found
that low knowledge subjects used a monetary incentive to make an inference
about their attitudes (as self-perception theory would expect), but high knowl-
edge subjects were unaffected by this simple external cue (see Chaiken
&
Bald-
win, 1981; and Fiske, Kinder,
&
Larter, 1983; for additional evidence).
Finally, we note that simple affective cues may be a more important deter-
minant of attitudes when prior knowledge is low rather than high. For example,
Srull (1983) had subjects rate their general knowledge about automobiles (cf.
Bettman
&
Park, 1980; Johnson
&
Russo, 1981). Following a mood-induction
procedure in which subjects were placed in a positive, negative, or neutral mood,
they were exposed to
an
attribute-oriented ad for a new car and then asked to
evaluate it. The attitudes of low knowledge subjects (as determined by a median
split) were significantly affected by the mood manipulation, but attitudes of high
knowledge subjects were not influenced by this simple affective cue.
3.
Testing the
ELM
In
general, research
on
prior knowledge has provided support for the ELM
view that when prior knowledge is low, simple cues in the persuasion context
affect influence, but when prior knowledge is high, message processing is biased
because previous knowledge enables the counterarguing of incongruent mes-
sages and the bolstering of congruent ones. However, more definitive support for
the ELM analysis of prior knowledge requires a study in which knowledge is
varied along with argument strength and a peripheral cue.
Fortunately, Wood, Kallgren, and Priesler (1985) reported such a study. In
this study, Wood and colleagues asked undergraduates to list their beliefs and
behaviors relevant to environmental preservation. Subjects were divided into
three groups based
on
a combination of the total number of beliefs and behaviors
listed (creating high, medium, and low knowledge groups). Subjects returned
1
to
2
weeks later and were exposed to one of four persuasive messages. The
messages differed in both the strength and length of the arguments presented.
Two of the messages contained three strong arguments favoring an antipreserva-
tion view and two messages contained three weak arguments. Two versions of
each argument were developed, however. One version contained short concise
statements of the arguments, and the other contained longer more wordy versions
of essentially the same information. The long and short versions were equated in
terms of strength and ease of comprehension.
ELABORATION
LIKELIHOOD
MODEL
OF
PERSUASION
45-
40-
35
30
25
169
a
Strong
\
y,
arguments
\
\
\
\o
Weak
\
\
\
\
-
-
arguments
-
I
I
Favorable
Unfavorable
b
LOW
High
Know ledge Knowledge
Fig.
6.
Prior
knowledge biases message processing. (a) Increased knowledge makes it easier
to counterargue weak than strong message arguments. (b) Length
of
arguments serves as a peripheral
cue
for
low but not high knowledge subjects (data adapted from
Wood,
Kallgren,
&
Priesler,
1985;
see Footnote
91.
After exposure to one of the versions of the message, subjects indicated
their attitudes on the topic of environmental preservation. Overall, a main effect
for knowledge was obtained indicating that as knowledge increased, subjects
were more resistant to the counterattitudinal appeal. In addition, individual cell
comparisons revealed that the attitudes of high knowledge subjects were affected
by argument quality, but the attitudes of low knowledge subjects were not.
A
closer inspection of this interaction pattern (graphed in Fig. 6a) indicates that
although high knowledge subjects were generally more resistant to all messages
than low knowledge subjects, this was especially true for the message containing
weak arguments.
As
noted previously, this particular interaction pattern suggests
that high knowledge subjects were better able (and perhaps more motivated) to
counterargue the incongruent message, but that it was more difficult to coun-
terargue the strong than the weak version of it (cf. Panel
IV,
Fig.
2).
In addition,
planned comparisons indicated that the attitudes of low knowledge subjects were
affected by argument length, but the attitudes of high knowledge subjects were
not
(see
Fig. 6b). In sum, low knowledge subjects’ attitudes were affected by the
simple cue
of
message length, but high knowledge subjects used their prior
knowledge in an attempt to defend their attitudes. They were more success-
ful in doing this when the arguments in the message were weak rather than
~trong.~
9For
ease of exposition we have graphed the data
based
on a median split on knowledge
(W.
Wood,
personal communication, October
18,
1984)
rather than the three-way split reported in the
published article.
As
might be expected, the three-way split only enhances the differences between
high and low knowledge groups, though the median split is based on a larger sample size.
170
RICHARD
E.
PEITY
AND
JOHN
T.
CACIOPW
B
.
FOREWARNINGS
Just as some variables generally enhance a person’s ability to engage in
biased processing of a persuasive message, such as prior knowledge,
so
too can
variables enhance a person’s
motivation
to process a message in a biased fashion,
even if ability is held constant. For example, McGuire (1964) argued that inocu-
lation treatments enhance resistance to persuasion mostly by increasing peoples’
motivation to defend their beliefs. The persuasion literature has identified other
variables that increase motivation to defend beliefs, and the most researched
category of variables is “forewarning” (see Smith, 1982).
Papageorgis (1968) noted that two conceptually distinct kinds of warnings
have been studied by persuasion researchers. One kind of treatment forewarns
message recipients of the upcoming topic and/or position of the persuasive
message (warning
of
message content). A second kind of treatment suggests
to
subjects that they are the targets of an influence attempt (forewarning of per-
suasive intent). Although some studies have explored
the
effects of combining
the
two
kinds of forewarnings (e.g., Allyn
&
Festinger, 1961; Brock, 1967), it is
possible to study their effects separately. Also, although many studies have
shown that forewarnings can reduce the persuasive impact of a message, other
studies have shown that forewarnings can enhance persuasion (e.g., Cooper
&
Jones, 1970; Mills, 1966). Importantly, which effect
is
obtained appears to
depend largely on a person’s motivation and ability to think about the issue. With
low motivation and/or ability, forewarnings have tended
to
either have no effect
or to enhance change in the direction of the advocacy. With high motivation and
ability, however, resistance to persuasion has generally resulted (e.g
.
,
Apsler
&
Sears, 1968; Freedman, Sears,
&
O’Conner, 1964; Petty
&
Cacioppo, 1979a).
According to the ELM, when motivation (e.g., personal relevance) and/or abil-
ity (e.g., prior knowledge)
to
think about an issue are low, forewarnings should
enhance the salience of various cues (e.g., attractive sources; e.g., Mills
&
Aronson, 1965) or motives (e.g., impression management; e.g., Cialdini, Levy,
Herman,
&
Evenbeck, 1973) in the situation that are capable of producing
attitude change without issue-relevant thinking. When motivation and ability
are
high, however, forewarnings should modify attitudes by affecting issue-relevant
thinking.
I.
Warning
of
Message Content
A forewarning of message content gives message recipients advance indica-
tion of what the message is about. McGuire and Papageorgis (1962) hypoth-
esized that the advance warning would motivate recipients to begin considering
information that would support their beliefs and counterargue opposing argu-
ments (i.e., biased processing). Consistent with this view, a content forewarning
ELABORATION LIKELIHOOD MODEL
OF
PERSUASION
171
is most effective when there is a sufficient time delay between the warning and
message to allow thinking (Freedman
&
Sears, 1965; Hass
&
Grady, 1975; Petty
&
Cacioppo, 1977). In addition, studies employing the thought-listing procedure
(e.g., Brock, 1967; Petty
&
Cacioppo, 1977) and psychophysiological assess-
ments (Cacioppo
&
Petty, 1979a) have supported the view that when confronted
with an impending counterattitudinal message on
an
involving issue, people use
the period between the forewarning and the message to bolster their initial
opinions. This “biased scanning” of arguments on the issue (cf. Janis
&
Gilmore, 1965) enables greater resistance to the subsequent message.
If accessing one’s issue-relevant information prior to a persuasive attack
assists in resisting the subsequent message, then a forewarning of the impending
a
Message
Conditions
’
1
‘
Actual
Thoughts Topic Thoughts
’
-
0
v)
36
E
-
-
35
No
message
control
0
Q)
U
3
4
e
3
c.
.-
2
Unfavorable
Unwarned Warned Unwarned Werned
U
nwarned
b
High
Relevance
Unwarned Warned Unwarned Warned
Fig. 7. Forewarnings bias message processing. (a) Forewarning
of
an involving counterat-
titudinal message topic and anticipatory topic-relevant thinking produce resistance to a message with
strong arguments (data
from
Petty
&
Cacioppo, 1977; Experiment
2).
(b) Forewarning
of
persuasive
intent produces resistance to a message with strong arguments
on
a counterattitudinal topic when the
issue is
of
high relevance (data from Petty
&
Cacioppo, 1979a).
172
RICHARD
E. PETTY
AND
JOHN
T.
CACIOPPO
message topic is not really necessary for resistance, but rather it is necessary for
people to access their issue-relevant knowledge in preparation for the message.
In a study testing this hypothesis (Petty
&
Cacioppo, 1977, Experiment 2), we
told students in an introductory psychology class that they soon would
be
hearing
a guest lecture from a counseling psychologist. Half of the students were warned
several minutes
in
advance of the lecture that the speaker would advocate that all
freshmen and sophomores be required to live in campus dorms (an involving
counterattitudinal issue). The remaining subjects were unaware of the topic of
the speech. After
3
min
of
sitting quietly, the students were given either three
additional minutes to list the thoughts that occurred to them during the preceding
minutes (actual thoughts) or they were instructed to list their thoughts on the
topic of requiring freshmen and sophomores to live in dorms (topic-relevant
thoughts). Following this procedure, the guest speaker presented a five-min
advocacy on the topic, and the students’ attitudes were measured.
A
control
group of subjects responded to the attitude measure prior to the warning and
advocacy.
The results of the study are depicted in Fig. 7a. The data for subjects
in
the
“actual thoughts” groups were in accord with previous research employing
involving counterattitudinal issues (e.g., Freedman
&
Sears, 1965). Specifically,
the unwarned subjects were influenced by the message, but the warned subjects
did not differ from controls. More interestingly, however, is that when subjects
listed their thoughts about the message topic prior to receiving it, the warning
had no unique effect. Subjects resisted the message whether they were warned or
not. The resistance of the unwarned group that accessed issue-relevant cognitions
prior to message exposure indicates that it is not the forewarning per se that
induces resistance. Rather, the accessing of attitude-supportive beliefs prior to
message exposure (which can be triggered by a warning) biases message process-
ing and thereby facilitates resistance (cf. Miller, 1965).
2.
Warning
of
Persuasive Intent
A forewarning of persuasive intent must work differently than
a
warning of
message content because a warning of intent does not indicate the topic
of
the
message. Thus, this kind of warning cannot enable a potential recipient to access
the relevant store of issue-relevant cognitions prior to message exposure. As
might be expected, then, unlike a warning of message content, a warning of
persuasive intent is equally effective in inducing resistance whether
it
immediate-
ly precedes or comes several minutes before message exposure (Hass
&
Grady,
1975). What then is the psychological process responsible for the resistance
conveyed by making the persuasive intent of a message salient? Brehm (1966,
1972) has argued that restricting a person’s perceived freedom to think or act in a
particular way arouses a psychological state of “reactance” that motivates peo-
ELABORATION LIKELIHOOD MODEL
OF
PERSUASION
173
pie to restore their freedom. When a speaker announces an intention to persuade
a recipient, this may be perceived as a direct threat to the person’s freedom to
hold a particular attitude (Hass
&
Grady, 1975). One way to demonstrate or
reassert freedom, of course, is to resist the persuasive message.
According to the
ELM,
a warning of persuasive intent may induce re-
sistance in one
of
three
ways. First, the warning may serve as a simple rejection
cue leading the person to discount the message without considering it (cue
effect). Second, the warning may lead the person to more carefully scrutinize the
message arguments leading to resistance when the arguments
are
weak but
not
when they are strong (objective processing). Finally, the warning may motivate
the recipient to actively counterargue the message drawing upon previous knowl-
edge in order to attack the message to the best
of
one’s ability (biased process-
ing). The results of several studies suggest that the latter process is responsible
for the resistance conveyed by a warning of persuasive intent.
For example,
in
one study, Kiesler and Kiesler (1964) varied whether the
information about persuasive intent preceded or came after the message. Per-
suasive intent only reduced persuasion when it came before the message. If the
statement of intent served as a simple rejection cue, it should have produced
resistance regardless of its position. In another study (Watts
&
Holt, 1979), a
warning of persuasive intent given before the message reduced persuasion only
when the message was not accompanied by distraction. The fact that distraction
during the message eliminated the effect of the forewarning is consistent with the
view that a warning works by affecting ongoing message processing. When this
processing
is
disrupted by distraction, the warning is ineffective.
Although these studies are consistent with the idea that a forewarning of
intent affects message processing rather than serving as a simple cue, they do not
indicate whether the processing is relatively objective or biased. In a study
designed to explore this issue (Petty
&
Cacioppo, 1979a), we told students that
they would
be
evaluating radio editorials. Some were further told that the tape
that they would hear “was designed specifically to
try
to persuade you and other
college students of the desirability of changing certain college regulations.
”
Others were simply told that the tape was prepared as
part
of a journalism class
project. In addition to the warning manipulation, the personal relevance of the
advocacy was manipulated. Some subjects were led to believe that the advocated
change would affect them personally because the change would be implemented
next year (high relevance), whereas others were led to believe that it would not
affect them either because the change would not take effect for 10 years (low
relevance-date) or because it was proposed for next year but for a different
university (low relevance-place). All subjects received a message containing five
strong arguments in favor of requiring seniors to take a comprehensive exam as a
requirement for graduation.
The results of this study revealed a main effect for warning and
a
warning
x
174
RICHARD
E.
PElTY
AND
JOHN
T.
CACIOPPO
relevance interaction on the postmessage attitude measure (see Fig. 7b).
Al-
though the warning decreased agreement overall, the effect was only significant
under the high relevance conditions. The fact that the warning worked better
under high than low relevance again suggests that the warning is not operating as
a simple rejection cue. As we detailed in Section VI,A, cues tend to work better
under low
than
high relevance conditions. Also, since the warning reduced
persuasion even though the arguments were strong, this suggests that the warning
induced biased rather than objective processing. When subjects were not warned,
increasing involvement enhanced persuasion as would be expected if the argu-
ments were strong and relevance increased subjects’ motivation to process the
arguments in
a
relatively objective manner (Petty
&
Cacioppo, 1979b). It ap-
pears that when
a
forewarning of persuasive intent was introduced under high
involvement, the nature
of
the information processing changed as subjects be-
came less objective, and more intent on finding fault with the message arguments
in order to reassert their attitudinal freedom. Consistent with this reasoning,
under high involvement, subjects who were warned generated significantly more
counterarguments and fewer favorable thoughts than unwarned subjects in a
postmessage thought listing.
C.
OTHER BIASING TREATMENTS
Although we have focused on two highly researched variables that appear to
enable or motivate biased information processing (prior knowledge and fore-
warning), other treatments may also bias the nature of message processing. We
have already noted that McGuire’s (1964) discussion of inoculation treatments
provides a cogent example.
In our own research we have suggested several procedures for biasing the
processing of a persuasive message. For example, in one study (Petty
&
Brock,
1979) we embedded a bogus personality assessment within an overall Barnum
personality description (e.g., Forer, 1949). Subjects who in one experiment were
led to believe that they had “closed-minded” personalities, subsequently gener-
ated a more one-sided profile of thoughts than subjects who were led to believe
that they were “open-minded.” In another study (Wells
&
Petty, 1980), we
attempted to bias thought production by instructing subjects to make vertical or
horizontal head movements while they processed a persuasive message. Whether
the message
was
pro or counterattitudinal, subjects who engaged in vertical (yes)
movements agreed with the message more than subjects who engaged in horizon-
tal head movements.
In
other studies we have found that excessive message
repetition (Cacioppo
&
Petty, 1979b) and the presence
of
hecklers (Petty
&
Brock, 1976) led to reduced agreement even though the arguments presented
were strong.
ELABORATION LIKELIHOOD
MODEL
OF
PERSUASION
175
The research on each of these independent variables (e.g., heckling, exces-
sive repetition) clearly indicates that they are unlikely to enhance objective
information processing. However, although there is some correlational evidence
(e.g., from thought listings) to suggest that some of these treatments bias the
nature of information processing, the studies may also be interpreted as indicat-
ing that the treatments serve as simple acceptance or rejection cues. For example,
horizontal head movements, hecklers, or excessive repetition may not facilitate
the production of counterarguments, but may instead induce negative affect that
becomes associated with the advocacy (see Zajonc
&
Markus,
1982).
Alter-
natively, the negative affect may bias information processing by increasing ac-
cess to other information linked to negative states (Bower,
1981;
Clark
&
Isen,
1982).
Research that includes these treatments along with message quality
and
other motivational and ability variables should allow more definitive distinction
of these possibilities (see also Section IX,B).
VIII.
Postulate
7:
Consequences
of
Elaboration
In the preceding sections of this article we have outlined how the
ELM
accounts for the initial attitude changes induced by persuasive messages. Postu-
late
7
deals with the different consequences of attitude changes induced via the
central and the peripheral routes. Specifically:
Attitude changes that result mostly from processing issue-relevant arguments (central
route) will show greater temporal persistence, greater prediction of behavior, and greater
resistance to counterpersuasion than attitude changes that result mostly from peripheral
cues.
There are several reasons why these differential consequences would be
expected. Recall that under the central route, attitude changes are based on a
thoughtful consideration of issue-relevant information and an integration of that
information into an overall position. Under the peripheral route, however, an
attitude
is
based on a simple cue that provides some affective association or
allows some relatively simple inference as to the acceptability of the message.
Thus, attitude changes induced via the central route involve considerably more
cognitive work than attitude changes induced under the peripheral route. The
process of elaborating issue-relevant arguments involves accessing the schema
for the attitude object in order to evaluate each new argument (e.g., by compar-
ing it
to
information previously stored in memory). Under the peripheral route,
however, the schema may be accessed only once to incorporate the affect or
inference elicited by a salient cue. Or, a peripheral schema unrelated to the issue
176
RICHARD
E.
PETIY
AND
JOHN
T.
CACIOPPO
schema may be invoked in order to evaluate the cue (i.e., is the source credi-
ble?). Under the central route, then, the issue-relevant attitude schema may be
accessed, rehearsed, and manipulated more times strengthening the interconnec-
tions among
the
components and rendering the schema more internally con-
sistent, accessible, enduring, and resistant than under the peripheral route (cf.
Crocker, Fiske,
&
Taylor,
1984;
McGuire,
1981).
lo
The greater the accessibility of the information supporting an attitude, the
greater the likelihood that the same attitude will be reported over time if people
consider their prior knowledge before reporting their attitudes. Even if people do
not scan their store of attitude-relevant information before reporting their at-
titudes in some circumstances (Lingle
&
Ostrom,
1981),
the greater accessibility
and endurance of the attitude itself would enhance the likelihood that the same
attitude would be reported at two points in time.
Also,
the greater the ac-
cessibility of the information supporting an attitude and the more well organized
it is, the greater the likelihood that this attitude-relevant knowledge can be used
to defend the attitude from subsequent attack. Finally, the greater the ac-
cessibility of the attitude itself, the greater the likelihood that it can guide behav-
ior (Fazio, Chen, McDonel,
&
Sherman,
1982).
In
sum, the greater memory trace and accessibility
of
attitudes and attitude-
relevant information for influence occumng via the central versus the peripheral
route renders people more
able
to report the same attitude over time, to defend
their beliefs, and to act
on
them. A motivational factor may also be relevant,
however. Specifically, the process of scrutinizing issue-relevant arguments may
generally be more deliberate than the processes
of
affective association and the
invocation of well-rehearsed (even automatic) decision rules (Cialdini,
1984).
Thus, changes induced under the central route may be accompanied by a subjec-
tive perception that considerable thought accompanied opinion formation. This
perception may induce more confidence in the attitude, and attitudes held with
more confidence may be more likely to be reported over time, slower to be
abandoned in the face of counterpropaganda, and more likely to be acted upon.
A.
PERSISTENCE
OF
PERSUASION
If extended issue-relevant thinking increases the temporal persistence of
opinion change, then conditions that foster issue-relevant elaboration should be
accompanied by greater attitudinal persistence than conditions that minimize
elaboration. Among all the ways to change attitudes,
role
playing
may be the
influence paradigm that requires the most issue-relevant thinking in order to
Wf
course,
if
a peripheral cue
is
repeatedly
associated with
an
attitude object, relative per-
sistence
of
influence may result (e.g., Weber,
1972).
ELABORATION LIKELIHOOD MODEL
OF
PERSUASION
177
produce persuasion. In role playing research people are required to generate or
improvise their own arguments for a message (e.g., King
&
Janis,
1956).
Impor-
tantly, research indicates that to the extent that people have sufficient knowledge
and skill to generate their own messages, the attitude changes induced by these
messages
are
especially persistent (e.g., Elms,
1966;
Watts,
1967).
In role playing studies subjects are instructed to think about
an
issue posi-
tion, whereas in other research the experimental conditions elicit issue-relevant
thought spontaneously. For example, in research on anticipatory attitude shifts,
subjects are induced to expect to discuss an issue or receive a message on some
topic, and attitudes
are
measured prior to the discussion or message presentation
(Cialdini
&
Petty,
1981).
In one anticipatory change study relevant to persistence
(Cialdini
et
a!.
,
1976),
we led college students to believe that they would discuss
a campus issue with another student who held a position opposite to their own.
Subjects were told that the discussion would take place either immediately or
1
week later, and the issue to be discussed was either one that was personally
important
to
the students or unimportant. While waiting for the discussion to
begin, subjects listed their thoughts on the issue and then reported their attitudes.
Although subjects in all conditions showed some anticipatory shifting of their
positions, only one group of subjects maintained their new issue positions after
they had been informed that the discussion was cancelled. This group, subjects
who expected to immediately discuss a personally important issue, were presum-
ably the most motivated to undertake the cognitive work necessary to prepare for
the discussion. Consistent with this analysis, these subjects listed significantly
more thoughts that supported their own positions in anticipation of the discussion
than subjects in the other cells.
In both the role playing research and the research on anticipatory shifts,
attitude changes that were accompanied by considerable issue-relevant cognitive
activity led to more persisting shifts than changes induced with less issue-rele-
vant thinking. However, in both of these paradigms, no persuasive messages
were presented. According to the
ELM,
the same result should hold
if
the
attitude changes resulted from exposure to a persuasive communication. Specifi-
cally, the greater the elaboration of the message arguments, the more persistent
the resulting attitude change should be. In a direct empirical test of this hypoth-
esis (Petty, Cacioppo,
&
Heesacker,
1985),
we had students listen to three
persuasive messages. Each message began with a description of the origin
of
the
message and provided a brief biography of the message source. The first two
messages served as filler material and were identical for all subjects. The third
message contained the experimental manipulations. Half of the subjects were led
to believe that the advocacy concerned an imminent change in policy at their own
university (high relevance) and half were led to believe that the advocacy con-
cerned a proposed change in policy at a distant university (low relevance). Half
of the students received a message from a very prestigious and credible source;
178
RICHARD
E.
PETTY
AND
JOHN
T.
CACIOPPO
this message contained six strong arguments in support of senior comprehensive
exams. The other half of the students received a message from a low prestige,
inexpert source; this message contained six weak arguments in support
of
senior
comprehensive exams.
Based on our previous research (Petty
&
Cacioppo, 1979b; Petty, Caciop-
po,
&
Goldman,
1981;
see Section VI,A), and pilot testing of the sources and
messages under high and low relevance conditions, we expected both the high
and low involvement groups who received the positive source-strong arguments
message to show equivalent amounts of initial persuasion. However, the change
in the high relevance group should be based mostly on a careful evaluation and
elaboration of the strong issue-relevant arguments, whereas the change in the low
relevance group should be based mostly on the positive source cue. Similarly,
the rejection of the advocacy in the high and low relevance groups exposed to the
negative source-weak arguments message should
be
equivalent initially, but in
the high involvement group the rejection should be based mostly on scrutiny of
the weak arguments, whereas in the low relevance group the rejection should be
based mostly on the negative source.
An
analysis of subjects’ immediate postmessage attitudes concerning senior
comprehensive exams provided support. Both high and low relevance groups of
subjects exposed
to
the strong message/source were more favorable than con-
trols, and both groups of subjects exposed to the weak message/source were less
favorable than controls. More interestingly, however, the degree of personal
relevance had an impact on whether or not these initial attitudes persisted. From
10
to
14
days following message exposure, subjects were called by phone and
were asked their opinions concerning a number of campus issues including the
general idea of senior comprehensive exams. An analysis of variance (ANOVA)
on the initial and delayed attitudes of high relevance subjects revealed only a
main effect for type of communication. The positive source-strong arguments
message was more effective than the negative source-weak arguments message
both initially and at the delayed testing. An analysis on the attitudes
of
low
relevance subjects, however, revealed a communication
X
time
of
measurement
interaction. For these subjects, the initial difference between
the
two message
conditions was
no
longer appparent at the delayed testing. In short, those sub-
jects who formed their initial attitudes based on a careful consideration of issue-
relevant arguments (high relevance) showed greater persistence
of
attitude
change than those subjects whose initial attitudes were based primarily on the
source cue (low relevance).
Other persuasion studies also support
the
view that conditions that foster
people’s ability or motivation to engage in issue-relevant cognitive activity en-
hance the persistence of persuasion.
Thus,
using more interesting or involving
issues about which subjects have more knowledge (e.g., Ronis, Baumgardner,
Leippe, Cacioppo,
&
Greenwald, 1977), providing more time to think about the
ELABORATION LIKELIHOOD MODEL
OF
PERSUASION
179
message (Mitnick
&
McGinnies, 1958), leading people to believe that they will
be
interviewed on the attitude issue (Chaiken, 1980), increasing message repeti-
tion (Johnson
&
Watkins, 1971), and reducing distraction (Watts
&
Holt, 1979)
have all been associated with increased temporal persistence of attitude change
(see
Cook
&
Flay, 1978; Petty, 1977; for reviews, and Petty
&
Cacioppo, 1986,
for a discussion of how the ELM differs from alternative models of attitude
persistence, such as Kelman, 1961).
B. ATTITUDE-BEHAVIOR LINK
The previous section provided support for the view that attitude changes
based primarily on thoughtful consideration (or self-generation) of issue-relevant
arguments produced more enduring persuasion than changes based primarily on
simple cues in the persuasion context. Research is also consistent with the view
that attitude changes induced via the central route
are
more predictive of behavior
than changes induced via the peripheral route.
As
we noted earlier, perhaps the most effortful form of processing occurs
when attitude change results from the self-generation of arguments. These
changes, then, should be especially predictive of behavior. In a relevant program
of research, Fazio and Zanna (1981) explored
the
consequences of attitudes
formed via direct rather than indirect experience. When an attitude is formed via
direct personal experience, the attitude is necessarily based on self-generated
information. When an attitude is based on indirect experience (i.e., a message
from others), less effortful processing may be involved. In some sense then, the
distinction between direct and indirect experience is analogous to the distinction
between attitudes based on role-playing (i.e., self-generation of arguments) ver-
sus passive exposure. Importantly, the research on direct versus indirect experi-
ence clearly indicates that the former attitudes are better predictors of behavior
than the latter (see Fazio, 1985). The ELM suggests that one reason for this is
that attitude formation based on direct experience may typically require more
effortful elaboration of the merits of the object (e.g., puzzle; Regan
&
Fazio,
1977) than attitude formation based on passive exposure.
In our own research, we have also found that conditions that foster a high
elaboration likelihood produce higher attitude-behavior correlations than condi-
tions in which the elaboration likelihood is low. For example, in one study
(described previously in Section VI,A,l), we exposed subjects to mock maga-
zine advertisements for a disposable razor under conditions of either high or low
personal relevance (Petty, Cacioppo,
&
Schumann, 1983). The ads that subjects
saw contained either strong or weak arguments for the product and featured
either a famous (likable) or an ordinary endorser. In addition to assessing product
attitudes in this study, we also asked subjects to rate how likely they were to
180
RICHARD
E.
PETTY
AND
JOHN
T.
CACIOPPO
purchase the product (behavioral intentions). Under high relevance, both at-
titudes and intentions were affected significantly by the manipulation of argu-
ment quality. Under low relevance, however, attitudes were affected by the
manipulation of endorser attractiveness, but behavioral intentions were not. The
peripheral cue of endorser attractiveness was sufficient to enhance liking for the
product when motivation to scrutinize product arguments was low, but was not
sufficient to produce a change in behavioral intentions. Overall, the attitude-
intention correlation under high relevance was
.59,
whereas under low relevance
it was
.36.
Just as increasing motivation to process issue-relevant arguments should
enhance the utility
of
attitudes in predicting behavior,
so
too should enhancing
ability to process the message. In a relevant study, Schumann, Petty, and
Cacioppo
(
1985)
exposed subjects to advertisements containing strong argu-
ments for a new pen either one, four, or eight times in the context of a simulated
television program. Each repetition of the message, of course, gives subjects an
additional opportunity to consider the product-relevant information. After mes-
sage exposure, subjects rated their attitudes toward the advertised pen, their
likelihood of purchasing this brand in the near future, and the amount
of
time
they spent thinking about the product during the program. Subjects reported
engaging in more thought about the product as repetition increased, and the
attitude-intention correlation also improved significantly with repetition.
Finally, we have obtained evidence that people who differ dispositionally in
their tendency to engage in and enjoy thinking also differ in the extent to which
their attitudes predict behavior (Cacioppo, Petty, Kao,
&
Rodriguez,
1985).
Specifically, we found that
the
attitudes toward the candidates in the
1984
presidential election predicted voting intentions and reported behavior better for
people who were high rather than low in their “need for cognition” (Cacioppo
&
Petty,
1982;
see Section V,E). In sum, when the experimental conditions or
dispositional factors enhanced peoples’ motivation or ability to elaborate issue-
relevant information, attitude-behavior correlations were higher than when elab-
oration was
low
(Cialdini
etal.,
1981;
see
also Pallak, Murroni,
,&
Koch,
1983;
Sandelands
&
Larson,
1985;
Sivacek
&
Crano,
1982).”
“Our
argument that the more issue relevant elaboration involved in attitude change the greater
the attitude-behavior correlation should
be,
may appear to conflict with a claim by Wilson, Dunn,
Bybee, Hyman, and Rotondo
(1984)
that analyzing reasons for one’s attitudes
reduces
attitude-
behavior consistency. However, in the research supporting the Wilson
et
al.
contention, one effect
of
having subjects think about the reasons behind their attitudes was to produce a change in attitudes.
Thus, Wilson
et
al.
compared the ability of an
init(a1
attitude to predict behavior with the ability of a
changed
attitude. The new attitude was less predictive than the old one. Importantly, the
ELM
addresses a comparison between two initial attitudes (one formed via the central and one formed via
the peripheral route)
or
two newly changed attitudes (one changed via the central and one changed via
the peripheral route). The
ELM
predicts that the central attitudes will predict behavior better than
comparable attitudes formed
or
changed via the peripheral route.
ELABORATION
LIKELIHOOD
MODEL OF PERSUASION
181
C. RESISTANCE
TO
COUNTERPERSUASION
The final consequence of the route to persuasion is that attitudes formed via
the central route should
be
more resistant to counterpropaganda than attitudes
formed via the peripheral route. Importantly, the
resistance
of an attitude to
attack is conceptually distinct from the temporal
persistence
of
an
attitude. Thus,
some attitudes may
be
highly persistent, but only if they are not attacked. Other
attitudes may be very transient even
in
a vacuum. Likewise, it is possible for
some attitudes to be very resistant
to
change, but only in the short term. Despite
the conceptual independence of persistence and resistance, we have already
outlined the reasons why the ELM holds that usually these two qualities will go
together. Attitudes based on extensive issue-relevant thinking will tend to be
both persistent and resistant, whereas attitudes based
on
peripheral cues will tend
to
be
transient and susceptible
to
counterpersuasion.
Attitudes for which persistence and resistance do
not
go together provide
an
intriguing target of study. Perhaps the most dramatic example of the possible
independence of persistence and resistance is found in cultural truisms. Truisms
such as “you should brush your teeth after every meal” tend to be highly
persistent in a vacuum, but very susceptible to influence if attacked. As McGuire
(1964)
noted, people have very little practice in defending these beliefs because
they have never been challenged. Furthermore, the ELM would contend that
these beliefs
are
highly susceptible to persuasion because they were probably
formed
with very little issue-relevant thinking. It is likely that people come to
accept many cultural truisms sometime during childhood. The truisms
are
con-
tinually presented by powerful, likable, and expert sources (e.g., parents, teach-
ers, television characters) with little or
no
justification. The continual pairing of
the belief with a positive cue results in a relatively persistent attitude, but one that
cannot be defended when subsequently attacked.
Most research
on
attitudinal resistance has focused
on
how various treat-
ments can help bolster an attitude that a person already has. For example, in an
important program of research, McGuire
(1964)
has provided impressive evi-
dence for the view that attitudes can be made more resistant by providing people
with the requisite motivation and/or ability to counterargue opposing messages.
The underlying logic of McGuire’s inoculation theory is that a threat to a pre-
viously wassailed belief motivates the person to defend that belief when it is
attacked in the future.
An
initial attack
on
a person’s belief also provides practice
in defending the belief.
In
another relevant program of research, Burgoon and his
colleagues (e.g., Burgoon, Cohen, Miller,
&
Montgomery,
1978)
investigated
how the manner in which an initial message is processed can affect susceptibility
to a subsequent message on the same topic (see review by Smith,
1982).
The work of McGuire and Burgoon has focused
on
how an initial belief held
by a person can be made more resistant or susceptible by providing some treat-
182
RICHARD
E.
PE3TY
AND
JOHN
T.
CACIOPPO
ment that enhances or reduces the person’s motivation and/or ability to coun-
terargue a subsequent opposing communication. This work is consistent with the
ELM in that it demonstrates that attitudes can be made more resistant by motivat-
ing or enabling people to engage in additional thought about the reasons
or
arguments supporting their attitudes.
To
date, however,
no
research has explicit-
ly tested the ELM prediction that the manner
in
which an attitude is formed
or
changed has important implications for the resistance of the attitude. Specifical-
ly, the ELM predicts that people who come to accept an issue position because of
a peripheral cue (e.g., source expertise) should
be
more susceptible to an attack-
ing message than people who adopt the same issue position based
on
a careful
scrutiny and elaboration
of
the message arguments.
IX.
Complicating
Factors
We have now presented the major postulates of the Elaboration Likelihood
Model and some research relevant to these postulates. In reviewing the evidence
for the ELM we have focused deliberately
on
variables and instances that were
straightforward and relatively unambiguous in interpretation. Although it would
be
nice if we could provide an exhaustive list
of
variables that serve as peripheral
cues and variables that affect message processing in either an objective or a
biased manner, we have already seen that this is not possible. For example, we
have argued that the effects
on
information processing of some variables may
shift from relatively objective to relatively biased as the variable reaches very
high levels. For example, although increasing personal relevance and message
repetition may generally enhance subjects’ motivation and/or ability to see the
merits
of
strong arguments and the flaws in weak ones, we have suggested that
when personal relevance or message repetition reaches very high levels, the
initially objective processing may become biased as the person becomes moti-
vated to reject the advocacy (Cacioppo
&
Petty, 1979b; Petty
&
Cacioppo,
1979b). In short, some variables have multiple effects
on
information process-
ing.
In
addition, we have seen that some variables may affect information pro-
cessing under certain conditions, but serve as peripheral cues in other contexts.
For example, we reviewed evidence in Section
VI,A
that manipulation of the
number of arguments in a message could serve as a peripheral cue when the
personal relevance of the message was low, but that increasing the number of
arguments in a message could increase the amount of information processing
activity when the personal relevance
of
the message was high (Petty
&
Cacioppo,
1984a). In this section we will comment briefly
on
these and other intricacies of
the ELM.
ELABORATION LIKELIHOOD
MODEL
OF
PERSUASION
183
A. VARIABLES
WITH
MULTIPLE EFFECTS
ON ELABORATION
In
most of the research that we have discussed
so
far, we have examined the
isolated effects of different source, message, recipient, and channel factors
on
information processing. However, in most natural persuasion situations, many
variables combine to create the overall persuasion context. For example, consid-
er a high need for cognition person who is part of a jury whose members share
responsibility for evaluating an expert witness who presents weak arguments
in
a
corporate tax case in a courtroom with noisy distractions. All of the many
variables present in this situation must be considered jointly to determine the
probable persuasive impact of the testimony. Normally, sharing cognitive re-
sponsibility with a group reduces information processing activity (Petty, Caciop-
po,
&
Harkins, 1983), but our message recipient dispositionally tends to like to
think (Cacioppo, Petty,
&
Moms,
1983) and is therefore less susceptible to
motivation loss in groups (Petty, Cacioppo,
&
Kasmer, 1985). Therefore,
moti-
vation
to process the message is likely to be high despite the group responsibility.
However, due to a lack
of
prior knowledge about corporate taxes and the distrac-
tions inherent in the situation, our message recipient may have little
ability
to
process the weak message arguments (Petty, Wells,
&
Brock, 1976; Wood,
Kallgren,
&
Priesler, 1985). Thus, the perceived expertise of the witness may
serve as a potent influence cue (Kiesler
&
Mathog, 1968).
Our example assumes that each
of
the features of the persuasion situation
(e.g., distraction, group responsibility) can be considered separately and inde-
pendently regardless of the levels of the other variables with which it
is
com-
bined. If
so,
one can roughly add (subtract) the effects of each variable to
determine the overall elaboration likelihood. Although this is often possible, as
we discuss next,
it
is also possible for one variable to have very different effects
on
information processing depending
on
the level of other variables.
For example, some variables may increase information processing at one
level of another factor, but may actually decrease processing at a different level
of
that factor.
In
one study exploring this possibility, we varied the personal
relevance of a message, whether concluding summaries of the message argu-
ments were framed as statements or as rhetorical questions (cf. Zillmann, 1972),
and whether the arguments presented were
strong
or weak.
In
this study (Petty,
Cacioppo,
&
Heesacker, 1981), all subjects heard over headphones a message
advocating that seniors take a comprehensive exam in their major as a require-
ment for graduation. The study was designed to test our view that summarizing
arguments as rhetorical questions (e.g., “Wouldn’t instituting a comprehensive
exam be an aid to those who seek admission to graduate and professional
schools?”) rather than as declarative statements, would motivate more thinking
about
the
arguments. If rhetoricals enhance relatively objective processing, then
184
RICHARD
E.
PElTY
AND
JOHN
T. CACIOPPO
their use should lead to more agreement if the message arguments are strong,
but
less agreement if the arguments are weak. However, this enhanced elaboration
with rhetoricals should be evident mostly when people are not naturally devoting
much effort to processing the message arguments, such as when the personal
II)
Low
Relevance
Strong
-.4
:j
<
--0
Weak
Stalemen1 Rhetorical
Message
Form
Ambiguous Relevance
Strong
60
0
Weak
LOW High
Source Attractiveness
151
\
I
9
Weak
One Three
Number
of
Sources
High
Relevance
?I
Statement Rhetorical
Message Form
4
Ambiguous Relevance
1
/Strong
LOW High
Source Expertise
6
Ambiguous
Relevance
Strong
:i/
Standing Reclining
Body
Position
Fig. 8. Additional variables that may affect information processing in a relatively objective
manner. (1) Effects of rhetorical questions
on
attitudes following strong and weak messages of low
relevance (data from Petty, Cacioppo,
&
Heesacker, 1981).
(2)
Effects of rhetorical questions
on
attitudes following strong and weak messages of high relevance (data from Petty, Cacioppo,
&
Heesacker. 1981).
(3)
Effects
of
social attractiveness
on
attitudes following
strong
and weak mes-
sages of uncertain relevance (data from Puckett
er
al.,
1983).
(4)
Effects of expertise
on
attitudes
of
field-dependent subjects following strong and weak messages of uncertain relevance (data from
Heesacker
er
al.,
1983).
(5)
Effects of multiple sources
on
attitudes following strong and weak
messages of uncertain relevance (data from Harkins
&
Petty, 1981a; Experiment
4).
(6)
Effects
of
recipient posture
on
attitudes following
strong
and weak messages
of
uncertain relevance (data from
Petty, Wells, Heesacker, Brock,
&
Cacioppo, 1983).
185
ELABORATION LIKELIHOOD MODEL
OF
PERSUASION
relevance of the message is low. When people are already naturally motivated to
process the message, such as when personal relevance is high, we expected that
the use of rhetoricals would either have no further effect on elaboration, or might
even
be
disruptive of ongoing information processing. The results of our study
supported these expectations. When personal relevance was low, the use
of
rhetorical questions increased elaboration (see Box
1,
Fig. 8), but when personal
relevance was high, subjects reported that the rhetorical questions were distract-
ing and argument scrutiny was reduced (see Box
2,
Fig. 8).
Burnkrant and Howard (1984) replicated our rhetoricals study making a few
key changes. Again, subjects were presented with the strong or weak version of
our senior comprehensive exam message that was made either high or low in
personal relevance. Instead of hearing the message over headphones, however,
subjects were presented with the message as a written communication. In addi-
tion, instead of summarizing each message argument as a rhetorical question
after the argument was presented, all rhetorical questions
preceded
the presenta-
tion of the message arguments. Bumkrant and Howard argued that these changes
should eliminate the distraction effect that we observed under high involvement.
First, presenting the message in print rather than orally gives subjects time to
stop the message to consider arguments fully (see Section V1,B). Thus, the
rhetoricals need not disrupt processing even if subjects are highly involved.
Second, placing the rhetorical questions at the beginning rather than at the end of
the arguments has the advantage of generating interest
and
curiosity while avoid-
ing the disadvantage of interrupting the train
of
thought concerning the argument
just presented. Their results were consistent with this reasoning. The use
of
introductory rhetorical questions in print enhanced argument elaboration re-
gardless of the personal relevance of the issue. When paired with the strong
arguments, rhetorical questions increased agreement, but when paired with the
weak arguments, rhetorical questions decreased agreement. The studies on rhet-
orical questions provide cogent examples of how an independent variable can
have different but predictable effects on elaboration depending on the level of
other variables, such
as
personal
relevance and modality of message presentation.
A second way in which the impact
of
one variable may depend on the level
of another factor is in whether the variable induces relatively objective or rela-
tively biased information processing. For example, we have demonstrated that
increasing the degree of personal relevance
of
a message can enhance a person's
motivation to process the message in a relatively objective manner (Petty
&
Cacioppo, 1979b). However, we have also seen that this processing may become
biased if personal relevance is combined with a threat, such as that induced by a
forewarning
of
persuasive intent (Petty
&
Cacioppo, 1979a). In short, personal
relevance per se may motivate increased processing, but when combined with
some threat, the processing may be directed in the defense of one's initial
position.
186
RICHARD
E.
PETTY
AND
JOHN
T.
CACIOPPO
B. VARIABLES THAT AFFECT ELABORATION AND SERVE AS
CUES
In Section
=,A
we argued that whether or not a particular variable en-
hances or diminishes processing, or motivates relatively objective or relatively
biased processing, may depend on the level of other variables in the persuasion
context. Similarly, whether a variable affects information processing or serves as
a peripheral cue may depend on the level
of
other elements in the persuasion
situation. We discuss this feature of the
ELM
below.
1.
Source
ExpertiseIAttractiveness
In Section
I,
we noted that one aspect of persuasion research that has
disappointed reviewers of the field is that even variables that were expected to be
quite simple in their effects on attitude change have instead proved to be quite
complex. We also noted that perhaps the
most
dramatic example of this was the
conflicting results of research on features of the message source (Eagly
&
Him-
melfarb,
1974).
Postulate
3
(Section IV) of the
ELM
outlines the several differ-
ent ways in which source (and other) factors can affect persuasion: they can serve
as arguments, they can serve as cues,
or
they can affect argument processing. In
the research that we have reviewed
so
far, we have focused on how source
factors operate when the elaboration likelihood is either very high or very low.
We have seen that when people are unmotivated and/or unable to process a
message, they rely on simple cues in the persuasion context, such as the expertise
or attractiveness of the message source, although other cues may be used if they
are more salient. Importantly, since subjects are either unmotivated or unable to
evaluate message arguments, a positive source tends to enhance persuasion and a
negative source tends to reduce persuasion, regardless of message quality (e.g
.
,
see top panels in Fig.
4).
On the other hand, when people are highly motivated and able to process
message arguments, strong arguments are more effective than weak ones despite
the presence
of
peripheral cues such as source credibility and attractiveness (e.g.,
see bottom panels in Fig.
4).
When motivation and ability to process are high,
people are concerned with evaluating the true merits of the advocacy. In order to
do this, they will scrutinize all available and inferred information in the immedi-
ate persuasion context, and attempt to relate it to information stored previously in
memory. Interestingly, a consideration of source factors may be part of a per-
son’s attempt to evaluate issue-relevant information when the elaboration like-
lihood
is high. For example, under some circumstances a source feature may
itself serve as a persuasive argument by providing information central to the
merits of the attitude object (e.g., a physically attractive source, without saying
anything, may provide persuasive visual testimony as to the effectiveness of a
ELABORATION LIKELIHOOD MODEL
OF
PERSUASION
187
beauty product; Kahle
&
Homer, 1985; Petty
&
Cacioppo, 1981b). Additionally,
a consideration
of
source information might help a person in evaluating the true
merits of any given argument (e.g., is the expert source biased or does the source
have a vested interest?).
It may now appear that the cases
of
high and low elaboration likelihood are
quite clear. Source factors affect agreement with a message by serving as simple
acceptance or rejection cues when the elaboration likelihood is low, but do not
serve as simple cues when the elaboration likelihood is high. Instead, they are
considered along with the message arguments in an attempt to evaluate the true
merits of the advocacy. These conclusions, however, are only part of the story of
how source factors impact on persuasion.
As
we noted in Section III,A, we view
elaboration likelihood as a continuum anchored at one end by the peripheral route
to persuasion, and at the other end by the central route. In all of our research
described in the previous sections, we have attempted to create and describe
relatively clear instances of central and peripheral routes to persuasion. Thus, for
example, in
our
research on motivation to process, subjects were either highly
involved with the topic (e.g., students were confronted with an advocacy that
had implications for their own graduation; Petty, Cacioppo,
&
Goldman, 1981),
or very uninvolved (e.g., the students were certain that there were no personal
implications of the advocacy). The extreme high and low elaboration likelihood
conditions have been quite useful for theory testing purposes and in explicating
the two routes to persuasion. However, these conditions represent only part of
the elaboration likelihood continuum.
Specifically, many day to day persuasion contexts are unlikely to be as high
or as low in elaboration likelihood as the conditions we have deliberately created
in
our
initial research. For example, people are sometimes uncertain as to the
personal relevance
of
an issue, or have moderate rather than very high
or
very
low knowledge on a topic. We have proposed that under more moderate condi-
tions, people use source characteristics to determine how much to think about the
message (Petty
&
Cacioppo, 1981a, 1984~). When the personal consequences or
prior knowledge on an issue
are
moderate or unclear, people may not
be
sure
if
the message is worth thinking about or
if
they are able to do
so.
Under these
circumstances, characteristics
of
the message source can help a person decide if
the message warrants (or needs) careful scrutiny. In our own research on moder-
ate levels of motivation to process, undergraduate students have been led to
believe that a change in policy was being advocated for their university, but they
were not told when or
if
this policy ever would be implemented. Thus, unlike our
studies
on
high and low personal relevance (see Section VJ,A), subjects could
not be certain whether or not the change in policy would affect them.
In one study (Puckett, Petty, Cacioppo,
&
Fisher, 1983), for example, we
told subjects that students
in
an evening undergraduate continuing education
course had written essays on the issue
of
whether comprehensive exams should
188
RICHARD
E.
PETTY AND
JOHN
T.
CACIOPPO
be given in a student’s major area of study as a prerequisite for obtaining a
bachelor’s degree. Each subject was given a folder containing a typed essay
along with a card containing a picture and a brief description of the author
of
the
essay. Two major variables were manipulated in the study:
(1)
the social attrac-
tiveness of the author (socially attractive authors were more physically attractive
and had better family backgrounds and more prestigious hobbies than the socially
unattractive authors), and
(2)
the quality of the arguments in the editorial (either
strong or weak). A third variable, age of the essay author, was also manipulated
but this factor had no effect on persuasion. After looking through the appropriate
folder, subjects were asked to rate their own opinions about the senior compre-
hensive exam issue. The major result indicated that the arguments were more
carefully processed when they were associated with a socially attractive than a
socially unattractive source. More specifically, the significant message quality
X
source attractiveness interaction was due to the joint tendencies for attractiveness
to enhance agreement with the proposal when the arguments presented were
strong, but for attractiveness to reduce agreement when the arguments presented
were weak
(see
Box
3,
in Fig.
8).
The latter effect (an attractive source reducing
agreement), of course, is opposite to what one normally would expect the effect
of attractive sources to be (see review by Chaiken,
1985).
In a study conceptually similar to the Puckett
et
al.
(1983)
study, we again
left the degree of personal relevance ambiguous and manipulated the quality of
the arguments presented in favor of senior comprehensive exams. This time,
however, subjects heard rather than read the message, and we varied source
expertise rather than social attractiveness (Heesacker, Petty,
&
Cacioppo,
1983).
Some subjects were led to believe that the source of the message was a professor
of education at Princeton University (high expertise), and others were led to
believe that the source was a local high school student (low expertise). The
subjects in this study were divided into those who were relatively field dependent
or independent as assessed by the embedded figures test (Ekstrom, French,
&
Harmon,
1962).
The data for field-dependent subjects showed a message quality
X
source expertise interaction (see Box
4,
Fig.
8).
Similar to the effect observed
for social attractiveness, the arguments were more carefully processed when they
were presented by the expert than by the inexpert source. Again, the interaction
was due to the joint tendencies for strong arguments to be more persuasive when
presented by an expert, but for weak arguments to
be
less persuasive when
presented by an expert, and again the latter effect is opposite
to
what one
normally would expect the effect
of
expertise to be.’*
I2Field-independent subjects showed only a main effect for argument quality, probably because
these subjects were generally more motivated and/or able to extract meaning
from
stimuli (Witkin,
Goodenough,
&
Oltman,
1979).
If
field-independent subjects generally have a higher elaboration
likelihood, then they would
be
more likely to process message arguments regardless
of
the source.
ELABORATION
LIKELIHOOD
MODEL
OF
PERSUASION
189
In sum, although the operation of source factors may seem quite simple on
the surface, the
ELM
indicates that their operation, although orderly and predict-
able, is quite complex. In separate experiments, we have seen that when personal
relevance is high, source factors can serve as persuasive arguments or assist in
the evaluation of arguments; when personal relevance is low, source factors can
serve as simple cues; and when personal relevance is moderate or ambiguous,
source factors can affect the extent of message processing. Since all of these
effects are obtained under different conditions, however, it is not surprising that
a great diversity
of
results has been observed in the literature.
2.
We have now seen how some source variables can serve as arguments in
some contexts, cues in other contexts, and affect argument processing in still
other situations. This general principle, which is stated explicitly in Postulate
3
(Section IV), was applied mostly to separate variables as we explicated the
various postulates of the
ELM.
However, it should now be clear that any
one
variable can serve in all
of
these roles. A few more examples should help to
elucidate how one variable can serve in multiple roles depending on the specific
features of the persuasion context.
First, consider the impact of the number of other people who endorse a
particular attitudinal position. Traditional analyses of the number of message
sources have assumed that the more people who are perceived to advocate a
position (up to some limit), the more conformity pressure that is induced, and the
more agreement that results (e.g., Asch, 1951; White, 1975). One popular
explanation for this conformity effect is that people shift toward the majority
view out of a desire to hold a correct opinion (Festinger, 1954). An alternative
point of view is that the more people who
are
associated with a particular
position, the more recipients may think about
the
position advocated (Bumstein
&
Sentis, 1981; Bumstein
&
Vinokur, 1977) or about the specific arguments
presented (Harkins
&
Petty, 1981a,b). This enhanced thinking might lead to
more or less agreement depending on the nature of the thinking. For example,
Harkins
&
Petty (1981a) found that when
the
personal relevance of a proposal
was left ambiguous, increasing the number of sources who presented strong
arguments enhanced persuasion, but that increasing the number of sources who
presented weak arguments reduced persuasion (see Box
5,
Fig.
8).
Other Variables with
Multiple
Functions
Importantly, we do not mean to suggest that expert/attractive sources invariably enhance information
processing when involvement is moderate. Under
some
conditions, for example, it may
be
more
adaptive and/or necessary to engage
in
more scrutiny
of
8
moderate or
low
than a clearly credible
source (Petty
&
Cacioppo, 1981a,
1984~).
190
RICHARD
E.
PETTY
AND
JOHN
T.
CACIOPPO
The initial (conformity) explanation of multiple source effects is consistent
with the view that the mere number of other people advocating a position serves
as a simple peripheral cue
as
to the validity of the advocacy. The second (infor-
mation processing) interpretation, however, is more consistent with the view that
the attitude changes induced by multiple sources follow the central route to
persuasion (Harkins
&
Petty, 1983). The ELM, of course, suggests that both of
these processes may operate in different situations. When the elaboration like-
lihood is very low (such as when personal relevance is low or distraction is high),
people will
be
unmotivated to evaluate the issue-relevant information presented
and may use the number of people who support the issue as a simple cue as to the
worth of the proposal. When the elaboration likelihood is moderate, people may
use the number of
sources
advocating a position
as
an
indication of whether the
message is worth considering. Finally, when the elaboration likelihood
is
very
high, message recipients will undertake a deliberate assessment
of
the message
arguments and the number of endorsers will have little further value as a motivat-
or of thought or as a simple acceptance cue.
No
experiment to date, however, has
examined the impact of the number of sources across the full elaboration like-
lihood continuum.
l3
Factors associated with the message source,
of
course,
are
not the only
variables that can both serve as cues and affect message processing. Message
variables can likewise serve in both roles. We have already discussed how the
number of arguments could serve as a simple cue when personal relevance was
low, but affect information processing when personal relevance was high (Petty
&
Cacioppo, 1984a). Similarly, recipient and context variables may serve
in
multiple roles. For example, we have shown that the physical posture of a
message recipient can affect the extent
of
elaboration under moderate involve-
ment conditions.
In
one study (Petty, Wells, Heesacker, Brock,
&
Cacioppo,
1983), people who were reclining comfortably during message exposure showed
greater attitudinal differentiation of strong from weak message arguments than
people who were standing (see Box
6,
Fig.
8).
If subjects were presented with a
message they were unmotivated or unable to elaborate, however, then posture (or
other factors related to comfort during message exposure) might serve as simple
positive or negative affective cues (e.g., Griffit
&
Veitch, 1971).
Importantly, even though the ELM holds
open
the possibility that variables
can affect agreement either by having an impact
on
information processing or by
serving as simple cues, the ELM specifies, in a general manner at least, the
conditions under which each process is likely to operate. Thus, a whole list of
source (e.g., credibility, attractiveness, number of sources), message (e.g.,
130Ur
analysis of multiple sources assumes that all sources
k
advocating the same position.
When conflicting positions are advocated
by
different numbers
of
people (as in minority influence),
the situation becomes
more
complex (see
Maass
&
Clark,
1984).
ELABORATION LIKELIHOOD
MODEL
OF
PERSUASION
191
number of arguments, use of rhetoricals, discrepancy), audience (e.g., recipient
posture, presence of hecklers, false physiological feedback), and other variables
may affect attitudes by modifying information processing under certain condi-
tions (e.g., ambiguous personal relevance), but affect attitudes by serving as
simple cues in other contexts (e.g., low prior knowledge).
X.
Summary
and Conclusions
At the most general level, we have outlined two basic routes to persuasion.
One route is based on the thoughtful (though sometimes biased) consideration of
arguments central to the issue, whereas the other is based on affective associa-
tions or simple inferences tied to peripheral cues in the persuasion context. When
variables in the persuasion situation render the elaboration likelihood high, the
first kind
of
persuasion occurs (central route). When variables in the persuasion
situation render the elaboration likelihood low, the second kind of persuasion
occurs (peripheral route). Importantly, there
are
different consequences of the
two routes to persuasion. Attitude changes via the central route appear to be more
persistent, resistant, and predictive of behavior than changes induced via the
peripheral route.
In the body of this article we have discussed a wide variety of variables that
proved instrumental in affecting the elaboration likelihood, and thus the route to
persuasion. In fact, one of the basic postulates of the Elaboration Likelihood
Model, that variables may affect persuasion by increasing or decreasing scrutiny
of message arguments, was highly useful
in
accounting for the effects of a
seemingly diverse list of variables (see Figs.
3
and
8).
The effects of these
variables had been explained with many different theoretical accounts in the
accumulated persuasion literature. The
ELM
was successful in tying the effects
of these variables to one underlying process. We have also seen that many
different variables could serve as peripheral cues, affecting persuasion without
issue-relevant thinking. Finally, we saw that some variables were capable of
serving in multiple roles, enhancing or reducing thinking in some contexts, and
serving as simple acceptance or rejection cues in others.
We began this article by noting that reviewers of the attitude change liter-
ature have been disappointed with the many conflicting effects observed, even
for ostensibly simple variables. For example, manipulations of source expertise
have sometimes increased persuasion, sometimes have had no effect, and have
sometimes decreased persuasion. Similarly, studies testing different theories
have sometimes found the theory to be useful in predicting attitude change, and
at other times have found the theory
to
be unpredictive. For example, self-
perception processes appear to operate under some conditions, but not others.
192
RICHARD
E.
PE'ITY
AND
JOHN
T.
CACIOPPO
The Elaboration Likelihood Model attempts to place these many conflicting
results and theories under one conceptual umbrella by specifying the major
processes underlying persuasion and indicating how many of the traditionally
studied variables and theories relate to these basic processes. Thus, we have seen
that
a
seemingly simple variable like source credibility actually is capable of
affecting persuasion in rather complex ways. The ELM, however, elucidates the
conditions under which these different effects are likely to operate. Similarly, we
have seen that a theoretical process such as self-perception, which emphasizes a
simple inference based on behavioral cues, is likely to operate when the elabora-
tion likelihood is relatively low but not when the elaboration likelihood is very
high.
We believe that perhaps the greatest strength of the Elaboration Likelihood
Model is that it specifies the major ways in which variables can have an impact
on persuasion, and it points to the major consequences
of
these different media-
tional processes. In one sense, the ELM
is
rather simple. It indicates that vari-
ables can affect persuasion
in
a limited number of ways:
A
variable can serve as a
persuasive argument, serve as a peripheral cue, or affect argument scrutiny in
either a relatively objective
or
a relatively biased manner. In confining the
mediational processes of persuasion to just these possibilities, the ELM provides
a simplifying and organizing framework that may be applied to many of the
traditionally studied source, message, recipient, and context variables. The pos-
tulates of the ELM do
not
ultimately indicate
why
certain arguments are strong or
weak, why certain variables serve as cues, or why certain variables affect infor-
mation processing. Instead, the ELM limits the mediational processes of persua-
sion to a finite set, and specifies, in a general way at least, the conditions under
which each mediational process is likely to occur and the consequences
of
these
processes.
In
doing this, the ELM may prove useful in providing a guiding set of
postulates from which
to
interpret previous work, and in suggesting new hypoth-
eses to be explored in future research.
ACKNOWLEDGMENTS
Research described in
this
article and preparation of
the
article were supported by grants
BNS
7913753,8217096,8414853,8418038, and 8444909 from the National Science Foundation.
We
are grateful to Icek Ajzen. Robert Cialdini, Chester Insko,
and
Abraham Tesser for their helpful
comments on an earlier draft of this article.
REFERENCES
Abelson, R. (1972). Are attitudes necessary. In
B.
T.
King
&
E. McGinnies (Eds.),
Attirudes,
Ajzen,
I.,
&
Fishbein, M. (1977). Attitude-behavior relations: A theoretical analysis and review of
conflict. and social change.
New
York
Academic
Press.
empirical research.
Psychological Bnllerin,
84,
888-918.
ELABORATION LIKELIHOOD MODEL OF PERSUASION
193
Ajzen,
I.,
&
Fishbein, M.
(1980).
Understanding arritudes and predicting social behavior.
En-
glewood Cliffs, NJ: Prentice-Hall.
Allport,
G.
W.
(1935).
Attitudes.
In
C.
Murchison (Ed.),
Handbook of socialpsychology
(Vol.
2).
Worchester, MA: Clark University Press.
Allyn, J.,
&
Festinger,
L.
(1961).
The effectiveness of unanticipated persuasive communications.
Journal of Abnormal and Social Psychology, 62,
35-40.
Anderson, N.
(1981).
Integration theory applied to cognitive responses and attitudes. In
R.
Petty,
T.
Osuom,
&
T. Brock (Eds.),
Cognitive responses
in
persuasion.
Hillsdale, NJ: Erlbaum.
Andreoli,
V.,
&
Worchel,
S.
(1978).
Effects of media, communicator, and message position on
attitude change.
Public Opinion Quarterly,
42,
59-70.
Apsler,
R.,
&
Sears,
D.
0.
(1968).
Warning, personal involvement, and attitude change.
Journal
of
Personality and Social Psychology,
9,
162- 166.
Asch,
S.
(1948).
The doctrine of suggestion, prestige, and imitation in social psychology.
Psycho-
logical Review,
55,
250-276.
Asch,
S.
(1951).
Effects of group pressure upon the modification and distortion of judgment. In
H.
Guetzkow (Ed.),
Groups, leadership, and men.
Pittsburgh: Carnegie.
Baron,
R.
S.,
Baron,
P.,
&
Miller,
N.
(1973).
The relation between distraction and persuasion.
Psychological Bullerin,
80,
310-323.
Batra, R.,
&
Ray, M.
(1984).
Advertising situations: The implications of differential involvement
and accompanying affective responses.
In
R.
I.
Hanis (Ed.),
Information processing research
in advertising.
Hillsdale, NJ: Erlbaum.
Batra,
R.,
&
Ray, M.
(1985).
How advertising works at contact.
In
L.
Alwitt
&
A. Mitchell (Eds.),
Psychological processes and advertising effects: Theory, research and application.
Hillsdale,
NJ:
Erlbaum.
Belch,
G.
E.
(1982).
The effects of television commercial repetition on cognitive responses and
message acceptance.
Journal of Consumer Research,
9,
56-65.
Bem,
D.
J.
(1967).
Self-perception: An alternative interpretation of cognitive dissonance phe-
nomena.
Psychological Review,
74,
183-200.
Bem,
D.
J.
(1972).
Self-perception theory. In
L.
Berkowitz (Ed).,
Advances
in
experimental social
psychology
(Vol.
6).
New York: Academic
Press..
Bettman, J.
R.,
&
Park, C.
W.
(1980).
Effects
of
prior knowledge and experience and phase of the
choice process
on
consumer decision processes: A protocol analysis.
Journal
of
Consumer
Research,
7,
234-248.
Bobrow,
D.
G.,
&
Norman,
D.
A.
(1975).
Some principles of memory schemata.
In
D.
G. Bobrow
&
A. Collins
(Eds.),
Represenfarion and understanding: Studies
in
cognitive science,
New
York: Academic Press.
Borgida, E.,
&
Howard-Pitney,
B.
(1983).
Personal involvement and the robustness of perceptual
salience effects.
Journal of Personality and Social Psychology,
45,
560-570.
Bower, G. H.
(1981).
Mood and memory.
American Psychologist,
11,
11-13.
Brehm,
J.
W.
(1966).
A theory of psychological reactance.
New York Academic Press.
Brehm, J. W.
(1972).
Responses to
loss
of
freedom: A theory of psychological reactance.
Mor-
ristown,
NJ:
General Learning Press.
Brickner, M. A,, Harkins,
S.
G.,
&
Ostrom,
T.
M.
(1985).
The effects of personal involvement:
Thought provoking implications for social loafing.
Journal of Personality and Social Psychol-
ogy,
in press.
Britton,
B.
K.,
&
Tesser, A.
(1982).
Effects of prior knowledge on use of cognitive capacity in three
complex cognitive tasks.
Journal
of
Verbal Learning and Verbal Behavior,
21,
421-
436.
Brock,
T.
C.
(1967).
Communication discrepancy and intent to persuade as determinants of coun-
terargument production.
Journal of Experimental Social Psychology.
3,
269-309.
194
RICHARD
E.
PETTY AND JOHN
T.
CACIOPPO
Burgoon, M., Cohen, M., Miller, M.,
&
Montgomery, C. (1978). An empirical test of a model of
resistance to persuasion.
Human Communications Research,
5,
27-39.
Burnkrant, R. E.,
&
Howard,
D.
J.
(1984). Effects of the use of introductory rhetorical questions
versus statements on information processing.
Journal
of
Personality and Social Psychology,
47,
Bumkrant,
R.
E.,
&
Sawyer, A. (1983). Effects of involvement
on
information processing intensity.
In
R.
J.
Hanis (Ed.),
Information processing research
in
advertising.
Hillsdale, NJ: Erlbaum.
Burnstein, E.,
&
Sentis, K. (1981). Attitude polarization in groups. In
R.
Petty, T. Ostrom,
&
T.
Brock
(Eds.),
Cognitive responses in persuasion
(pp. 197-216). Hillsdale,
NJ:
Erlbaum.
Burnstein, E.,
&
Vinokur, A., (1977). Persuasive argumentation and social comparison as determi-
nants of attitude polarization.
Journal
of
Experimental Social Psychology,
13,
3 15-332.
Burnstein, E., Vinokur, A.,
&
Trope,
Y.
(1973). Interpersonal comparison versus persuasive argu-
mentation: A more direct test
of
alternative explanations for group induced shifts in individual
choice.
Journa/ of Experimental Social Psychology,
9
236-245.
Cacioppo,
J.
T., Harkins,
S.
G.,
&
Petty,
R.
E. (1981). The nature of attitudes and cognitive
responses and their relationships to behavior. In R. Petty, T. Ostrom,
&
T. Brock
(Eds.),
Cognitive responses in persuasion.
Hillsdale,
NJ:
Erlbaum.
Cacioppo,
J.
T.,
&
Petty,
R.
E.
(1979a). Attitudes and cognitive response: An electrophysiological
approach.
Journal of Personality and Social Psychology,
37,
2181-2199.
Cacioppo,
J.
T.,
&
Petty, R. E. (1979b). Effects of message repetition and position on cognitive
responses, recall, and persuasion.
Journal of Personality and Social Psychology,
37,
97- 109.
Cacioppo,
J.
T.,
&
Petty,
R.
E. (1980a). Persuasiveness
of
communications is affected by exposure
frequency and message quality: A theoretical and empicial analysis of persisting attitude change.
In
J.
H.
Leigh
&
C.
R.
Martin (Eds.),
Current issues and research in advertising.
Ann Arbor:
University of Michigan Graduate School of Business Administration.
Cacioppo,
J.
T.,
&
Petty, R. E. (1980b). Sex differences in influenceability: Toward specifying the
underlying processes.
Personality and Social Psychology Bulletin,
6,
65 1-656.
Cacioppo,
J.
T.,
&
Petty, R. E. (1981a). Electromyograms
as
measures of extent and affectivity of
information processing.
American Psychologist.
36,
441-456.
Cacioppo,
J.
T.,
&
Petty,
R.
E.
(1981b). Electromyographic specificity during covert information
processing.
Psychophysiology.
18,
518-523.
Cacioppo,
J.
T.,
&
Petty,
R.
E. (1981~). Social psychological procedures for cognitive response
assessment: The thought listing technique. In T. Merluzzi, C. Glass,
&
M.
Genest
(Eds.),
Cognitive assessment.
New
York
Guilford.
Cacioppo,
J.
T.,
&
Petty, R. E. (1982). The need for cognition.
Journal of Personality and Social
Cacioppo,
J.
T.,
&Petty, R. E. (1984a). The Elaboration Likelihood Model.
Advances in Consumer
Research,
11,
673-675.
Cacioppo,
J.
T.,
&
Petty, R. E. (1984b). The need for cognition: Relationship to attitudinal
pro-
cesses. In
R.
McGlynn,
J.
Maddux, C. Stoltenberg,
&
J.
Harvey
(Eds.),
Social perception
in
clinical and counseling psychology.
Lubbock: Texas Tech University Press.
Cacioppo.
J.
T.,
&
Petty,
R.
E. (1985). Central and peripheral routes to persuasion: The role of
message repetition.
In
A. Mitchell
&
L. Alwitt
(Eds.),
Psychological processes
and
advertising
eflects.
Hillsdale,
NJ
Erlbaum.
Cacioppo,
J.
T.,
&
Petty,
R.
E.
(1986). Staiking rudimentary processes of social influence: A
psychophysiological approach. In M.
P.
Zanna,,J. M. Olson,
&
C. P. Herman
(Eds.).
Social
influence: The Ontario symposium
(Vol.
5).
Hillsdale,
NJ
Erlbaum, in press.
Cacioppo,
J.
T., Petty, R.
E.,
&
Kao, C. (1984). The efficient assessment
of
need for cognition.
Journal of Personality Assessment,
48,
306-307.
1218- 1230.
Psychology, 42,
1 16- 1
3
1.
ELABORATION LIKELIHOOD MODEL
OF
PERSUASION
195
Cacioppo, J. T., Petty, R. E., Kao, C.,
&
Rodriguez, R.
(1985).
Central andperipheral routes
to
persuasion:
An
individual difference perspective.
Unpublished manuscript, University of Iowa,
Iowa City.
Cacioppo, J. T., Petty,
R.
E.,
&
Moms,
K.
(1983).
Effects of need for cognition on message
evaluation, recall, and persuasion.
Journal of Personaliry and Social Psychology,
45,
805-8 18.
Cacioppo,
J.
T.,
Petty, R.
E.,
&
Sidera,
J.
(1982).
The effects of a salient self-schema on the
evaluation of proattitudinal editorials: Top-down versus bottom-up message processing.
Journal
of Experimental Social Psychology,
18,
324-338.
Cacioppo,
J.
T., Petty,
R.
E.,
&
Stoltenberg, C.
(1985).
Processes
of
social influence:
The
elabora-
tion likelihood model of persuasion. In P. Kendall
(Ed.),
Advances in cognitive behavioral
research and therapy
(Vo1.4).
New York: Academic Press.
Calder, B.
J.,
Insko, C.,
&
Yandell,
B.
(1974).
The relation
of
cognitive and memorial processes to
persuasion in a simulated jury trial.
Journal of Applied Social Psychology,
4,
62-93.
Calder,
B.
J.,
&
Stemthal,
B.
(1980).
Television commercial wearout: An information processing
view.
Journal of Marketing Research,
17,
173-186.
Cantor,
G.
N.
(1968).
Children’s “like-dislike” ratings of familiarized and nonfamiliarized visual
stimuli.
Journal
of
Experimental Child Psychology,
6,
65 1-657.
Cantor,
N.,
&
Mischel,
W.
(1979).
Prototypes in person perception. In
L.
Berkowitz
(Ed.),
Ad-
vances
in
experimental social psychology
(Vol. 12).
New York: Academic Press.
Chaiken,
S.
(1980).
Heuristic versus systematic information processing and the use of source versus
message cues in persuasion.
Journal
of
Personality and Social Psychology,
39,
752-756.
Chaiken,
S.
(1985).
Physical appearance and social influence. In C.
P.
Herman, M. P. Zanna,
&
E.
T. Higgins
(Eds.),
Physical appearance-, stigma and social behavior: The Ontario symposium
(Vol. 4).
Hillsdale, NJ: Erlbaum, in press.
Chaiken,
S.,
&
Baldwin,
M.
W.
(1981).
Affective-cognitive consistency and the effect of salient
behavioral information on the self-perception of attitudes.
Journal of Personality and Social
Chaiken,
S.,
&
Eagly, A. H.
(1976).
Communication modality
as
a determinant of message per-
suasiveness and message comprehensibility.
Journal
of
Personality and Social Psychology,
34,
Chaiken,
S.,
&
Eagly, A. H.
(1983).
Communication modality as a determinant of persuasion: The
role of communicator salience.
Journal of Personality and Social Psychology,
45,
241-256.
Cialdini, R.
B.
(1984).
Principles of automatic influence. In J. Jacoby
&
C.
S.
Craig
(Eds.),
Personal selling: Theory, research, and practice.
Lexington, MA: Heath.
Cialdini, R. B., Levy, A., Herman, P.,
&
Evenbeck,
S.
(1973).
Attitudinal politics: The strategy of
moderation.
Journal
of
Personality and Social Psychology,
25,
100-108.
Cialdini, R.
B.,
Levy,
A., Herman, P., Kozlowski,
L.,
&
Petty,
R.
E.
(1976).
Elastic shifts of
opinion: Determinants of direction and durability.
Journal of Personality and Social Psychol-
Cialdini,
R.
B.,
&
Petty,
R.
E.
(1981).
Anticipatory opinion effects.
In
R.
Petty, T. Ostrom,
&
T.
Brock
(Eds.),
Cognitive responses in persuasion
(pp.
217-235).
Hillsdale, NJ: Erlbaum.
Cialdini,
R.
B., Petty,
R.
E.,
&
Cacioppo,
J.
T.
(1981).
Attitude andattitudechange.
AnnualReview
Clark,
M.
S.,
&
Isen,
A.
M.
(1982).
Toward understanding the relationship between feeling states
and social behavior. In A. Hastorf
&
A. Isen
(Eds.),
Cognitive social psychology.
New York:
Elsevier-North Holland.
Cohen, A.
(1957).
Need for cognition and order of communication as determinants of opinion
change. In C. Hovland
et al.
(Eds.),
The order ofpresentation
in
persuasion.
New Haven, CT:
Yale University
Press.
PSyChOlOgy,
41,
1-12.
605-614.
ogy,
34,
663-612.
of PSycholOgy,
32,
351-404.
196
RICHARD
E.
PETTY
AND
JOHN
T.
CACIOPPO
Cohen, A,, Stotland, E.,
&
Wolfe,
D.
(1955).
An experimental investigation of need for cognition.
Journal of Abnormal and Social Psychology,
51,
291-294.
Cook, T.
D.,
&
Flay,
B.
(1978).
The temporal persistence of experimentally induced attitude change:
An evaluative review.
In
L.
Berkowitz (Ed.),
Advances in experimental social psychology
(Vol.
11).
New York Academic Press.
Cooper, H. M.
(1979).
Statistically combining independent studies: meta-analysis of sex differences
in conformity research.
Journal of Personality and Social Psychology,
37,
I3
1
-
146.
Cooper,
J.,
&
Croyle,
R.
T.
(1984).
Attitudes and attitude change.
Annual Review
of
Psychology,
Cooper, J.,
&
Jones, R. A.
(1970).
Self-esteem and consistency as determinants of anticipatory
opinion change.
Journal of Personality and Social Psychology.
14,
312-320.
Craik, F.
1.
M.
(1979).
Human Memory.
Annual Review of Psychology,
30,
63-102.
Crocker, J., Fiske,
S.
T.,
&
Taylor,
S.
E.
(1984).
Schematic bases of belief change. In R. Eiser
(Ed.),
Attitudinal judgment.
New York: Springer-Verlag.
Deaux,
K.
(1984).
From individual differences to social categories: Analysis
of
a decade’s research
on gender.
American Psychologist,
39,
105-1 16.
Deci, E.
L.,
&
Ryan, R.
M.
(1980).
The empirical exploration of intrinsic motivational processes.
In
L.
Berkowitz (Ed.),
Advances in experimental socialpsychology
(Vol.
13).
New York: Academ-
ic Press.
Eagly,
A.
H.
(1967).
Involvement as a determinant
of
response to favorable and unfavorable
information.
Journal of Personality and Social Psychology Monograph,
7(3
F’t.
2).
Eagly, A. H.
(1974).
Comprehensibility of persuasive arguments as a determinant of opinion change.
Journal of Personality and Social Psychology,
29,
758-773.
Eagly, A. H.
(1978).
Sex differences in influenceability.
Psychological Bulletin,
85,
86-1 16.
Eagly, A. H.,
&
Carli,
L.
(1983).
Sex of researchers and sex-typed communications as determinants
of sex differences in influenceability.
Psychological Bulletin.
90,
1-20,
Eagly, A.
H.,
&
Chaiken,
S.
(1984).
Cognitive theories
of
persuasion. In
L.
Berkowitz
(Ed.),
Advances
in
experimental social psychology
(Vol.
17).
New York:. Academic
Press.
Eagly, A. H.,
&
Himmelfarb,
S.
(1974).
Current trends in attitude theory and research. In
S.
Himmelfarb
&
A. Eagly (Eds.),
Readings in attitude change.
New York: Wiley.
Eagly, A. H.,
&
Himmelfarb,
S.
(1978).
Attitudes and opinions.
AnnualReview OfPsychology,
29,
517-554.
Eagly, A. H.,
&
Manis,
M.
(1966).
Evaluation of message and communication as a function
of
involvement.
Journal of Personality and Social Psychology,
3,
483-485.
Eagly, A. H.,
&
Warren, R.
(1976).
Intelligence, comprehension, and opinion change.
Journal of
Personality.
44,
226-242.
Eagly, A. H., Wood,
W.,
&
Chaiken,
S.
(1978).
Causal inferences about communicators
and their effect on opinion change.
Journal
of
Personality and Social Psychology,
36,
424-
435.
Ekstrom,
R.
B., French, J.
W..
&
Harmon,
H.
H.
(1962).
Kit of factor referenced cognitive tests.
Princeton, NJ: Educational Testing Service.
Elms, A.
C.
(1966).
Influence of fantasy ability on attitude change through role-playing.
Journal of
Personality and Social Psychology,
4,
36-43.
Fazio,
R.
H.
(1985).
How do attitudes guide behavior? In R.
M.
Sorrentino
&
E.
T. Higgins
(Eds.),
The handbook of morivation and cognition: Foundations of social behavior.
New York:
Guilford
.
Fazio, R. H., Chen,
I.,
McDonel, E.,
&
Sherman,
S.
J.
(1982).
Attitude accessibility, attitude-
behavior consistency, and the
strength
of
the object-evaluation association.
Journal
of
Experi-
mental Social Psychology,
18,
339-357.
35,
395-426.
ELABORATION LIKELIHOOD MODEL
OF
PERSUASION
197
Fazio,
R.
H.,
&
Zanna, M. P.
(1981).
Direct experience and attitude behavior consistency. In L.
Berkowitz
(Ed.),
Advances
in
experimental social psychology,
(Vol.
14,
pp.
161-202).
New
York:
Academic
Press.
Festinger,
L.
(1950).
Informal social communication.
Psychological Review,
57,
271-282.
Festinger,
L.
(1954).
A theory
of
social comparison processes.
Human Relations,
7,
117-140.
Festinger,
L.
(1957).
A
theory of cognitive dissonance.
Stanford, CA: Stanford University Press.
Festinger, L.,
&
Maccoby,
N.
(1964).
On
resistance to persuasive communications.
Journal
of
Abnormal and Social Psychology,
68,
359-366.
Fishbein, M.
(1980).
A theory of reasoned action: Some applications and implications.
In
H. Howe
&
M.
Page
(Eds.),
Nebraska symposium
on
motivation.
1979.
Lincoln: University of Nebraska
Press.
Fishbein, M.,
&
Ajzen,
I.
(1972).
Attitudes and opinions.
AnnualReview of Psychology,
23,
487-
544.
Fishbein, M.,
&
Ajzen,
I.
(1975).
Belief, attitude, intention,
and
behavior: An introduction to theory
and research.
Reading, MA: Addison-Wesley.
Fishbein, M.,
&
Ajzen,
I.
(1981).
Acceptance, yielding and impact: Cognitive processes in persua-
sion. In
R.
Petty, T. Ostrom,
&
T.
Brock
(Eds.),
Cognirive responses
in
persuasion.
Hillsdale,
NJ:
Erlbaum.
Fiske,
S.
T., Kinder, D. R.,
&
Larter, W. M.
(1983).
The novice and the expert: Knowledge-based
strategies in political cognition.
Journal of Experimental Sociul Psychology,
19,
38 1-400.
Folkes,
V.
(1985).
Mindlessness or mindfulness: A partial replication and extension
of
Langer,
Blank and Chanowitz.
Journal of Personality and Social Psychology,
48,
600-604.
Forer,
B.
(1949).
The fallacy of personal vlaidation.
Journal of Abnormal and Social Psychology,
Freedman,
J.
L.
(1964).
Involvement, discrepancy, and change.
Journal
of
Abnormal and Social
Freedman,
J.
L.,
&
Sears,
D.
0.
(1965).
Warning, distraction and resistance to influence.
Journal of
Personality and Social Psychology,
1,
262-266.
Freedman,
J.
L., Sears,
D.
O.,
&
O'Conner,
E.
F.
(1964).
The effects of anticipated debate and
commitment
on
the polarization of audience opinion.
Public Opinion Quarterly,
28,615-627.
Gom,
G.
(1982).
The effects of music in advertising
on
choice behavior: A classical conditioning
approach.
Journal of Marketing Research,
46,
94-101.
Gom,
G.,
&
Goldberg,
M.
(1980).
Children's responses
to
repetitive TV commercials.
Journal
of
Consumer Research,
6,
421-425.
Greenwald, A.
G.
(1968).
Cognitive learning, cognitive response to persuasion, and attitude change.
In
A. Greenwald,
T.
Brock,
&
T. Ostrom
(Eds.),
Psychological foundations of attitudes
(pp.
148-170).
New York: Academic
Press.
Greenwald. A.
G.
(1980).
The totalitarian ego: Fabrication and revision of personal history.
Ameri-
can Psychologist,
35,
603-618.
Greenwald, A.
G.
(1981).
Ego task analysis: An integration
of
research
on
ego-involvement.
In
A.
,
.~
Hastorf
&
A. Isen
(Eds.),
Cognitive social psychology.
Amsterdam: Elsevier.
Griffit,
W.,
&
Veitch. R.
(1971).
Hot and crowded: Influences of population density and temper-
ature
on
interpersonal affective behavior.
Journal of Personality and Social Psychology,
17,
Harkins,
S.
G.,
Latan6, B.,
&
Williams,
K.
D.
(1980).
Social loafing: Allocaing effort or taking it
easy.
Journal of Experimental social Psychology,
16,
457-465.
Harkins,
S.
G.,
&
Petty,
R.
E.
(1981a).
The effects of source magnification of cognitive effort
on
attitudes: An information processing view.
Journal of Personaliry and Social Psychology,
40,
44,
118-123.
Psychology,
69,
290-295.
92-98.
40 1-413.
198
RICHARD
E.
PETTY AND JOHN
T.
CACIOPPO
Harkins,
S.
G.,
&
Petty,
R.
E. (1981b). The multiple source effect in persuasion: The effects of
distraction.
Personality and Social Psychology Bulletin,
7,
627-635.
Harkins,
S.
G.,
&
Petty,
R.
E. (1982). Effects of task difficulty and task uniqueness
on
social
loafing.
Journal
of
Personality and Social Psychology,
43,
1214-1229.
Harkins,
S.
G.,
&
Petty,
R.
E. (1983). Social context effects in persuasion: The effects
of
multiple
sources and multiple targets.
In
P. Paulus
(Ed.),
Basic group processes.
New York Springer-
Verlag.
Harkness, A.
R.,
DeBono, K. G.,
&
Borgida, E. (1985). Personal involvement and strategies for
making contingency judgments: A stake in the dating game makes a difference.
Journal
of
Personality and Social Psychology,
49,
22-32.
Harrison,
A. A. (1977). Mere exposure. In
L.
Berkowitz (Ed.),
Advances in experimental social
psychology
(Vol. 10). New
York:
Academic
Press.
Harvey,
0.
J.,
Hunt,
D.
E.,
&
Schroder, H.
M.
(1961).
Conceptual systems and personality
orgonizarion.
New Yo&: Wiley.
Hass,
R.
G. (1981). Effects of source characteristics
on
cognitive responses and persuasion.
In
R.
Petty,
T.
Ostrom,
&
T.
Brock
(Eds.),
Cognitive responses in persuasion.
Hillsdale, NJ:
Erlbaum.
Hass,
R.
G.,
&
Grady,
K.
(1975). Temporal delay, type of forewarning, and resistance to influence.
Journal
of
Experimental Social Psychology,
11,
459-469.
Heesacker,
M.,
Petty,
R.
E.,
&
Cacioppo, J. T. (1983). Field dependence and attitude change:
Source credibility can alter persuasion by affecting message-relevant thinking.
Journal
of
Per-
SOMlity.
51,
653-666.
Heider,
F.
(1946). Attitudes and cognitive organization.
Journal
of
Psychology,
21,
107-1 12.
Higgins,
T.,
Herman, C. P.,
&
Zanna,
M.
P.
(Eds.),(1981).
Social cognition: The Ontario sym-
posium
(Vol., 1). Hillsdale, NJ: Erlbaum.
Himmelfarb.
S.,
&
Eagly. A. H. (1974). Orientations to the study of attitudes and their change. In
S.
Himmelfarb
&
A. Eagly
(Eds.),
Readings in artitude change.
New York: Wiley.
Hovland,
C.
I.,
Harvey,
0.
J.,
&
Sherif, M. (1957). Assimilation and contrast effects in reactions
to
communications and attitude change.
Journal
of
Abnormal and Social Psychology,
55,
244-
252.
Hovland, C.
l.,
Janis,
I.,
&
Kelley,
H.
H.
(1953).
Communication andpersuasion.
New Haven, CT:
Yale University Press.
Hovland, C.
I.,
&
Weiss, W. (1951). The influence of source credibility on communication effec-
tiveness.
Public Opinion Quarterly,
15,
635-650.
Ingham, A,, Levinger,
G.,
Graves,
J.,
&
Peckham,
V.
(1974). The Ringelmann effect: Studies of
group size and group performance.
Journal
of
Experimenral Social Psychology,
10,
371-384.
Insko,
C.
A. (1981). Balance theory and phenomenology.
In
R.
Petty,
T.
Ostrom,
&
T. Brock
(Eds.).
Cognitive responses
in
persuasion.
Hillsdale, NJ: Erlbaum.
Insko, C.
A.
(1984). Balance theory, the Jordan paradigm, and the Wiest tetrahedron.
In
L.
Berkowitz
(Ed.),
Advances
in
experimental social psychology
(Vol. 18). New York: Academic
Press.
Insko,
C.
A., Lind, E. A.,
&
LaTour,
S.
(1976). Persuasion, recall, and thoughts.
Representative
Research
in
Social Psychology.
7,
66-78.
Insko,
C. A., Turnbull,
W.,
&
Yandell, B. (1974). Facilitating and inhibiting effects of distraction
on
attitude change.
Sociomerry.
37,
508-528.
Janis,
1.
L.,
&
Gilmore,
J.
B. (1965). The influence of incentive conditions
on
the success of role
playing in modifying attitudes.
Journal
of
Personality and Social Psychology,
1,
17-27.
Janis,
I.
L.,
Kaye,
D.,
&
Kirschner, P. (1965). Facilitating effects of “eating while reading”
on
responsiveness to persuasive communications.
Journal
of
Personality and Social Psychology,
1,
181- 186.
ELABORATION LIKELIHOOD MODEL
OF
PERSUASION
199
Johnson,
E.
J.,
&
Russo, J.
E.
(1981).
Product familiarity and learning new information.
Advances
Johnson, H.
H.,
&
Watkins,
T.
A.
(1971).
The effects of message repetitions on immediate and
Kahle,
L.
R.,
&
Homer, P. M.
(1985).
Physical attractiveness of the celebrity endorser: A social
Kahneman,
D.,
Slovic,
P.,
&
Tversky, A.
(Eds.)
(1982).
Judgment under uncertainty: Heuristics
Karabenick,
S.
A.
(1983).
Sex-relevance of content and influenceability: Sistrunk and McDavid
Katz,
D.
(1960).
The functional approach
to
the study
of
attitudes.
Public Opinion Quurterly
24,
Keating, J. P.,
&
Brock, T. C.
(1974).
Acceptance
of
persuasion and the inhibition of counterargu-
mentation under various distraction tasks.
Journal
of
Experimental Social Psychology,
10,301
-
309.
Kelley,
H.
H.
(1967).
Attribution theory in social psychology. In
D.
Levine
(Ed.),
Nebraska
symposium
on
motivation
(Vol.
15).
Lincoln: University of Nebraska
Press.
Kelman,
H.
C.
(1961).
Processes of opinion change.
Public Opinion Quurrerly, 25,
57-78.
Kelman,
H.
C.,
&
Eagly, A.
H.
(1965).
Attitude toward the communicator, perception of commu-
nication content, and attitude change.
Journal
of
Personality and Social Psychology,
1,
63-
78.
Kelman,
H.
C.,
&
Hovland, C.
1.
(1953).
Reinstatement of the communicator in delayed measure-
ment of opinion change.
Journal
of
Abnormul and Social Psychology,
48,
327-335.
Kerr, N.,
&
Bruun,
S.
(1981).
Ringelmann revisited: Alternative explanations for the social loafing
effect.
Personality and Social Psychology Bulletin,
7,
224-231.
Kiesler, C. A., Collins, B.,
&
Miller, N.
(1969).
Attitude change:
A
critical analysis
of
theoretical
approaches.
New York: Wiley.
Kiesler, C. A.,
&
Kiesler,
S.
(1964).
Role
of
forewarning in persuasive communications.
Journal
of
Abnormul and Social Psychology,
68,
547-549.
Kiesler,
S.
B.,
&
Mathog, R.
(1968).
The distraction hypothesis in attitude change.
Psychological
Reports.
23,
1123-1133.
Kiesler, C. A,,
&
Munson, P. A.
(1975).
Attitudes and opinions.
Annual Review
of
Psychology, 26,
King, B.
T.,
&
Janis.
I.
L.
(1956).
Comparison of the effectiveness of improvised versus non-
improvised role-playing in producing opinion change.
Humun Relations,
9,
177- 186.
Kleinhesselink, R. R., &Edwards, R. E.
(1975).
Seeking and avoiding belief-discrepant information
as a function of its perceived refutability.
Journal
of
Personality and Social Psychology,
31,
Kohlberg, L.
(1963).
The development
of
children's orientations toward a moral order.
I.
Sequence
in the development of moral thought.
Vita Humuna, 6,
11-33.
Kunst-Wilson, W. R.,
&
Zajonc,
R.
B.
(1980).
Affective discrimination of stimuli that cannot
be
recognized.
Science,
207,
557-558.
Lammers, H.
B.,
&
Becker,
L.
A.
(1980).
Distraction: Effects on the perceived extremity of a
communication and on cognitive responses.
Personality and Social Psychology Bulletin, 6,
Landman, J.,
&
Manis, M.
(1983).
Social cognition: Some historical and theoretical perspectives. In
L.
Berkowitz
(Ed.),
Advances
in
experimental social psychology
(Vol.
16).
New York: Aca-
demic
Press.
Langer,
E.
(1978).
Rethinking the role of thought in social interaction. In J. Harvey,
W.
Ickes,
&
R.
Kidd
(Eds.),
New directions in attribution research
(Vol.
2).
Hillsdale, NJ: Erlbaum.
in
Consumer Research,
8,
15
1
-
155.
delayed attitude change.
Psychonomic Science,
22,
101- 103.
adaptation perspective.
Journal
of
Consumer Research,
11,
954-961.
and biases.
London and New York Cambridge University Press.
revisited.
Personality and Social Psychology Bulleiin.
9,
243-252.
163-204.
41 5-456.
787-790.
26 1-266.
200
RICHARD
E.
PETTY
AND
JOHN
T.
CACIOPPO
Langer, E., Blank, A,,
&
Chanowitz, B.
(1978).
The mindlessness of ostensibly thoughtful action.
JOUrM! of Personality and social Psychology,
36,
635-642.
Lastovicka.
J.,
&
Gardner,
D.
(1979).
Components of involvement.
In
J.
Maloney
&
B.
Silverman
(Eds.),
Arrirude research
phys
for high stakes.
Chicago: American Marketing Associ-
ation.
Latank,
B.,
Williams,
K.,
&
Harkins,
S.
G.
(1979).
Many hands make light the work: The causes
and consequences
of
social
loafing. Journal
of
Personality and Social Psychology,
37,
822-
832.
Lepper,
M.
R., Greene, D.,
&
Nisbett,
R.
E.
(1973).
Undermining children’s intrinsic interest with
extrinsic reward A test of the “overjustification” hypothesis.
Journal of Personality and Social
Lingle,
J.
H.,
&
Ostrom, T. M.
(1981).
Principles of memory and cognition in attitude formation. In
R. E. Petty, T. Ostrom,
&
T. Brock
(Eds.),
Cognitive responses in persuasion.
Hillsdale, NJ:
Erlbaum.
Linville, P.
(1982).
The complexity-extremity effect and age based stereotyping.
Journal
of
Person-
ality and Social Psychology,
42,
193-210.
Lord,
C.
G.,
Ross, L.,
&
Lepper, M. R.
(1979).
Biased assimilation and attitude polarization: The
effects
of
prior theories on subsequently considered evidence.
Journal of Personality and social
Lowin, A.
(1967).
Approach and avddance as alternate modes
of
selective exposure to information.
Journal of Personality and Social Psychology,
6,
1-9.
Maass, A., &Clark, R.
D.
(1984).
Hidden impact
of
minorities. Fifteen years of minority influence
research.
Psychological Bulletin,
95,
428-450.
Maddux,
J.
E.,
&
Rogers,
R.
W.
(1980).
Effects of source expertness, physical attractiveness, and
supporting arguments
on
persuasion: A case
of
brains over beauty.
Journal of Personality and
Social Psychology,
38,
235-244.
Markus,
H.
(1977).
Self-schemata and processing information about the self.
Journal
of
Personaliry
and Social Psychology,
35,
63-78.
McDavid,
J.
(1959).
Personality and situational determinants of conformity.
Journal
of
Abnormal
and Social Psychology,
58,
241-246.
McGuire,
W.
J.
(1964).
Inducing resistance to persuasion: Some contemporary approaches.
In
L.
Berkowitz
(Ed.),
Advances
in
experimental social psychology
(Vol.
I).
New York: Academic
Press.
McGuire,
W.
J.
(1968).
Personality and attitude change:
An
information-processing theory. In A.
Greenwald, T. Brock,
&
T. Ostrom
(Eds.),
Psychological foundarions of anirudes.
New York:
Academic
Press.
McGuire, W.
1.
(1969).
The nature of attitudes and attitude change. In
G.
Lindzey
&
E. Aronson
(Eds.),
The handbook of social psychology
(2nd ed.,
Vol.
3).
Reading, MA: Addison-
Wesley.
McGuire, W.
J.
(1981).
The probabilogical model of cognitive structure and attitude change.
In
R.
E. Petty, T.
M.
Ostrom,
&
T. C. Brock (Eds.),
Cognitive responses
in
persuasion.
Hillsdale,
NJ: Erlbaum.
McGuire,
W.
J.
(1985).
Attitudes and attitude change.
In
G.
Lindzey
&
E.
Aronson
(Eds.),
Hand-
book of
social
psychology
(3rd ed.,
Vol.
2.).
New York Random House.
McGuire, W.
J.,
&
Papageorgis,
D.
(1962).
Effectiveness
of
forewarning in developing resistance to
persuasion.
Public Opinion Quarterly,
26,
24-34.
Millar,
M.
G.,
&
Tesser, A.
(1984).
Thought-induced attitude change: The efecrs of schema
structure and commitment.
Unpublished manuscript. University of Georgia, Athens.
Miller,
N.
(1965).
Involvement and dogmatism as inhibitors of attitude change.
Journal of Experi-
menrd
Social Psychology,
1,
121-132.
P~chology.
28,
129-137.
P~y~hol~gy.
37,
2098-2109.
ELABORATION LIKELIHOOD MODEL
OF
PERSUASION
20
1
Miller,
N.,
&
Baron, R.
S.
(1973).
On
measuring counterarguing.
Journal
for
the Theory
of
Social
Behavior,
3,
101-118.
Miller, N.,
&
Colman,
D.
(1981). Methodological issues in analyzing the cognitive mediation of
persuasion. In
R.
E. Petty, T.
M.
Ostrom,
&
T. C. Brock
(Eds.),
Cognitive responses
in
persuasion.
Hillsdale, NJ: Erlbaum.
Miller,
N.,
Mamyama,
G.,
Beaber, R.,
&
Valone,
K.
(1976). Speed of speech and persuasion.
Journal
of
Personality and Social Psychology,
34,
615-625.
Miller, R.
L.,
Brickman, P.,
&
Bolen, D. (1975). Attribution versus persuasion as a means for
modifying behavior.
Journa/
of
Personality and Social Psychology,
31,
430-441.
Mills, J. (1966). Opinion change as a function of the communicator’s desire to influence and liking
for
the audience.
Journal
of
Experimental Social Psychology,
2,
152-159.
Mills, J.,
&
Aronson,
E.
(1965). Opinion change
as
a function of the communicator’s attractiveness
and desire to influence.
Journal
of
Personality and Social Psychology,
1,
173-177.
Mitnick,
L.,
&
McGinnies, E. (1958). Influencing ethnocentrism
in
small discussion groups through
a
film
communication.
Journal
of
Abnormal and Social Psychology,
56,
82-92.
Moscovici,
S.
(1980). Toward a theory of conversion behavior.
In
L.
Berkowitz
(Ed.),
Advances
in
experimental social psychology
(Vol. 13). New York: Academic Press.
Nisbett, R.
E.,
&
Wilson,
T.
D.
(1977). Telling more than
we
can know: Verbal reports on mental
processes.
Psychological Review,
84,
231 -259.
Osterhouse,
R.
A.,
&
Brock, T. C. (1970). Distraction increases yielding
to
propaganda by inhibit-
ing counterarguing.
Journal
of
Personality and Social Psychology,
15,
344-358.
Ostrom,
T.
M.,
&
Brock, T. C. (1968). A cognitive model of attitudinal involvment. In R. Abelson
et al.
(Eds.),
Theories
of
cognitive consistency: A sourcebook.
Chicago: Rand McNally.
Pallak,
M.
S.,
Mueller,
M.,
Dollar,
K.,
&
Pallak,
J.
(1972). Effect of commitment on
respon-
siveness to an extreme consonant communication.
JOUrMl
of
Personality and Social Psychol-
Pallak,
S.
S.,
Murroni,
E.,
&
Koch,
J.
(1983). Communicator attractiveness and expertise, emo-
tional versus rational appeals, and persuasion.
Social Cognition.
2,
122-141.
Papageorgis, D. (1968). Warning and persuasion.
Psychological Bulletin,
70,
271-282.
Petty,
R.
E.
(1977).
A
cognitive response analysis
of
the temporal persistence
of
attitude changes
induced by persuasive communications.
Unpublished doctoral dissertation. Ohio State Univer-
sity, Columbus.
Petty,
R.
E.,
&
Brock, T. C. (1976). Effects of responding or
not
responding to hecklers
on
audience
agreement with a speaker.
Journal
of
Applied Social Psychology,
6,
1-17.
Petty, R. E.,
&
Brock, T. C. (1979). Effects of “Barnum” personality assessments
on
cognitive
behavior.
Journal
of
Consulting and Clinical Psychology,
47,
201-203.
Petty,
R.
E.,
&
Brock,
T.
C. (1981). Thought disruption and persuasion: Assessing the validity of
attitude change experiments. In R. Petty, T. Ostrom,
&
T. Brock
(Eds.),
Cognitive responses in
persuasion
(pp. 55-79). Hillsdale,
NJ:
Erlbaum.
Petty, R. E.,
&
Cacioppo, J.
T.
(1977). Forewarning, cognitive responding, and resistance to
persuasion.
Journal
of
Personality and Social Psychology,
35,
645-655.
Petty,
R.
E.,
&
Cacioppo,
J.
T.
(1978).
A
cognitive response approach to attitudinalpersistence.
Paper presented at the annual meeting of the American Psychological Association, Toronto,
Canada.
Petty,
R.
E.,
&
Cacioppo, J. T. (1979a). Effects of forewarning of persuasive intent and involvement
on
cognitive responses and persuasion.
Personality and Social Psychology Bulletin,
5,
173-
176.
Petty,
R.
E.,
&
Cacioppo,
J.
T. (1979b). Issue-involvement can increase
or
decrease persuasion by
enhancing message-relevant cognitive responses.
JourM!
of
Personality and
Social
Psychology,
ogy,
23,
429-436.
37,
1915-1926.
202
RICHARD E. PEnY AND JOHN
T.
CACIOPPO
Petty,
R.
E.,
&
Cacioppo,
J.
T.
(1981a).
Attitudes and persuasion: Classic and contemporary
approaches.
Dubuque, IA: Wm.
C.
Brown.
Petty,
R.
E.,
&
Cacioppo,
J.
T. (1981b). Issue involvement as a moderator of the effects
on
attitude
of advertising content and context.
Advances
in
Consumer Research,
8,
20-24.
Petty,
R.
E.,
&
Cacioppo,
J.
T. (1983a). Central and peripheral mutes to persuasion: Application to
advertising.
In
L.
Percy
&
A. Woodside
(Eds.),
Advertising and consumer psychology
(pp. 3-
23). Lexington, MA: Heath.
Petty,
R.
E.,
&
Cacioppo,
J.
T. (1983b). The role of bodily responses in attitude measurement and
change.
In
J.
T. Cacioppo
&
R.
E.
Petty
(Eds.),
Social psychophysiology: A sourcebook.
New
York: Guilford.
Petty,
R.
E.,
&
Cacioppo,
J.
T. (1984a). The effects
of
involvement
on
responses to argument
quantity and quality: Central and peripheral routes to persuasion.
Journal
of
Personality and
Social Psychology,
46,
69-81.
Petty,
R.
E.,
&
Cacioppo.
J.
T. (1984b). Motivational factors in consumer response to advertise-
ments.
In
R.
Geen,
W. Beatty,
&
R.
Arkin,
Human motivation: Physiological. behavioral, and
social approaches
(pp. 418-454). Boston: Allyn
&
Bacon.
Petty,
R.
E.,
&
Cacioppo,
J.
T. (19%). Source factors and the elaboration likelihood model of
persuasion.
Advances in Consumer Research,
11,
668-672.
Petty,
R.
E.,
&
Cacioppo,
J.
T. (1986).
Communication and persuasion: Central and peripheral
routes
to
attitude change.
New York: Springer-Verlag, in press.
Petty,
R.
E., Cacioppo,
J.
T.,
&
Goldman,
R.
(1981). Personal involvement
as
a determinant of
argument-based persuasion.
Journal
of
Personality and Social Psychology,
41,
847-855.
Petty,
R.
E., Cacioppo,
J.
T.,
&
Harkins,
S.
G.
(1983). Group size effects
on
cognitive effort and
attitude change.
In
H. Blumberg, A. Hare, V. Kent,
&
M. Davies
(Eds.),
Small
groups and
social interaction
(Vol.
1,
pp. 165-181). London: Wiley.,
Petty,
R.
E.,
Cacioppo,
J.
T.,
&
Heesacker, M. (1981). The use of rhetorical questions in persua-
sion:
A
cognitive response analysis.
Journal
of
Personality and Social Psychology,
40,
432-
440.
Petty,
R.
E.,
Cacioppo,
J.
T.,
&
Heesacker, M. (1984). Central and peripheral routes to persuasion:
Application to counseling.
In
R.
McGlynn,
J.
Maddux, C. Stoltenberg,
&
J.
Harvey
(Eds.),
Social perception in clinical and counseling psychology
(pp. 59-89). Lubbock: Texas Tech
University Press.
Petty,
R.
E., Cacioppo,
J.
T.,
&
Heesacker,
M.
(1985).
Persistence
of
persuasion:
A
test
of
the
Elaboration Likelihood Model.
Unpublished manuscript. University of Missouri, Columbia.
Petty,
R.
E., Cacioppo,
J.
T.,
&
Kasmer.
J.
(1985).
Efecrs
of
need
for
cognition on social loafing.
Paper presented at the Midwestern Psychological Association Meeting, Chicago.
Petty,
R.
E.. Cacioppo,
J.
T.,
&
Schumann,
D.
(1983).
Central
and peripheral routes
to
adveflising
effectiveness: The moderating role of involvement.
Journal
of
Consumer Research,
10,
134-
148.
Petty,
R.
E., Cacioppo,
J.
T.,
&
Schumann,
D.
(1984). Attitude change and personal selling.
In
J.
Jacoby
&
S.
Craig
(Eds.),
Personal selling: Theory, research, and practice
(pp. 29-55).
Lexington, MA: Heath.
Petty,
R.
E.,
Harkins,
S.
G.,
&
Williams, K.
D.
(1980). The effects
of
group diffusion of cognitive
effort
on
attitudes: An information processing view.
Journal
of
Personality and Social Psychol-
Petty,
R.
E.,
Harkins,
S.
G.,
Williams, K.
D.,
&
Latanb, B. (1977). The effects of group size
on
cognitive effort and evaluation.
Personality and Social Psychology Bulletin,
3,
579-
582.
Petty,
R.
E..
Ostrom,
T.
M.,
&
Brock, T. C. (Eds.) (1981).
Cognitive responses in persuasion.
Hlllsdale,
NJ:
Erlbaum.
ogy,
38,
81-92.
ELABORATlON LIKELIHOOD
MODEL
OF
PERSUASION
203
Petty, R.
E.,
Wells,
G.
L.,
&
Brock. T. C.
(1976).
Distraction can enhance or reduce yielding to
propaganda: Thought disruption versus effort justification.
Journal of Personality and Social
Petty, R. E., Wells,
G.
L.,
Heesacker, M., Brock, T.,
&
Cacioppo,
J.
T.
(1983).
The effects of
recipient posture
on
persuasion:
A
cognitive response analysis.
Personality and Social Psychol-
ogy
Bulletin.
9,
209-222.
Puckett,
J.,
Petty, R.
E.,
Cacioppo,
J.
T.,
&
Fisher, D.
(1983).
The relative impact of age and
attractiveness stereotypes
on
persuasion.
JOUrMl of Gerontology,
38,
340-343.
Regan, D.
T.,
&
Cheng,
J.
B.
(1973).
Distraction and attitude change:
A
resolution.
Journal of
Experimental Social Psychology,
9,
138- 147.
Regan,
D.
T.,
&
Fazio, R.
(1977).
On
the consistency between attitude and behavior: Look to the
method of attitude formation.
Journal of Experimental Social Psychology,
13,
28-45.
Rhine,
R.,
&
Severance,
L.
(1970).
Ego-involvement, discrepancy, source credibility, and attitude
change.
Journal of Personality and Social Psychology,
16,
175-190.
Ronis, D.
L.,
Baumgardner, M., Leippe,
M.,
Cacioppo,
J.
T.,
&
Greenwald,
A.
G.
(1977).
In
search
of
reliable persuasion effects:
I.
A
single session procedure for studying persistence of
persuasion.
JOUrMl
of
Personality and Social Psychology,
35,
548-569.
Rosenberg, M.
(1956).
Cognitive structure and attitudinal affect.
JOUrMl
of
Abnormal and Social
Ross, E. A.
(1908).
Social psychology: An outline and a sourcebook.
New York: Macmillan.
Ross,
L.
(1981).
The “intuitive scientist” formulation and its developmental implications.
In
J.
H.
Flaveil
&
L. Ross
(Eds.),
Social cognitive development: Frontiers andpossible futures.
London
and New York: Cambridge University
Press.
Ross,
L.,
Lepper, M.,
&
Hubbard, M.
(1975).
Perseverance in self-perception and social perception:
Biased attributional processes in the debriefing paradigm.
JOUrMl
of
Personality and Social
Sadler,
O.,
&
Tesser,
A.
(1973).
Some effects
of
salience and time upon interpersonal hostility and
attraction during social isolation.
Sociometry,
36,
99-1 12.
Sandelands,
L.
E.,
&
Larson,
J.
R.
(1985).
When measurement causes task attitudes: A note from the
laboratory.
Journal
of
Applied Psychology,
70,
116-121.
Sawyer,
A.
G.
(1981).
Repetition, cognitive responses and persuasion.
In
R. Petty, T. Ostrom,
&
T.
Brock
(Eds.),
Cognitive responses in persuasion
(pp.
237-261).
Hillsdale,
NJ:
Erlbaum.
Schneider, W.,
&
Shiffrin, R. M.
(1977).
Controlled and automatic human information processing:
I.
Detection,
search,
and attention.
Psychological Review,
84,
1-66.
Schumann,
D.,
Petty, R.
E.,
&
Cacioppo,
J.
T.
(1985).
Effects
of
involvement, repetition, and
variation
on
responses
to
advertisements.
Unpublished manuscript. University
of
Missouri,
Columbia.
Sherif, C. W., Kelly, M., Rodgers, H.
L.,
Sarup,
G.,
&
Tittler, B.
(1973).
Personal involvement,
social judgment, and action.
Journal
of
Personality and Social Psychology,
27,
311-327.
Sherif, C. W., Sherif, M.,
&
Nebergall,
R.
E.
(1965).
Artirude and attitude change: The social
judgment-involvement approach.
Philadelphia: Saunders.
Sherif, M.
(1977).
Crisis
in
social psychology: Some remarks towards breaking through the crisis.
Personality and Social Psychology Bulletin,
3,
368-382.
Sherif, M.,
&
Hovland, C.
I.
(1961).
Social judgment: Assimilation and contrast effects
in
commu-
nication and attitude change.
New Haven,
CT:
Yale University
Press.
Sherif, M.,
&
Sherif, C.
W.
(1967).
Attitude as the individual’s own categories: The social judg-
ment-involvement approach to attitude and attitude change.
In
C. W. Sherif
&
M. Sherif
(Eds.),
Attitude, ego-involvement. and change.
New York Wiley.
Sherman,
S.
J.,
&
Fazio,
R.
H.
(1983).
Parallels between attitudes and traits as predictors
of
behavior.
Journal of Personality,
51,
308-345.
Psychology,
34,
874-884.
PSychOlOgy.
53,
367-372.
Psychology,
32,
880-892.
204
RICHARD
E. PEITY
AND
JOHN
T.
CACIOPPO
Sistrunk, F.,
&
McDavid,
J.
W.
(1971).
Sex variable in conforming behavior.
Journalof Personaliry
Sivacek,
J.,
&
Crano, W. D.
(1982).
Vested interest as a moderator of attitude-behavior consisten-
Smith, B.
L.,
Lasswell,
H.
D.,
&
Casey,
R.
D.
(1946).
Propaganda, communication, andpublic
Smith, M.
J.
(1982).
Persuasion and human action.
Belmont, CA: Wadsworth.
Snyder, M.,
&
DeBono,
K.
G.
(1985).
Appeals
to
image and claims about quality: Understanding
the psychology of advertising.
Journal
of
Personality and Social Psychology,
49,
586-597.
Srull,
T.
K.
(1983).
The role of prior knowledge in the acquisition, retention, and use of new
information.
Advances in Consumer Research,
10,
572-576.
Staats, A. W.,
&
Staats, C.
K.
(1958).
Attitudes established by classical conditioning.
Journal
of
Abnormal and Social Psychology,
57,
37-40.
Staats, C.
K.,
&
Staats, A. W.
(1957).
Meaning established by classical conditioning.
Journal
of
Experimental Psychology,
54,
74-80.
Steiner,
1.
(1972).
Group process and productivity.
New York: Academic Press.
Sternthal, B., Dholakia, R.,
&
Leavitt, C.
(1978).
The persuasive effect of source credibility: A test
of
cognitive response analysis.
JOUrMl
of
Consumer Research,
4,
252-260.
Taylor,
S.
E.
(1975).
On infemng one’s attitude from one’s behavior: Some delimiting conditions.
Journal
of
Personaliry and Social Psychology,
31,
126-131.
Taylor,
S.
E.,
&
Fiske,
S.
(1978).
Salience, attention, and attributions: Top of the head phenomena.
In
L.
Berkowitz (Ed.),
Advances in experimental social psychology
(Vol.
11).
New York:
Academic Press.
and Social Psychology,
17,
200-207.
cy.
JOUrMl
of
Personality and Social Psychology,
43,
210-221.
opinion.
Princeton,
NJ:
Princeton University Press.
Taylor,
S.
E.,
&
Fiske,
S.
(1984).
Social cognition.
Reading, MA: Addison-Wesley.
Tesser, A.
(1976).
Thought and reality constraints as determinants
of
attitude polarization.
Journal
of
Tesser, A.
(1978).
Self-generated attitude change. In L. Berkowitz
(Ed.),
Advances in experimental
Tesser,
A.,
&
Conlee, M.
C.
(1975).
Some effect of time and thought on attitude polarization.
Tesser,
A.,
&
Leone, C.
(1977).
Cognitive schemas and thought as determinants
of
attitude change.
Thurstone,
L. L.
(1928).
Attitudes can be measured.
American Journal
of
Sociology,
33,
529-544.
Tsal, Y.
(1984).
The
role
of
attention in processing information from advertisements.
Unpublished
manuscript. Cornell University, Ithaca, New York.
Vinokur, A.,
&
Burnstein,
E.
(1974).
The effects of partially shared persuasive arguments on group-
induced shifts: A group problem solving approach.
Journal
of
Personality and Social Psychol-
ogy,
29,
305-315.
Watts, W. A.
(1967).
Relative persistence of opinion change induced by active compared to passive
participation.
Journal
of
Personality and Social Psychology,
5,
4-
15.
Watts, W. A,,
&
Holt,
L.
E.
(1979).
Persistence of opinion change induced under conditions
of
forewarning and distraction.
Journal
of
Personality and Social Psychology,
37,
778-789.
Weber,
S.
J.
(1972).
Opinion change is a function
of
the associarive learning
of
content and source
factors.
Unpublished doctoral dissertation. Northwestern University, Evanston, Illinois.
Wells,
G.
L.,
&
Petty, R. E.
(1980).
The effects of overt head-movements on persuasion: Com-
patibility and incompatibility of responses.
Basic and Applied Social Psychology,
1,
219-230.
White,
G.
L.
(1975).
Contextual determinants
of
opinion judgments: Field experimental probes of
judgmental relativity boundary conditions.
Journal
of
Personality and Social Psychology,
32,
Research in Personality,
10,
183-194.
social psychology
(Vol.
1
I).
New York: Academic Press.
Journal
of
Personaliry and Social Psychology,
31,
262-270.
Journal
of
Experimental Social Psychology,
13,
340-356.
1047-1054.
ELABORATION LIKELIHOOD MODEL
OF
PERSUASION
205
Wicker,
A.
(1971).
An examination of the “other variable” explanation of attitude-behavior incon-
sistency.
Journal
of
Personality and social Psycho~ogy.
19,
18-30.
Wilson, T. D., DUM,
D.,
Bybee,
J.,
Hyman, D.,
&
Rotondo,
1.
(1984).
Effects of analyzing reasons
on attitude-behavior consistency.
Journal
of
Personality and Social Psychology,
47,
5-
16.
Wirkin, H.
A,,
Goodenough, D. R.,
&
Oltman, P.
K.
(1979).
Psychological differentiation: Current
status.
Journal
of
Personality and Social Psychology,
37,
1127- 1145.
Witt,
W.
(1976).
Effects
of
quantification in scientific writing.
Journal
of
Communication,
26,
67-
69.
Wood, W.
(1982).
Retrieval of attitude-relevant information from memory: Effects on susceptibility
to persuasion and on intrinsic motivation.
Journal
of
Personalify and Social Psychology,
42,
Wood, W., Kallgren, C.,
&
Priesler,
R.
(1985).
Access
to
attitude relevant infomation in memory
as a determinant of persuasion.
Journal
of
Experimental Social Psychology,
21, 73-85.
Wyer, R.
S.,
&
Srull,
T.
(1984).
The handbook
of
social cognition.
Hillsdale,
NJ:
Erlbaurn.
Yalch, R. F.,
&
Elmore-Yalch, R.
(1984).
The effect of numbers on the route to persuasion.
Journal
Zajonc, R. B.
(1968).
Attitudinal effects
of
mere exposure.
Journal
of
Personality and Social
Zajonc, R. B.
(1980).
Feeling and thinking: Preferences need no inferences.
American Psychologist,
Zajonc,
R.
B.,
&
Markus,
H.
(1982).
Affective and cognitive factors in preferences.
Journal
of
Consumer Research,
9,
123-
13
I.
Zanna, M. P., Kiesler, C. A.,
&
Pilkonis, P.
A.
(1970).
Positive and negative attitudinal affect
established by classical conditioning.
Journal
of
Personality and Social Psychology,
14, 321
-
328.
Zanna,
M.
P.,
&
Pack,
S.
J.
(1975).
On the self-fulfilling nature of apparent sex differences in
behavior.
Journal
of
Experimental Social Psychology,
11,
583-591.
Zillmann,
D.
(1972).
Rhetorical elicitation
of
agreement in persuasion.
Journal
of
Personality and
Social Psychology,
21,
159-165.
Zimbardo, P.
G.
(1960).
Involvement and communication discrepancy as determinants of opinion
conformity.
Journal
of
Abnormal and Social Psychology,
60,
86-94.
798-810.
of
Consumer Research,
11,
522-527.
Psychology Monograph Supplement,
9,
1-27.
35,
151-175.
