Annu. Rev. Psychol. 2000. 51:539–570
Copyright ᭧ 2000 by Annual Reviews. All rights reserved
0084–6570/00/0201–0539$12.00 539
ATTITUDE CHANGE: Persuasion and Social
Inﬂuence
Wendy Wood
Department of Psychology, Texas A&M University, College Station, Texas 77843;
e-mail: wlw@psyc.tamu.edu
Key Words inﬂuence, motives, fear appeals, social identity
Abstract This chapter reviews empirical and theoretical developments in
research on social inﬂuence and message-based persuasion. The review emphasizes
research published during the period from 1996–1998. Across these literatures, three
central motives have been identiﬁed that generate attitude change and resistance.
These involve concerns with the self, with others and the rewards/punishments they
can provide, and with a valid understanding of reality. The motives have implications
for information processing and for attitude change in public and private contexts.
Motives in persuasion also have been investigated in research on attitude functions
and cognitive dissonance theory. In addition, the chapter reviews the relatively unique
aspects of each literature: In persuasion, it considers the cognitive and affective mechanisms
underlying attitude change, especially dual-mode processing models, recipients’
affective reactions, and biased processing. In social inﬂuence, the chapter
considers how attitudes are embedded in social relations, including social identity
theory and majority/minority group inﬂuence.
CONTENTS
Introduction ......................................................................................540
Motives for Agreeing with Others.........................................................540
Public Versus Private Inﬂuence ............................................................. 542
Motives in Persuasion Research............................................................544
Functional Theories........................................................................... 544
Cognitive Dissonance Theory................................................................ 546
Multiple Attitudes .............................................................................. 548
Sources of Multiple Attitudes ................................................................ 548
Inﬂuence and Multiple Attitudes............................................................. 549
Dual-Mode Processing Models of Persuasion..........................................551
Cognitive Response Mediation of Attitude Change....................................... 552
Dual-Mode Processing Models and Social Inﬂuence .................................... 553
Motivated Processing and Bias Correction................................................ 554
Affect and Inﬂuence ...........................................................................555
Effects of Mood ................................................................................ 555
Fear Appeals ................................................................................... 556
Group and Self-Identity.......................................................................557
540 WOOD
Social Consensus and Validity of Information ............................................ 558
Multiple Motives Instigated by Groups..................................................... 559
Opinion Minority and Majority Groups.................................................. 560
Conclusion........................................................................................561
INTRODUCTION
This chapter reviews the research on attitude change from what traditionally have
been two separate areas of inquiry, the study of message-based persuasion and
the study of social inﬂuence. In the persuasion paradigm, inﬂuence appeals typically
include detailed argumentation that is presented to individual recipients in
a context with only minimal social interaction. Social inﬂuence appeals, in contrast,
usually consist solely of information about the source’s position, but these
are delivered in more complex social settings that may include interaction among
participants. Because of the marked continuities in the theoretical analyses and
in the empirical ﬁndings that have emerged across these research areas in the past
few years, this review draws from both literatures. It emphasizes in particular
research published during 1996 to 1998, since the prior review of Petty et al
(1997).
Giving social inﬂuence research a signiﬁcant role in the current review requires
that limited attention be given to some other research areas that have been featured
prominently in the past. Work on attitude structure and on attitude-behavior relations
has continued to ﬂourish, and Eagly & Chaiken (1998) provide an excellent
review elsewhere. Also noteworthy, despite some overlap with the current review,
are Petty & Wegener (1998a), Cialdini & Trost (1998), and Chaiken et al (1996b).
Another research area beyond the scope of this chapter is the extensive work on
intergroup attitudes and stereotypes (Brewer & Brown 1998, Fiske 1998).
MOTIVES FOR AGREEING WITH OTHERS
A hallmark of social inﬂuence research is the delineation of the multiple motives
that spur agreement or disagreement with others. For over 40 years, the central
organizing perspective in this area has been a dual-motive scheme that differentiates
between informational inﬂuence, which involves accepting information
obtained from others as evidence about reality, and normative inﬂuence, which
involves conformity with the positive expectations of ‘‘another,’’ who could be
‘‘another person, a group, or one’s self’’ (Deutsch & Gerard 1955:629).
Yet contemporary theories of motives for attitude change and resistance appear
to be converging instead on a tripartite distinction (e.g. Chaiken et al 1996a,
Cialdini & Trost 1998, Johnson & Eagly 1989, Wood 1999; for an early presentation
of this kind of framework, see Kelman 1958). Although these typologies
each possess unique features, a common thread is the recognition that attitude
ATTITUDE CHANGE 541
change can be motivated by normative concerns for (a) ensuring the coherence
and favorable evaluation of the self, and (b) ensuring satisfactory relations with
others given the rewards/punishments they can provide, along with an informational
concern for (c) understanding the entity or issue featured in inﬂuence
appeals. Thus, for example, Cialdini & Trost (1998) identify the behavioral goals
of social inﬂuence recipients as managing the self-concept, building and maintaining
relationships, and acting effectively. Similarly, Chaiken et al (1996a) distinguished
between people’s ego-defensive motives to achieve a valued, coherent
self-identity, impression-related motives to convey a particular impression to others,
and validity-seeking motives to accurately assess external reality.
Social inﬂuence researchers traditionally assumed that informational and normative
motives are each associated with unique mechanisms that generate attitude
change and with unique forms of change. The desire for an informed, correct
position supposedly orients message recipients to process the content of the appeal
and results in enduring private change in judgments. The desire to meet normative
expectations supposedly yields less informational analysis and public, contextdependent,
transitory judgment change. This view has been challenged by dualmode
processing models of persuasion (Eagly & Chaiken 1993, Petty & Wegener
1998a), especially by the demonstration that informational, accuracy-seeking
motives can lead either to extensive processing and enduring attitude change or
to more superﬁcial processing and temporary change. In the dual-mode framework,
motives for change are not preferentially related to change mechanisms or
outcomes.
Two recent studies support the persuasion analysis by providing evidence that
normative and informational motives affect inﬂuence through a common set of
information-processing mechanisms (Chen et al 1996, Lundgren & Prislin 1998).
Lundgren & Prislin (1998) found that, when participants were motivated to be
accurate, they selected arguments to read on both sides (i.e. pro and con) of the
target issue, generated thoughts that were relatively balanced in evaluation of
both sides, and indicated relatively neutral attitudes. In contrast, when participants
were motivated to convey a favorable impression to an interaction partner, they
selected arguments to read that were congruent with the view ostensibly held by
the partner and generated thoughts and attitudes that were congenial with their
partner’s position. Finally, when participants were motivated to defend their own
position, they selected arguments to read that supported their view, generated
thoughts supportive of their position, and indicated relatively polarized attitudes.
Furthermore, analyses to test mediation revealed that the favorability of participants’
thoughts (at least partially) mediated the effects of motives on attitude
change. Thus, it appears that the attitude effects emerged in part because accuracy
motives generated a relatively open-minded processing orientation, impression
motives generated an agreeable orientation, and defense motives generated a protective
orientation that maintained existing judgments.
These two studies also challenge the assumption that recipients’ motives are
associated with unique forms of attitude change. Regardless of the initial motive
542 WOOD
directing attitude judgments, the attitudes participants expressed to their ostensible
discussion partners persisted when participants later indicated their judgments
privately (Chen et al 1996, Lundgren & Prislin 1998). Especially
impressive is the persistence of attitudes designed to convey a favorable impression.
Contrary to classic theories of social inﬂuence, attitudes directed by
impression-related normative motives were no more ‘‘elastic’’ than were attitudes
directed by accuracy-seeking, informational motives. Instead, it seems that
impression and defense motives, much like the accuracy motives studied extensively
in message-based persuasion research, can yield careful, systematic processing
of relevant information that results in stable judgments. This ﬁnding
augments the results of earlier research in which impression motives were linked
to superﬁcial processing and temporary judgment shifts (e.g. Cialdini et al 1976).
The factors that determine whether motives instigate extensive or more superﬁcial
processing are discussed below.
Public Versus Private Inﬂuence
In social inﬂuence paradigms, researchers often have diagnosed the motive for
attitude change from the continuity of recipients’ judgments across public and
private settings. In public settings, recipients believe that the source of the appeal
or members of their experimental group have surveillance over their responses,
whereas in private settings, recipients believe that these others are unaware of
their judgments. Supposedly, attitudes that maintain across public and private
measures are internalized responses that result from the thoughtful processing
associated with accuracy motives, whereas attitudes that are expressed in public
but not private reﬂect normative pressures such as acceptance from the source or
group.
Recent empirical ﬁndings suggest instead that lack of continuity between public
and private judgments is not reliably diagnostic of recipients’ motives. As
described above, enduring attitude change is not the unique province of informational
motives; it also can emerge from social motives such as the desire to
accommodate others (Chen et al 1996, Lundgren & Prislin 1998). Evidence of
judgment stability across public and private settings has emerged also in inﬂuence
in the Asch-type judgment paradigm. In this research, participants are exposed to
others’ obviously incorrect judgments of line length and participants’ agreement
with others typically is interpreted as public, superﬁcial acquiescence to social
pressure. Yet the meta-analytic synthesis by Bond & Smith (1996) of 97 studies
using the Asch-type social inﬂuence paradigm revealed no greater agreement in
public than in private contexts of attitude expression. It seems, then, that social
motives for agreement affected attitudes in public as well as private settings.
The lack of systematic differences between public and private expressions of
judgment can be attributed to a number of factors. One is that recipients’ motives
for agreement can have relatively extended effects that generalize to new contexts
in which the original motives are no longer salient or relevant (e.g. Hardin &
ATTITUDE CHANGE 543
Higgins 1996, Higgins 1981, Ruscher & Duval 1998). Extended effects can occur
when the initial motivated judgment is retrieved in new settings or when the
information on which the judgment was based is retrieved, given that the motivated
processing yielded a biased representation of the original information. Thus,
because motives affect the judgments and the judgment-relevant information
available in memory, initial motivations for processing may have effects that
transcend context, and positions stated in public contexts may be maintained in
private. Kassin & Kiechel (1996) provide a compelling example of the extended
consequences of motivated processing. They simulated procedures sometimes
used in the interrogation of crime suspects by (falsely) accusing research participants
of an act of negligence while they were typing data into a computer. When
participants were subjectively uncertain about their innocence (because they were
typing at a fast speed), they accepted a witness’s report of their actions and (incorrectly)
confessed to the allegation. For the majority of participants, the confession
was not mere compliance. Over half reported in a subsequent discussion that they
had performed the negligent act, and over a third actually confabulated details in
support of the false allegation.
Furthermore, the distinction between public and private settings suggests an
overly simpliﬁed view of social impact, one that equates social presence with
surveillance. Allport’s (1985) famous deﬁnition of social psychology provided a
considerably more differentiated view of social impact, in which the effects of
others emerge whether their presence is ‘‘actual, imagined, or implied.’’ Because
important features of social impact may hold across public and private contexts,
attitudes that are affected by these features may also hold across settings. For
example, the manipulation by Baldwin & Holmes (1987) of social impact
involved simply instructing female participants to think about two of their older
relatives. The women were later given sexually explicit material to read in a
supposedly unrelated context, and they reported not liking it much. Presumably,
others’ conservative moral standards were activated in the initial manipulation
and continued to exert impact on subsequent experiences.
Theoretical perspectives need to progress beyond the simple distinction
between public and private attitude expression and consider whether the features
of social pressure that are relevant to attitude change are stable across settings.
For example, in a meta-analytic synthesis of the minority-inﬂuence literature
(Wood et al 1994), the inﬂuence of opinion-minority, low-consensus sources
proved comparable in public and private settings. Thus, it seemed that attitude
change was not controlled by surveillance and the fear that aligning with a deviant
minority source in public would lead to social embarrassment and rejection by
others. Agreement did vary, however, with another feature of the inﬂuence context;
how directly attitudes were measured. ‘‘Direct’’ measures assess attitudes
on the issue in the appeal, and recipients are aware that their (public or private)
judgment can align them with the source’s position. ‘‘Indirect’’ measures might,
for example, assess attitudes on issues tangentially related to the appeal, and
recipients are less aware that their judgments can align them with the inﬂuence
544 WOOD
source. Minority impact was smaller on direct than on indirect measures. Wood
et al (1994) concluded that recipients’ resistance on direct measures is due to
their own personal knowledge that their judgments could align them with a deviant
minority source. It seems, then, that minority inﬂuence was inhibited by recipients’
concern for the favorability and integrity of their self-concept and their
place in their reference group, and that these motives held in both public and
private contexts (see below).
The current analysis of attitude expressions in public and private contexts also
calls into question the common assumption that when public and private judgments
differ in accuracy, privately expressed ones are generally more trustworthy
because public expressions may be biased to achieve social motives. Although
some features of public contexts (e.g. politeness norms) may compromise the
accuracy of attitude expressions, other features appear to enhance thoughtful analyses
and sometimes to increase accuracy. Cowan & Hodge (1996) demonstrated
that to the extent public contexts enhance perceived accountability for judgments,
people give especially thoughtful, reasoned responses in public. Similarly, Lambert
et al (1996) argued that the expectation of public discussion focuses people
on their own attitudes and encourages them to bolster their beliefs; thus, attitudes
were found to play a greater role in guiding thought and action in public than in
private settings.
Finally, given that in social-inﬂuence paradigms respondents often give their
judgments ﬁrst publicly and then again privately, continuity across judgment contexts
can emerge from the effects of initial judgments on subsequent ones.
Research on the impact of behavior on attitudes has demonstrated that people’s
interpretations of their public statements and other attitude-relevant behaviors can
instigate shifts in privately held attitudes to correspond to public acts (see Chaiken
et al 1996b). This research also has demonstrated that public-attitude statements
that are of questionable veracity (e.g. when a public statement is given with low
choice or high reward) do not affect the attitudes expressed in private settings
(see below). However, Maio & Olson (1998) provide intriguing evidence that
even under low-choice conditions, the act of providing an attitude judgment can
enhance the accessibility of one’s own attitude in memory; accessible attitudes
then may affect subsequent attitude-relevant judgments in seemingly unrelated
contexts.
MOTIVES IN PERSUASION RESEARCH
Functional Theories
The motives underlying attitude change in message-based persuasion paradigms
have been investigated primarily in research on attitude functions (see Eagly &
Chaiken 1998). In addition to the basic adaptive function of enabling people to
evaluate and appraise stimuli in their environment, attitudes also are thought to
serve more speciﬁc functions. Functions identiﬁed in early work include securing
ATTITUDE CHANGE 545
utilitarian outcomes, ego defense, value expression, and social adjustment (Katz
1960, Smith et al 1956). These functions are reﬂected in the tripartite motive
scheme suggested above: Accuracy motives correspond generally to a utilitarian
concern with maximizing rewards and minimizing punishments, self-concept
motives correspond to concerns for defending the ego against potential threats
and for expressing one’s values, and social relation motives correspond to concerns
for social adjustment and for obtaining social rewards and avoiding social
punishments.
In one account of the role of attitude functions in inﬂuence, persuasive attempts
are likely to be effective to the extent that the function of, or reason for holding,
the position outlined in the appeal matches the function underlying recipients’
attitudes (Lavine & Snyder 1996, Murray et al 1996: Study 2). For example,
Lavine & Snyder (1996) reported that for people who are generally sensitive to
the social consequences of their behavior (i.e. high self-monitors), appeals that
emphasized the social adjustive functions of voting (e.g. enhancing one’s attractiveness
to others) elicited more favorable evaluations and greater attitude change
than appeals that emphasized its value-expressive functions (e.g. a way to express
values). For people who rely on inner dispositions (i.e. low self-monitors),appeals
with value-expressive arguments yielded more favorable evaluations and were
more persuasive. Furthermore, certain issues may be associated with certain attitude
functions for most people. Although not speciﬁcally couched within a functional
framework, the analysis by Rothman & Salovey (1997) of health-related
messages suggested that inﬂuence is greatest when the orientation of an appeal
matches the orientation intrinsic to the health issue itself. Recommendations to
perform illness-detecting behaviors (e.g. breast self-exams) potentially incur risk
and thus loss-framed appeals are likely to be effective, whereas recommendations
for preventative behaviors (e.g. exercise) potentially incur positive outcomes and
thus gain-framed appeals are likely to be effective.
In another account, matching between attitude function and message orientation
does not always enhance persuasion but instead enhances careful thought
about an appeal. Petty & Wegener (1998b) demonstrated that matched messages
increased scrutiny of message content but enhanced persuasion only when the
message contained strong, cogent arguments and not when it contained weak
arguments. Yet because functionally matched messages potentially address important
aspects of recipients’ self-concepts, this careful processing will not always
be objective and unbiased. Such appeals may instigate a thoughtful but defensive
orientation, as recipients try to maintain valued aspects of the self. For example,
Tykocinski et al (1994) reasoned that messages framed to match people’s current
experiences and concerns can elicit distress by identifying seemingly relevant
goals that have not been adopted. Thus, such messages are especially likely to
yield counterarguing and resistance. Similarly, Marsh et al (1997) reported that
persuasive messages that address an important attitude function (i.e. for college
students, the value-relevant issue of sororities/fraternities on college campuses)
are processed carefully yet defensively and as a result are minimally inﬂuential.
546 WOOD
Cognitive Dissonance Theory
This classic motivational theory of how attitudes change to maintain cognitive
consistency continues to spark interest. The original notion of Festinger (1957)
that dissonance arises from psychological inconsistency between linked cognitions
has been modiﬁed extensively in subsequent research. In Cooper & Fazio’s
(1984) ‘‘new look’’ approach, dissonance arises not from simple inconsistency
but rather from freely chosen behavior that brings about some foreseeable negative
consequence.
A central question for dissonance researchers has been the motivational bases
for dissonance and the cause of the aversive state of dissonance arousal. In Aronson’s
(1992) self-concept analysis, dissonance arises from inconsistent cognitions
that threaten the consistency, stability, predictability, competence, or moral goodness
of the self-concept. In Steele’s (1988) self-afﬁrmation theory, dissonance
arises from the violation of general self-integrity. From these self-related perspectives,
negative consequences are powerful inducers of dissonance because it
is inconsistent with most people’s self-views to act in a way that results in foreseeable
aversive consequences. An alternate perspective on dissonance arousal,
which has yet to be integrated into mainstream theorizing, is the argument by
Joule & Beauvois (1998) that dissonance reduction is oriented toward rationalizing
behavior rather than attaining psychological consistency.
In an interesting integration that recognizes that both self-concept threat and
aversive consequences can instigate dissonance, Stone & Cooper (see Petty &
Wegener 1998a) proposed that dissonance arises when people fail to behave in a
manner consistent with some valued self-standard. The speciﬁc motivation behind
dissonance supposedly depends on the type of self-standard involved. Dissonance
can emerge from behavior that is inconsistent with personal self-standards and
does not reﬂect the way people want to be (ideal self) or think they should be
(ought self), or dissonance can emerge from behavior that generates aversive
consequences and does not reﬂect how others want them to be (normative self-
standards).
Several studies support the conclusion that dissonance motivation can emerge
in contexts devoid of negative consequences. Participants in a study by HarmonJones
et al (1996) freely engaged in the nonconsequential behavior of privately
taking a counterattitudinal position, yet they experienced increased arousal and
attitude shifts toward their expressed position. Similarly, Prislin & Pool (1996)
found little evidence that dissonance arises only when behavior has identiﬁably
negative consequences and instead concluded that dissonance emerges when
behavior and its consequences challenge existing ideas about the self.
The hypocritical advocacy paradigm was developed to study dissonance motivation
in the absence of negative consequences. In this research, participants
advocate a proattitudinal position, are made aware of their past failures to act in
accordance with this position, and (in order to reduce dissonance) then engage in
acts congruent with the position (e.g. Fried 1998, Stone et al 1997). Although the
ATTITUDE CHANGE 547
lack of immediate negative consequences to this proattitudinal advocacy might
appear to indicate that such consequences are not critical for producing dissonance,
proponents of the negative consequences view can rightly note that highlighting
past failures to uphold one’s stated beliefs identiﬁes aversive
consequences of past acts. However, Fried & Aronson (1995) argued for the
importance of the self in instigating dissonance by noting that dissonance in this
paradigm does not emerge when people are only reminded of past transgressions
(i.e. negative consequences) and do not engage in proattitudinal advocacy and
experience the self-related implications of this behavior.
The second major issue addressed in dissonance research is the multiple routes
or modes through which dissonance can be reduced. Although self-afﬁrming
behavior that reestablishes personal integrity has been shown to reduce dissonance
(Steele 1988), self-afﬁrmations are not always the mode of choice. When multiple
routes are available, people apparently prefer to reduce dissonance directly by
changing attitudes and behaviors (i.e. modifying the inconsistent cognitions)
rather than to alleviate it indirectly through self-afﬁrmations (Stone et al 1997).
Even people with high self-esteem, who should possess the resources to reduce
dissonance by focusing on positive aspects of the self-concept, have been found
instead to modify cognitions (Gibbons et al 1997). Other research has identiﬁed
boundary conditions for the usefulness of self-afﬁrmations. According to Blanton
et al (1997), afﬁrmations do not reduce dissonance if they remind people of the
violated self-standard (e.g. reassurance of one’s compassion when one has acted
in a noncompassionate way), presumably because such afﬁrmations make it more
difﬁcult to justify the dissonance-inducing act and lead one to dwell on the dissonant
behavior.
Other research on modes of dissonance reduction has revealed individual differences
in the route of choice. People who are high in attributional complexity
and characteristically search for abstract, complex explanations appear to reduce
the dissonance caused by counterattitudinal advocacy by considering possible
external justiﬁcations for the attitude-discrepant act rather than by changing attitudes
(Stadler & Baron 1998). In addition, suggesting that research has only begun
to tap the variety of modes available, Burris et al (1997) documented the
dissonance-reducing effects of transcendence (reconciling inconsistent beliefs
under a broader principle) and reafﬁrmation of the attacked belief. A recognition
of the full range of routes available for dissonance reduction can help to account
for some of the seeming inconsistencies in the dissonance literature. For example,
when Fried (1998) modiﬁed the hypocrisy paradigm to make public participants’
past failures to live up to their attitudes, participants did not perform attitudecongruent
compensating behaviors to reduce dissonance. Instead, they decreased
the dissonance caused by public transgressions by changing their attitudes to be
congruent with the behavioral transgression.
A new perspective worth watching is the development of formal mathematical
models of dissonance-related processes. Parallel constraint satisfaction systems
represent dissonance as a dynamic, holistic process and provide a means to eval-
548 WOOD
uate consistency and other aspects of the relation between beliefs (Read et al
1997). In one application of this approach, researchers specify a network of attitudes
and other cognitions, simulate changes in the network (according to prespeciﬁed
rules of how the attitudes/cognitions are related to each other) until it
reaches a state of overall consistency, and compare the results of the simulation
to data from research participants. Shultz & Lepper (1996) were able to use this
technique to successfully account for judgment change in several classic dissonance
paradigms.
MULTIPLE ATTITUDES
Sources of Multiple Attitudes
A central assumption of much attitude theorizing is that people’s evaluations of
a given object are stable across time, context, and form of assessment. Empirical
evidence of this coherence has emerged with increasing clarity in recent years,
in part due to improved understanding of the determinants of coherence in attitudinal
responses [e.g. strong attitudes (Petty & Krosnick 1995)] and in part to
improved methods to document coherence [e.g. compatibility in measurement
(Ajzen 1996)]. However, empirical evidence that people can hold multiple attitudes
toward a given object is emerging as well, and this proﬁle of dissociation
has been explained in a number of ways.
The evidence for multiple attitudes has sometimes been dismissed as reﬂecting
epiphenomena (e.g. context effects, ‘‘nonattitudes’’ or weakly held attitudes). Yet
multiple attitudes also can stem from more enduring effects. For example, attitudes
that vary with context can represent temporary constructions (e.g. differential
use of rating scales, anchoring, and adjustment effects) or more internalized
tendencies to respond that are stably linked to certain contextual features
(McConnell et al 1997). In addition, multiple evaluations of an attitude object
can emerge from attitude structure. They can reﬂect the superﬁcial responses
associated with attitudes that have minimal cognitive and affective bases or the
more stable reactions associated with attitudes that have inconsistent components,
as when attitudes are structurally inconsistent (e.g. Chaiken et al 1995, Prislin et
al 1998), held with ambivalence (Priester & Petty 1996, Thompson et al 1995),
or associated with varied affective responses [e.g. immediate versus anticipated
future affect (Richard et al 1996a,b, Van der Pligt et al 1998)].
In recent years, multiple attitudes have begun to generate interest in their own
right, as increased theoretical understanding provides a basis for predicting both
the coherence and the dissociation that occurs in attitude judgments (e.g. Mackie
& Smith 1998, Wilson & Hodges 1992). This is a still-developing research area
that encompasses a somewhat diverse set of effects, including context-dependent
attitudes, multiply categorizable attitude objects, explicit versus implicit attitudes,
subjective construals, and issue framing. The common theme linking these various
ATTITUDE CHANGE 549
research areas is that people’s responses to a particular attitude object can reﬂect
diverse evaluations, cognitive representations and interpretations, and affective
reactions.
One source of dissociations is the variety of motivations that can underlie
attitudes. Dissociation is implied in the claim of social inﬂuence theorists that
people are motivated to adopt attitudes of relevant reference groups to the extent
that the group identity is salient and desirable (Kelman 1958, Turner 1991). That
is, people may possess multiple cognitive representations of attitudes on an issue
that is important to more than one of the groups with which they identify. Preliminary
support for this idea was provided in ﬁndings by Wood & Matz (unpublished
data) that college students’ attitudes toward welfare programs were more favorable
when their social identities as religious people were salient (relative to a
base-line attitude measure obtained with no salient group identity), whereas students’
attitudes were less favorable when their identities as political conservatives
were salient. Furthermore, these attitude shifts emerged most strongly among
participants who considered religious or conservative social groups to be self-
deﬁning.
Dissociations also can arise from cognitive processes. To the extent that an
attitude object (e.g. yogurt) is relevant to a diverse set of issues and values, people
can select from multiple categories (e.g. dairy products, health foods) when construing
the object. A number of factors determine the category selected, including
the accessibility of category attitudes (ER Smith et al 1996). Furthermore, a single
attitude object may generate multiple representations in memory when attituderelevant
information is stored separately from overall evaluations of the object
(Hastie & Park 1986). Indeed, McConnell et al (1997) demonstrated that different
attitudes can be generated in different contexts for a single attitude object when
perceivers do not attempt, on-line, to form an integrated attitude. Multiple cognitive
representations of an issue also can emerge from information processing
in social inﬂuence settings. After being exposed to the judgments of others in an
inﬂuence appeal, recipients may later retrieve this information without recognizing
its source and, under some circumstances, unwittingly adopt it as their own
response (Betz et al 1996).
Dissociations also can arise as by-products of the variety of processes through
which attitude-relevant judgments are generated. Explicit, conscious judgments
differ in a number of ways from judgments that are implicitly held, including the
kinds of information considered (Ajzen 1996, Greenwald & Banaji 1995). For
example, Wittenbrink et al (1997) suggest that explicit and implicit measures of
racial prejudice are only moderately correlated because people are more likely to
base explicit judgments on an egalitarian ideology; thus explicit attitudes toward
racial minorities are more favorable than implicit ones.
Inﬂuence and Multiple Attitudes
Inﬂuence strategies have capitalized on the multiple cognitive representations,
affective reactions, and evaluations that people can hold concerning a given
object. Asch (1940) argued early on that the primary process in inﬂuence is not
550 WOOD
change in attitudes toward an object but rather change in the deﬁnition and meaning
of the object. When meaning changes, attitudes change accordingly. The link
between meaning and evaluation is suggested in the ﬁnding by Bosveld et al
(1997) that people are more favorable toward ‘‘afﬁrmative action’’ when others
claim that it refers to equal opportunity rather than, for example, reverse discrimination.
Even subtle aspects of the way an issue is framed or represented in an
appeal, such as the apparent location and time at which a proposal will take place,
can affect recipients’ attitudes (Liberman & Chaiken 1996). This latter ﬁnding
raises troubling questions for persuasion research that has varied personal relevance
through the supposed time or location of a message proposal (e.g. instituting
senior comprehensive exams at one’s own university or another university).
Typically, it has been assumed that recipients are responding to comparable attitude
objects regardless of whether the time and location of the message proposal
renders it personally relevant.
Issue framing is an inﬂuence strategy that capitalizes on multiple attitudes. In
this approach, the importance and relevance of certain consequences or attributes
of an issue are emphasized over other potential consequences (Ball-Rokeach &
Loges 1996, Nelson & Kinder 1996, Nelson et al 1997, Price & Tewksbury 1997).
For example, Eagly & Kulesa (1997) illustrate how persuasive appeals from both
sides of environmental debates have framed desired positions as achieving values
favored by recipients (e.g. a healthy economy based on tourism or based on
logging and industry) and framed the nondesired position as achieving values
most recipients reject.
Readers may wonder how appeals that use framing differ from ones that use
the belief change strategies typical of message-based persuasion paradigms. Standard
persuasion appeals typically address intraattitudinal relations and describe
the speciﬁc attributes possessed by an attitude object. In contrast, as implied in
the label, framing appeals typically address interattitudinal relations and place an
issue or object in the context of other attitude issues, values, and goals. Furthermore,
framing appears to highlight the relevance or importance of existing knowledge
structures and values. Nelson et al (1997) reported that framing effects are
typically stronger for participants familiar with an issue (i.e. possessing the relevant
knowledge structures) than for those unfamiliar with it. Despite these unique
features of framing, established models of attitude change seem appropriate for
understanding its effects. For example, expectancy-value formulations can
account for framing effects through variations in the salience and likelihood of
relevant goals and values, as well as through variations in the evaluation given
to them (Ajzen 1996).
The mechanisms underlying framing effects probably depend, as in standard
persuasion paradigms, on the extent to which recipients are motivated and able
to process message content and other relevant information (see discussion of the
dual-mode information processing models in next section). For example, the
impetus to adopt a new interpretation or frame for an issue can arise from motivational
goals such as rejecting a strongly disliked group identity. In a demon-
ATTITUDE CHANGE 551
stration of this process, college students informed that their attitudes corresponded
to those of a hated group, the Ku Klux Klan, shifted their own interpretations of
the issue away from a seemingly racist construal; they then were able to shift
their attitudes away from those of the Klan (Wood et al 1996, Pool et al 1998).
DUAL-MODE PROCESSING MODELS OF PERSUASION
Persuasion research has continued in the highly successful tradition of the dualmode
processing theories, the heuristic/systematic model (Chaiken et al 1996a)
and the elaboration likelihood model (Petty & Wegener 1998a). The central tenet
of these theories is that the determinants and processes of attitude change depend
on people’s motivation and ability to process issue-relevant information. When
people are not highly motivated (e.g. the issue is not personally involving) or
they have low ability (e.g. they are distracted), attitude judgments are based on
easily available attributes of a source, message, or situation that are evaluated via
efﬁcient processing strategies. For example, they might use the heuristic rule,
consensus implies correctness, and thus agree with a majority position. When
people are both motivated and able to process information carefully, then attitudes
are based on a more thoughtful, systematic assessment of relevant information.
According to the elaboration likelihood model, such high-elaboration processes
include learning message content, generation of cognitive responses, and
dissonance-induced reasoning (Petty & Wegener 1998a). For evaluations of the
similarities and differences between the heuristic/systematic model and the elaboration
likelihood model, see Chaiken et al (1996a), Eagly & Chaiken (1993),
and Petty & Wegener (1998a).
Research has continued apace identifying the factors that enhance systematic,
thoughtful processing. According to the heuristic/systematic model, people are
motivated to engage in systematic thought in order to achieve a sufﬁcient ‘‘desired
level of conﬁdence’’ in their judgments. Factors that have been found to increase
systematic processing (presumably by decreasing actual conﬁdence or increasing
desired conﬁdence) include the following: framing of persuasive messages in an
unexpected manner (Smith & Petty 1996); self-relevance of messages, either
because recipients self-reference or are made self-aware (Turco 1996); and use
in messages of token phrases that ambiguously signal broader values, such as,
for Democrats, ‘‘family values’’ rhetoric (Garst & Bodenhausen 1996). Systematic
processing also has been found when recipients hold strong, accessible attitudes
on the message topic (Fabrigar et al 1998), hold ambivalent attitudes (Maio
et al 1996), or enjoy effortful cognitive activity (i.e. are high in need for cognition;
for review, see Cacioppo et al 1996). In addition, recipients engage in systematic
processing in circumstances in which careful thought is likely to generate judgment
conﬁdence, such as when recipients believe in their own efﬁcacy and ability
to evaluate (Bohner et al 1998b) and when the message is presented in accessible,
not overly complex language (Hafer et al 1996).
552 WOOD
Systematic processing also may be implicated in resistance to inﬂuence. Pfau’s
(1997) insightful review of resistance research focused in particular on inoculation
procedures, in which recipients receive information that strengthens their attitudes
before exposure to persuasive attack. He argues that inoculation effects emerge
when anticipated threats (e.g. warning of a potential attack to one’s attitudes)
motivate thoughtful processing to support one’s own position or to counter opposing
ones. Furthermore, resistance to attitude change appears to increase at midlife
(Visser & Krosnick 1998). Middle-aged people appear to have especially strong
attitudes that enable them to counterargue opposing positions.
Heuristic processing is used when a low-effort strategy yields attitudes with a
sufﬁcient level of conﬁdence (Chaiken et al 1996a). A variety of heuristic cues
have been identiﬁed in recent research. For example, familiar sayings can provide
cues to agreement (Howard 1997), and the subjective experience of thinking about
an issue can be a cue, so that, for example, people adopt positions when they can
easily generate supportive arguments and reject positions when they cannot (Rothman
& Schwarz 1998). In addition, relevant to understanding the heuristic cues
used in everyday contexts are Dickerson’s (1997) observations of politicians bolstering
their favored positions with references to unbiased, expert sources and
sources with an apparent allegiance to the opposition.
Cognitive Response Mediation of Attitude Change
Research has continued to address the mechanics underlying systematic, highelaboration
processing. One question is whether the valence of recipients’ cognitive
responses mediates persuasion or whether valence represents either an
alternate measure of attitude change or an after-the-fact justiﬁcation of change.
To test causal precedence, Romero et al (1996) and Killeya & Johnson (1998)
directly manipulated thoughts and examined the effects on attitudes. Consistent
with the perspective in which thoughts mediate change, acceptance of the position
in the appeal corresponded to the extent and valence of the generated thoughts.
Furthermore, correlational analyses in these studies, as well as in several studies
that assessed recipients’ spontaneously generated thoughts (e.g. Friedrich et al
1996, Hafer et al 1996), proved consistent with the mediational role of thoughts.
Conclusions concerning causal order are complicated, however, by the few
studies that reordered the typical causal sequence tested in correlational designs
and found that attitudes can successfully mediate the effects of independent variables
on thoughts (Friedrich et al 1996, Maio & Olson 1998). Yet the evidence
of mutual mediation through attitudes as well as thoughts does not necessarily
represent a challenge to the cognitive response model. Mutual mediation could
reﬂect the simultaneous use of dual processing modes. People may engage simultaneously
in effortful processing, in which an appeal instigates thoughts that then
affect attitudes, along with less effortful processing, in which attitudes are directly
affected by an appeal through, for example, heuristic analyses (Chaiken et al
ATTITUDE CHANGE 553
1996a). When appeals have a direct effect on attitudes, then thoughts may reﬂect
an after-the-fact justiﬁcation for attitude judgments.
Another question is whether cognitive responses provide a sufﬁcient model of
mediation. Munro & Ditto (1997) and Zuwerink & Devine (1996) report that
attitude change on prejudice-related issues is linked to message recipients’ subjective
experience of affect in addition to the favorability of their thoughts.
Although it is possible that the insufﬁciency of thoughts as a mediator stems from
measurement limitations (e.g. poor reliability or validity of coding of thought
protocols), it also is possible that for prejudice and other attitudes with a strong
affective basis, changes in affective reactions impact attitudes independently of
cognitive responses.
Dual-Mode Processing Models and Social Inﬂuence
Dual-mode models also can provide a framework for understanding attitude
change in social inﬂuence settings. For example, careful scrutiny of other group
members’ answers to a judgment task appears to depend on participants’ motivation
to perform well and their ability to conduct their own evaluations of the
task. Baron et al (1996) reported that highly motivated participants relied on their
own evaluations except when judgment stimuli were presented too brieﬂy to
identify the correct answer; then they appeared to adopt the heuristic strategy of
relying on others’ judgments. In contrast, participants who were only moderately
motivated used the heuristic-like strategy of relying on others’ estimates regardless
of whether they could determine the correct answer themselves. The heuristic
cues important in social inﬂuence settings include not only others’ judgments,
but also aspects of social interaction and others’ self-presentation. For example,
group members who are more conﬁdent in their judgments have been found to
be more inﬂuential in discussions, regardless of their actual task accuracy (Zarnoth
& Sniezek 1997).
The group discussion research by Kelly et al (1997) also can be interpreted
from a dual-mode perspective. When members were motivated and able to
achieve accurate solutions (i.e. the task had a seemingly correct solution, time
was sufﬁcient), group discussions entailed considerable reasoning and argumentation.
Furthermore, systematic reasoning was apparently successful in yielding
valid solutions; more systematic thought during discussion was associated with
greater solution accuracy. Also interpretable from this perspective is the ﬁnding
by Shestowsky et al (1998) that in dyadic discussions, participants who were
motivated to engage in cognitive activities (i.e. were high in need for cognition)
had the greatest impact on group decisions. Although no direct evidence was
obtained that motivated participants engaged in careful analysis during the discussion
or that this is what made them inﬂuential, participants’ self-ratings suggested
that those who valued thinking activities presented many arguments,
presented valid arguments, and tried hard to be persuasive.
554 WOOD
Motivated Processing and Bias Correction
Research has continued to document how recipients’ motives (i.e. to defend self,
maintain desired relations with others, have accurate judgements) instigate and
direct systematic processing and yield more favorable evaluations of goalpromoting
than hindering information (Munro & Ditto 1997, Zuwerink & Devine
1996). Motives also can yield selective use of heuristic cues, such as relying on
social consensus when it provides adequate support for one’s desired position
(Giner-Sorolla & Chaiken 1997).
A variety of factors appear to motivate biased processing, including recipients’
broader values (Eagly & Kulesa 1997, Maio & Olson 1998, Seligman et al 1996)
and self-interests (Giner-Sorolla & Chaiken 1997), and attitude issues that are
highly important and involving (Zuwerink & Devine 1996). However, the role of
ability in biased processing remains unclear. Although some have speculated that
bias emerges when people rely on their (presumably predominantly attitudesupportive)
personal beliefs and knowledge to evaluate an issue, Biek et al (1996)
found knowledge alone to be insufﬁcient; thoughtful biased processing emerged
only when knowledgeable people were also highly motivated (e.g. by strong
affect) to hold a particular position.
Asymmetries also have been noted in motivated processing. People sometimes
respond more intensely to threatening information that disconﬁrms their desired
view than to congenial information that conﬁrms it (see also Cacioppo et al 1997).
Ditto et al (1998) speculate that because threatening stimuli are likely to require
an immediate behavioral response, it is adaptive for people to conduct an objective,
critical analysis of preference-inconsistent information while responding
more passively to congenial information (see also Edwards & Smith 1996). It
seems likely, however, that people use a variety of processing strategies to meet
defensive goals. For example, a strategy of defensive inattention to challenging
information can explain the ﬁnding by Slater & Rouner (1996) that people more
carefully process congenial than threatening health-related information (see
below).
An interesting question that has emerged in the past few years is whether
people are aware of and can counteract biases and shortcuts in their information
processing strategies. Wegener & Petty (1997) suggest that people engage in bias
correction processes to the extent that they believe factors unrelated to the true
qualities of the attitude issue have inﬂuenced their judgments and to the extent
that they are motivated and able to counteract the bias. Although an implicit
assumption seems to be that validity-seeking motives instigate bias correction,
other motives also are plausible (e.g. the self-related motive of being an objective,
impartial judge, the other-related motive of conveying this impression to others).
In a test of bias correction, Petty et al (1998) simulated a context in which it
would be illegitimate to rely on heuristic rules. Participants were instructed to
avoid letting their biases about a seemingly likable or unlikable source inﬂuence
ATTITUDE CHANGE 555
their judgments of the source’s proposal. The result was an apparent overcorrection
in which the dislikable source was more persuasive than the likable
one. Furthermore, because the correction instructions appeared to affect processing
independently of participants’ motivation to scrutinize the message, Petty et
al (1998) concluded that careful message scrutiny does not spontaneously include
attempts to counteract potential biases.
Bias correction also emerges in jury trials when the evidence presented is
subsequently ruled inadmissible by the court. In experimental trial simulations,
whether such evidence is discounted appears to depend in part on the reason for
it being inadmissible. Evidence excluded because it was unreliable or because it
was presented for some personal motive had little effect on judgments, whereas
evidence excluded for procedural reasons continued to exert impact (Fein et al
1997, Kassin & Sommers 1997). In addition, consistent with Wegener & Petty’s
(1997) model, people attempting to compensate for inadmissible evidence have
been found to both over- and undercorrect, depending on whether they believe
the evidence was likely to have a strong or a weak impact on their judgments
(Schul & Goren 1997).
AFFECT AND INFLUENCE
Effects of Mood
Several models have been developed to explain the effects of mood on information
processing and attitude change. According to Wegener & Petty (1996),
mood effects vary with elaboration likelihood. Direct effects of mood on agreement
emerge through low-elaboration processes, including association of a persuasive
appeal with positive or negative feelings (e.g. classical conditioning) and
use of heuristic rules based on those feelings (e.g. ‘‘I feel bad so I must dislike
it’’). When people are more extensively processing, how an attitude object makes
them feel can serve as a persuasive argument. It also can bias the information
considered, such as when people attend more to messages that match their mood
or when they recall such information more accurately (Rusting 1998). When
amount of elaboration is at some middle level, people respond strategically to
‘‘manage’’ moods. Happy people selectively process in order to maintain their
positive mood (e.g. attending to information that makes them feel good), whereas
sad people are less selective because there is greater potential for any activity to
be mood enhancing.
Alternatively, in the feelings-as-information account, moods signal appropriate
processing strategies (Bless et al 1996, Schwarz 1997, Schwarz & Clore 1996).
Similar to the discussion in the prior section concerning asymmetrical effects of
motives on processing, positive moods suggest a benign environment appropriate
for heuristic strategies, whereas negative moods indicate a potential problem that
may require systematic evaluation. Yet all negative emotions do not appear to
556 WOOD
have the same processing implications. Ottati et al (1997) found that anger (marginally)
reduced systematic processing relative to neutral moods, perhaps because
anger implies agonistic contexts requiring quick response.
Controversy has emerged concerning whether negative affect enhances processing
through mood management, as suggested by the elaboration model of
Wegener & Petty (1996), or through signaling a problematic situation, as suggested
by the feelings-as-information account of Schwarz & Clore (1996). At
present, the available empirical data can be interpreted as supporting either perspectives.
In addition, research ﬁndings have been taken to support a third, affectinfusion
model (Forgas 1995), in which affect infuses thoughts and behaviors
primarily when people engage in systematic, substantive processing. Although
Forgas (1998) concluded that empirical tests support affect infusion, the ﬁnding
that moods bias and direct systematic processing of inﬂuence appeals also is
consistent with the other two models considered here.
Fear Appeals
The effects of fear-inducing appeals have been of particular interest in the health
domain. Several theories have outlined the proximal beliefs through which fearinducing
appeals affect inﬂuence. According to protection motivation theory
(Rogers 1983), appeals that are threatening and that offer an effective means of
coping with the threat instigate danger control processes, which include accepting
the recommended coping strategy and changing the maladaptive behavior
(Prentice-Dunn et al 1997, Sturges & Rogers 1996). Rogers & Prentice-Dunn
(1997) concluded that about half of the studies to date that have performed appropriate
tests have supported the theory’s prediction of maximum acceptance when
perceived threat and coping are both high.
The extended parallel process model (Witte 1992, 1998) has linked inﬂuence
to fear-control as well as danger-control responses. When threat is greater than
coping, fear reactions can instigate message rejection through defensive responses
(McMahan et al 1998, Witte et al 1998). Consistent with this view, Aspinwall &
Brunhart (1996) reported that people who are not very optimistic about their own
health coping strategies are less likely to attend to threatening health information
than those who are more optimistic. Because fear-control responses inhibit the
adoption of self-protective acts suggested in the appeal, they can account for the
sometimes obtained ‘‘boomerang’’ shifts in attitudes away from messages when
people do not believe they can cope effectively (Rogers & Prentice-Dunn 1997).
Research also has continued to investigate the effects of fear on message processing
and to integrate the study of fear appeals into the broader models of mood
and inﬂuence covered above. Fear, like other affective responses, appears to
impact extent of processing. At low-to-moderate levels, fear functions like personal
involvement and increases processing (Rogers & Prentice-Dunn 1997).
Increased processing (at least of strong messages) may explain the often-obtained
ﬁnding that fear facilitates inﬂuence and acceptance of new coping strategies (e.g.
ATTITUDE CHANGE 557
Dillard et al 1996, Millar & Millar 1996). But fear also can bias processing in a
way that justiﬁes existing coping (or noncoping) behaviors (Biek et al 1996). At
high levels, fear appears to reduce systematic processing (e.g. Jepson & Chaiken
1990). In general, the nonlinear effects of fear on processing are compatible with
the idea that fear is multidimensional and involves both arousal and negative
affect. Measurement scales that have been designed to tap the activation component
have found that greater reported fear (i.e. arousal) enhances inﬂuence,
whereas scales that have been designed to tap the negative tension component
have found that greater fear (i.e. negative affect) inhibits inﬂuence or has minimal
effect (Celuch et al 1998, LaTour & Rotfeld 1997).
GROUP AND SELF-IDENTITY
Social identity theory (Tajfel 1981, 1982) has sparked considerable interest in
group inﬂuence as well as in other aspects of group behavior (see Brewer &
Brown 1998). In the social identity view, when people categorize themselves as
an ingroup member, the ingroup serves as a reference for social comparison, and
people adopt the prototypic ingroup attitudes and beliefs as their own. Building
on this analysis, Turner (1982, 1991) proposed that groups exert inﬂuence through
a speciﬁc process, which he called referent informational inﬂuence. In this view,
agreement from others categorized as similar to self enhances one’s subjective
certainty and suggests that the shared attitudes reﬂect external reality and the
objective truth of the issue. Disagreement from similar others yields subjective
uncertainty and motivates people to address the discrepancy through, for example,
mutual social inﬂuence or attributional reasoning to explain the disagreement.
Empirical data from a variety of research paradigms are congruent with the
social identity approach. Content analyses by Reicher & Hopkins (1996a,b) of
political and social speeches illustrate the central role of ingroup and outgroup
deﬁnitions in everyday persuasion. For example, their analysis of an antiabortion
speech delivered to medical doctors revealed that it deﬁned doctors and antiabortion
activists as a common ingroup in their concern for others’ welfare,
deﬁned antiabortion activity as consonant with the medical identity, and deﬁned
abortion-rights proponents as a derogated outgroup. Other work has supported
the social identity claim that inﬂuence stems from prototypic group attitudes. In
the small group discussions examined by Kameda et al (1997), ﬁnal decisions
were inﬂuenced most strongly by ‘‘cognitively central’’ group members, whose
initial beliefs about the discussion topic overlapped the most with other members.
The inﬂuence advantage of prototypic members was independent of whether their
initial judgment preferences placed them in the group majority or minority.
Kameda et al (1997) argued that the shared beliefs and knowledge provided social
validation for other members’ views and a basis on which others could recognize
prototypic members’ expertise.
558 WOOD
The self-categorization analysis differs from standard persuasion models in
locating the determinants of attitude change in people’s construction of group
identity rather than in their understanding of attitude issues. One kind of evidence
presented in support of this analysis is that inﬂuence varies with group membership
(e.g. Haslam et al 1995). Other evidence is that the inﬂuence of a group does
not depend on recipients learning the content of the inﬂuence appeal (McGarty
et al 1994; see also Haslam et al 1996). Although some research has suggested
that ingroup inﬂuence is accompanied by acceptance and learning of the message
(e.g. Mackie et al 1992), Haslam et al (1996) argue that such learning occurs after
adoption of the ingroup position, as people try to understand the group view in
order to be an effective group member.
The self-categorization theory claim that group identity can have a direct effect
on inﬂuence is compatible with a heuristic-like reasoning process in which people
are persuaded because of the seeming validity of ingroup positions (e.g. my kind
of people believe X). Yet evidence also exists for other kinds of processing.
Empirical research has identiﬁed a number of conditions under which group identity
motivates recipients to conduct a systematic, careful evaluation and interpretation
of the appeal, and attitude change or resistance depends on this evaluation.
Systematic processing has been found when the identity of the source group is
relevant to recipients’ own self-deﬁnitions (Wood et al 1996), when the issue is
relevant to recipients’ membership group (Crano & Chen 1998, Mackie et al
1990), when the message position is representative of ingroup consensus and thus
is informative about the true group norms (van Knippenberg & Wilke 1992), and
when the inﬂuence appeal evaluates an ingroup member and thus has implications
for recipients’ self-evaluations (Budesheim et al 1996). Systematic processing
also emerges when recipients are unable to process heuristically, such as when
the group position is provided after the message (Mackie et al 1992), and when
the context enables careful analysis, such as when people have sufﬁcient time to
process a message from a salient, potentially important ingroup (Hogg 1996).
Thus, the empirical data suggest that group inﬂuence does not operate through a
single process. People’s motives to align with or differentiate from social groups
can yield heuristic analyses and other forms of relatively superﬁcial information
processing, or they can yield careful, systematic processing of relevant
information.
Social Consensus and Validity of Information
Social identity and self-categorization theories offer a new perspective on the
question of what makes information inﬂuential (see also Moscovici 1976). In
message-based persuasion paradigms, strong, cogent arguments typically have
been deﬁned as ones that link the attitude issue to highly valued outcomes and
that generate favorable reactions from recipients, whereas weak arguments are
ones that link the issue to less-valued outcomes and that generate more negative
reactions (Eagly & Chaiken 1993). Recent work within this tradition, for example,
ATTITUDE CHANGE 559
has examined how inserting weak arguments into an otherwise strong message
can impair persuasion (Friedrich et al 1996, Friedrich & Smith 1998).
Social identity perspectives instead have focused on the social determinants
of information validity. This view was developed in opposition to Festinger’s
(1954) claim that people prefer objective reality testing and use social reality
testing (e.g. comparison with others’ views) primarily when objective standards
are not available. In self-categorization theory, reality testing is a single process
in which people use the normative standards of relevant reference groups to
achieve a valid, consensually shared understanding of reality congruent with their
social identity (Turner 1991). In a strong statement of the implications of this
position, Turner & Oakes argued that consensual judgments ‘‘are rationally more
likely to reﬂect a deeper truth about the world, not because agreement always
indicates accuracy, but because they have emerged from, and survived processes
of discussion, argument, and collective testing’’ (1997:369).
Empirical data, however, suggest that the relation between social consensus
and the apparent accuracy of information is contingent on a variety of factors.
Subjective validity emerges from consensus that is established through the convergence
of independent rather than dependent views and through validation by
an individual’s own, private cognitive processing (Levine 1996, Mackie & Skelly
1994). Furthermore, social consensus appears to be more impactful for some kinds
of issues than for others, presumably because ingroup consensus implies subjective
validity more strongly for some issues than for others. For example, majority
consensus has an especially strong impact on judgments of personal preference,
and less impact on judgments of objective stimuli, presumably because consensus
indicates preferences likely to be shared with similar others (Crano & HannulaBral
1994, Wood et al 1994).
Multiple Motives Instigated by Groups
In social identity and self-categorization theories, the motive for inﬂuence derives
largely from the desire to establish and maintain a positive evaluation of the self.
In support, Pool et al (1998) demonstrated that people maintain a favorable selfview
by shifting their attitudes to align with positively valued groups and to
deviate from negatively valued ones. Attitude change motivated by social identity
also can be driven by other self-related concerns, such as striving to be true to
oneself and to achieve a coherent, certain self-view (Abrams & Hogg 1988).
The tripartite analysis of motives presented in this chapter suggests additional
reasons why people adopt or reject group positions (see also Wood 1999). One
involves the positive and negative outcomes provided by others. For example,
going along with others in order to get along with them (i.e. receive social rewards
and avoid punishments) is likely to be important in close relationships and other
ingroup settings in which social harmony is valued. Another potential motive for
adopting or rejecting group positions is in order to achieve an accurate, valid
560 WOOD
understanding of reality. This possibility challenges perspectives that have associated
group inﬂuence with bias. For example, in an analysis of democratic participation,
Pratkanis & Turner (1996) proposed that active, deliberative discussion
and analysis of political issues occurs when citizens view issues as personally
relevant, whereas uncritical acceptance of political solutions and propaganda provided
by a ruling elite occurs when citizens wish to assume a certain social identity
(e.g. party allegiance). However, to the extent that some social and group identities
(e.g. informed voter, responsible citizen) motivate people to adopt valid,
accurate positions, then group-related motives are not necessarily associated with
biased processing.
OPINION MINORITY AND MAJORITY GROUPS
The seeming paradox that ‘‘few can inﬂuence many’’ has continued to spark
interest in minority inﬂuence. This work typically deﬁnes minorities as sources
advocating infrequent, low consensus positions.
Recent research has progressed beyond the original notion that minority
sources are inﬂuential because they elicit informational conﬂict and challenge
recipients’ understanding of issues (Moscovici 1985). Instead, the most consistent
ﬁnding appears to be that the social identity of minorities inhibits inﬂuence.
Minorities are most inﬂuential when their impact is assessed on ‘‘indirect’’ judgment
measures—ones on which recipients are relatively unaware that their judgments
could align them with the deviant minority source (Wood et al 1994). Yet
even on indirect measures, evidence of minority impact often fails to emerge
(Martin 1998). Minorities seem to face two impediments to exerting inﬂuence.
They are, by deﬁnition, not important reference groups. Thus source group identity
is unlikely to motivate people to attend to and evaluate an appeal (De Vries
et al 1996). Furthermore, even when people process the appeal, the source’s low
consensus, deviant position is likely to yield a negatively biased processing orientation.
As a result, recipients resist inﬂuence (Mugny 1980, Erb et al 1998) or
demonstrate ‘‘boomerang’’ shifts away from the minority view (Pool et al 1998,
Wood et al 1996).
In a series of elegant experiments, Crano and his colleagues (Alvaro & Crano
1996, 1997; Crano & Alvaro 1998; Crano & Chen 1998) demonstrated that these
impediments to minority inﬂuence can be surmounted by ingroup minority
sources (see also David & Turner 1996). It appears that the distinctiveness of
ingroup minority positions encourages careful message processing, yet with the
lenient, open-minded orientation typically accorded to ingroup members. The
deviance of the minority view attenuates acceptance on direct attitude measures
but the imbalance in attitude structure that results from message processing yields
attitude change toward the minority view on measures indirectly related to the
appeal.
ATTITUDE CHANGE 561
Several lines of research have documented the kinds of thought recipients
generate to minority appeals. Especially in problem-solving contexts in which
novel solutions are valued (Nemeth 1986), minorities encourage recipients to
think about issues in a divergent manner and to consider novel ideas and solution
strategies (Butera et al 1996, Erb et al 1998, Gruenfeld et al 1998, Nemeth &
Rogers 1996, Peterson & Nemeth 1996). Attributional reasoning is also likely to
mediate inﬂuence, given that minority positions are often unexpected and require
explanation. Research ﬁndings generally support the analysis by Eagly et al
(1981) that inﬂuence is impaired when the advocated position can be attributed
to a potentially biasing attribute of the source or situation, such as the source’s
personal self-interest (Moskowitz 1996). In like manner, inﬂuence is enhanced
when the advocated position can be attributed to external reality or the truth about
the issue (Bohner et al 1996, 1998a). Also suggesting attributional reasoning is
the ﬁnding that minorities are inﬂuential when they express their views in contexts,
such as face-to-face interaction, in which deviancy can incur costs (McLeod
et al 1997). Advocacy then suggests a source who is sufﬁciently committed and
certain to risk social rejection.
A useful goal for future research will be to identify the speciﬁc aspect(s) of
minority identity that motivate agreement or resistance. Following the threemotive
scheme of this chapter, it may be that the minority source has relevance
for recipients’ personal identities and represents, for example, a deviant other
from whom they wish to differentiate or a valued innovator they wish to emulate
(Wood et al 1996). Another motive concerns recipients’ relations with others,
such as the desire to be lenient with ingroup minority sources (Crano & Chen
1998). Finally, recipients could achieve an accurate understanding of an issue by
agreeing with minorities whose positions appear to reﬂect external reality (Bohner
et al 1996).
CONCLUSION
This chapter has drawn from the literatures on message-based persuasion and
social inﬂuence to identify common themes in attitude change research. In these
traditionally separate areas of investigation, theoretical and empirical work has
begun to delineate the motives underlying recipients’ responses to inﬂuence
appeals and the variety of cognitive and affective processes involved in attitude
change and resistance.
The Annual Review chapter on attitudes traditionally has focused on messagebased
persuasion research that examines attitudes at the individual level. From
this standpoint, the current inclusion of social inﬂuence ﬁndings highlights the
sometimes neglected point that attitudes are social phenomena, that they emerge
from and are embedded in social interaction. Yet, enthusiasm for aggregating
knowledge across these two areas of investigation should not overwhelm the
562 WOOD
equally important point that each area is associated with unique predictors and
processes. Models of inﬂuence via complexly argued persuasive messages will
need to address unique factors that are not as important in the study of simple
messages, including recipients’ ability to engage in extensive cognitive processing
and their knowledge about the message topic. In like manner, models of inﬂuence
in groups and other complex social contexts address unique factors such as the
likelihood of information exchange and the interaction structure in the social
setting.
A challenge for future investigation will be to continue to develop models of
social and cognitive processes that are sufﬁciently inclusive to capture attitude
change in the variety of social and informational settings in which it occurs. An
important aspect of this challenge is to place persuasion and social inﬂuence
within a framework that recognizes cross-cultural and ethnic effects. It is appropriate
to end the chapter with a noteworthy example of such an approach, the
meta-analytic synthesis by Bond & Smith (1996) of social inﬂuence experiments
from 17 countries. The usefulness of integrating cultural-level phenomena into
attitude theories is evident in the greater levels of conformity in experiments
conducted in nations characterized by collective than by individualistic values.
ACKNOWLEDGMENTS
Preparation of this chapter was supported by National Science Foundation grant
#SBR 95–14537. The author thanks William D Crano, Alice H Eagly, David
Matz, Richard E Petty, and Radmila Prislin for their helpful comments on an
earlier draft of the chapter.
Visit the Annual Reviews home page at www.AnnualReviews.org.
LITERATURE CITED
Abrams D, Hogg MA. 1988. Comments on the
motivational status of self-esteem in social
identity and intergroup discrimination. Eur.
J. Soc. Psychol. 18:317–34
Ajzen I. 1996. The directive inﬂuence of attitudes
on behavior. See Gollwitzer & Bargh
1996, pp. 385–403
Allport GW. 1985. The historical background
of social psychology. See Lindzey & Aronson
1985, 1:1–46
Alvaro EM, Crano WD. 1996. Cognitive
responses to minority- or majority-based
communications: factors that underlie
minority inﬂuence. Br. J. Soc. Psychol.
35:105–21
Alvaro EM, Crano WD. 1997. Indirect minority
inﬂuence: evidence for leniency in
source evaluation and counterargumentation.
J. Pers. Soc. Psychol. 72:949–64
Aronson E. 1992. The return of the repressed:
Dissonance theory makes a comeback. Psychol.
Inq. 3:303–11
Asch SE. 1940. Studies in the principles of
judgments and attitudes. II. Determination
of judgments by group and by ego standards.
J. Soc. Psychol. 12:433–65
Aspinwall LG, Brunhart SM. 1996. Distinguishing
optimism from denial: optimistic
beliefs predict attention to health threats.
Pers. Soc. Psychol. Bull. 22:993–1003
ATTITUDE CHANGE 563
Baldwin MW, Holmes JG. 1987. Salient private
audiences and awareness of the self. J.
Pers. Soc. Psychol. 52:1087–98
Ball-Rokeach SJ, Loges WE. 1996. Making
choices: media roles in the construction of
value-choices. See Seligman et al 1996, pp.
277–98
Baron RS, Vandello JA, Brunsman B. 1996.
The forgotten variable in conformity
research: impact of task importance on
social inﬂuence. J. Pers. Soc. Psychol.
71:915–27
Betz AL, Skowronski J, Ostrom TO. 1996.
Shared realities: social inﬂuence and stimulus
memory. Soc. Cogn. 14:113–40
Biek M, Wood W, Chaiken S. 1996. Working
knowledge, cognitive processing, and attitudes:
on the determinants of bias. Pers.
Soc. Psychol. Bull. 22:547–56
Blanton H, Cooper J, Skurnik I, Aronson J.
1997. When bad things happen to good
feedback: exacerbating the need for selfjustiﬁcation
with self-afﬁrmations. Pers.
Soc. Psychol. Bull. 23:684–92
Bless H, Clore GO, Schwarz N, Golisano V,
Rabe C, Wolk M. 1996. Mood and the use
of scripts: Does a happy mood really lead
to mindlessness? J. Pers. Soc. Psychol.
71:665–79
Bohner G, Erb HP, Reinhard MA, Frank E.
1996. Distinctiveness across topics in
minority and majority inﬂuence: an attributional
analysis and preliminary data. Br.
J. Soc. Psychol. 35:27–46
Bohner G, Frank E, Erb HP. 1998a. Heuristic
processing of distinctiveness information in
minority and majority inﬂuence. Eur. J.
Soc. Psychol. 28:855–60
Bohner G, Rank S, Reinhard MA, Einwiller S,
Erb HP. 1998b. Motivational determinants
of systematic processing: Expectancy moderates
effects of desired conﬁdence on processing
effort. Eur. J. Soc. Psychol. 28:185–
206
Bond R, Smith PB. 1996. Culture and conformity:
a meta-analysis of studies using
Asch’s (1952, 1956) line judgment task.
Psychol. Bull. 119:111–37
Bosveld W, Koomen W, Vogelaar R. 1997.
Construing a social issue: effects on attitudes
and the false consensus effect. Br. J.
Soc. Psychol. 36:263–72
Brewer MB, Brown RJ. 1998. Intergroup relations.
See Gilbert et al 1998, 2:554–94
Budesheim TL, Houston DA, DePaola SJ.
1996. Persuasiveness of in-group and outgroup
political messages: the case of negative
political campaigning. J. Pers. Soc.
Psychol. 70:523–34
Burris CT, Harmon-Jones E, Tarpley WR.
1997. ‘‘By faith alone’’: religious agitation
and cognitive dissonance. Basic Appl. Soc.
Psychol. 19:17–31
Butera F, Mugny G, Legrenzi P, Perez JA.
1996. Majority and minority inﬂuence, task
representation and inductive reasoning. Br.
J. Soc. Psychol. 35:123–36
Cacioppo JT, Gardner WL, Berntson GG.
1997. Beyond bipolar conceptualizations
and measures: the case of attitudes and
evaluative space. Pers. Soc. Psychol. Rev.
1:3–25
Cacioppo JT, Petty RE, Feinstein J, Jarvis B.
1996. Dispositional differences in cognitive
motivation: the life and times of individuals
varying in need for cognition. Psychol.
Bull. 119:197–253
Celuch K, Lust J, Showers L. 1998. A test of
a model of consumers’ responses to product
manual safety information. J. Appl. Soc.
Psychol. 28:377–94
Chaiken S, Giner-Sorolla R, Chen S. 1996a.
Beyond accuracy: defense and impression
motives in heuristic and systematic information
processing. See Gollwitzer & Bargh
1996, pp. 553–78
Chaiken S, Pomerantz EM, Giner-Sorolla R.
1995. Structural consistency and attitude
strength. See Petty & Krosnick 1995, pp.
387–412
Chaiken S, Wood W, Eagly AH. 1996b. Principles
of persuasion. See Higgins & Kruglanski
1996, pp. 702–42
Chen S, Schechter D, Chaiken S. 1996. Getting
at the truth or getting along: accuracy- versus
impression-motivated heuristic and sys-
564 WOOD
tematic processing. J. Pers. Soc. Psychol.
71:262–75
Cialdini RB, Levy A, Herman CP, Kozlowski
LT, Petty RE. 1976. Elastic shifts of opinion:
determinants of direction and durability.
J. Pers. Soc. Psychol. 34:663–72
Cialdini RB, Trost MR. 1998. Social inﬂuence:
social norms, conformity, and compliance.
See Gilbert et al 1998, 2:151–92
Cooper J, Fazio RH. 1984. A new look at dissonance
theory. Adv. Exp. Soc. Psychol.
17:229–66
Cowan G, Hodge C. 1996. Judgments of hate
speech: the effects of target group, publicness,
and behavioral responses of the target.
J. Appl. Soc. Psychol. 26:355–74
Crano WD, Alvaro EM. 1998. The context/
comparison model of social inﬂuence:
mechanisms, structure, and linkages that
underlie indirect attitude change. Eur. Rev.
Soc. Psychol. 8:175–202
Crano WD, Chen X. 1998. The leniency contract
and persistence of majority and minority
inﬂuence. J. Pers. Soc. Psychol. 74:
1437–50
Crano WD, Hannula-Bral KA. 1994. Context/
categorization model of social inﬂuence:
minority and majority inﬂuence in the formation
of a novel response norm. J. Exp.
Soc. Psychol. 30:247–76
David B, Turner JC. 1996. Studies in selfcategorization
and minority conversion: Is
being a member of the out-group an advantage?
Br. J. Soc. Psychol. 35:179–99
Deutsch M, Gerard HB. 1955. A study of normative
and informational social inﬂuences
upon individual judgment. J. Abnorm. Soc.
Psychol. 51:629–36
De Vries NK, De Dreu CKW, Gordijn EH,
Schuurman M. 1996. Majority and minority
inﬂuence: a dual role interpretation. Eur.
Rev. Soc. Psychol. 7:145–72
Dickerson P. 1997. ‘It’s not just me who’s saying
this . . . .’ The deployment of cited others
in televised political discourse. Br. J.
Soc. Psychol. 36:33–48
Dillard JP, Plotnick CA, Godbold LC, Freimuth
VS, Edgar T. 1996. The multiple
affective outcomes of AIDS PSAs: Fear
appeals do more than scare people. Commun.
Res. 23:44–72
Ditto PH, Scepansky JA, Munro GD, Apanovitch
AM, Lockhart LK. 1998. Motivated
sensitivity to preference-inconsistent information.
J. Pers. Soc. Psychol. 75:53–69
Eagly AH, Chaiken S. 1993. The Psychology
of Attitudes. Orlando, FL: Harcourt Brace
Jovanovich
Eagly AH, Chaiken S. 1998. Attitude structure
and function. See Gilbert et al 1998, 1:269–
322
Eagly AH, Chaiken S, Wood W. 1981. An attribution
analysis of persuasion. In New
Directions in Attribution Research, ed. JH
Harvey, WJ Ickes, RF Kidd, 3:37–62. Mahwah,
NJ: Erlbaum
Eagly AH, Kulesa P. 1997. Attitudes, attitude
structure, and resistance to change: implications
for persuasion on environmental
issues. In Environment, Ethics, and Behavior:
The Psychology of Environmental Valuation
and Degradation, ed. MH Bazerman,
DM Messick, AE Tenbrunsel, KA WadeBenzoni,
pp. 122–53. San Francisco: New
Lexington
Edwards K, Smith EE. 1996. A disconﬁrmation
bias in the evaluation of arguments. J.
Pers. Soc. Psychol. 71:5–24
Erb H, Bohner G, Schmalzle K, Rank S. 1998.
Beyond conﬂict and discrepancy: cognitive
bias in minority and majority inﬂuence.
Pers. Soc. Psychol. Bull. 24:620–33
Fabrigar LR, Priester JR, Petty RE, Wegener
DT. 1998. The impact of attitude accessibility
on elaboration of persuasive messages.
Pers. Soc. Psychol. Bull. 24:339–52
Fein S, McCloskey AL, Tomlinson TM. 1997.
Can the jury disregard that information?
The use of suspicion to reduce the prejudicial
effects of pretrial publicity and inadmissible
testimony. Pers. Soc. Psychol.
Bull. 23:1215–26
Festinger L. 1954. A theory of social comparison
processes. Hum. Relat. 7:117–40
Festinger L. 1957. A Theory of Cognitive Dissonance.
Evanston, IL: Row Peterson
ATTITUDE CHANGE 565
Fiske ST. 1998. Stereotyping, prejudice, and
discrimination. See Gilbert et al 1998,
2:357–414
Forgas JP. 1995. Mood and judgment: the
affect infusion model (AIM). Psychol. Bull.
117:39–66
Forgas JP. 1998. Asking nicely? The effects of
mood on responding to more or less polite
requests. Pers. Soc. Psychol. Bull. 24:173–
85
Fried CB. 1998. Hypocrisy and identiﬁcation
with transgressions: a case of undetected
dissonance. Basic Appl. Soc. Psychol.
20:145–54
Fried CB, Aronson E. 1995. Hypocrisy, misattribution,
and dissonance reduction. Pers.
Soc. Psychol. Bull. 21:925–33
Friedrich J, Fetherstonhaugh D, Casey S, Gallagher
D. 1996. Argument integration and
attitude change: suppression effects in the
integration of one-sided arguments that
vary in persuasiveness. Pers. Soc. Psychol.
Bull. 22:179–91
Friedrich J, Smith P. 1998. Suppressive inﬂuence
of weak arguments in mixed-quality
messages: an exploration of mechanisms
via argument rating, pretesting, and order
effects. Basic Appl. Soc. Psychol. 20:293–
304
Garst J, Bodenhausen GV. 1996. ‘‘Family values’’
and political persuasion: impact of
kin-related rhetoric on reactions to political
campaigns. J. Appl. Soc. Psychol. 26:1119–
37
Gibbons FX, Eggleston TJ, Benthin AC. 1997.
Cognitive reactions to smoking relapse: the
reciprocal relation between dissonance and
self-esteem. J. Pers. Soc. Psychol. 72:184–
95
Gilbert DT, Fiske ST, Lindzey G, eds. 1998.
The Handbook of Social Psychology, Vols.
1, 2. Boston: McGraw-Hill. 4th ed.
Giner-Sorolla R, Chaiken S. 1997. Selective
use of heuristic and systematic processing
under defense motivation. Pers. Soc. Psychol.
Bull. 23:84–97
Gollwitzer PM, Bargh JA, eds. 1996. The Psychology
of Action: Linking Cognition and
Motivation to Behavior. New York:
Guilford
Greenwald AG, Banaji MR. 1995. Implicit
social cognition: attitudes, self-esteem, and
stereotypes. Psychol. Rev. 102:4–27
Gruenfeld DH, Thomas-Hunt MC, Kim PH.
1998. Cognitive ﬂexibility, communication
strategy, and integrative complexity in
groups: public versus private reactions to
majority and minority status. J. Exp. Soc.
Psychol. 34:202–26
Hafer CL, Reynolds K, Obertynski MA. 1996.
Message comprehensibility and persuasion:
effects of complex language in counterattitudinal
appeals to laypeople. Soc. Cogn.
14:317–37
Hardin CD, Higgins ET. 1996. Shared reality:
How social veriﬁcation makes the subjective
objective. In Handbook of Motivation
and Cognition, ed. RM Sorrentino, ET Higgins,
3:28–84. New York: Guilford
Harmon-Jones E, Brehm JW, Greenberg J,
Simon L, Nelson DE. 1996. Evidence that
the production of aversive consequences is
not necessary to create cognitive dissonance.
J. Pers. Soc. Psychol. 70:5–16
Haslam SA, McGarty C, Turner JC. 1996.
Salient group memberships and persuasion:
the role of social identity in the validation
of beliefs. In What’s Social About Social
Cognition? Research on Socially Shared
Cognition in Small Groups, ed. JL Nye,
AM Brower, pp. 29–56. Thousand Oaks,
CA: Sage
Haslam SA, Oakes PJ, McGarty C, Turner JC,
Onorato RS. 1995. Contextual changes in
the prototypicality of extreme and moderate
outgroup members. Eur. J. Soc. Psychol.
25:509–30
Hastie R, Park B. 1986. The relationship
between memory and judgment depends on
whether the judgment task is memorybased
or on-line. Psychol. Rev. 93:258–68
Higgins ET. 1981. ‘‘The communication
game’’: implications for social cognition
and persuasion. In Social Cognition: The
Ontario Symposium, ed. ET Higgins, CP
566 WOOD
Herman, MP Zanna, 1:343–92. Mahwah,
NJ: Erlbaum
Higgins ET, Kruglanski AW, eds. 1996. Social
Psychology: Handbook of Basic Principles.
New York: Guilford
Hogg MA. 1996. Intragroup processes, group
structure and social identity. In Social
Groups and Identities: Developing the
Legacy of Henri Tajfel, ed. WP Robinson,
pp. 65–93. Oxford, UK: Butterworth-
Heinemann
Howard DJ. 1997. Familiar phrases as peripheral
persuasion cues. J. Exp. Soc. Psychol.
33:231–43
Jepson C, Chaiken S. 1990. Chronic issuespeciﬁc
fear inhibits systematic processing
of persuasive communications. J. Soc.
Behav. Pers. 5:61–84
Johnson BT, Eagly AH. 1989. Effects of
involvement on persuasion: a metaanalysis.
Psychol. Bull. 106:290–314
Joule R, Beauvois J. 1998. Cognitive dissonance
theory: a radical view. Eur. Rev. Soc.
Psychol. 8:1–32
Kameda T, Ohtsubo Y, Takezawa M. 1997.
Centrality in sociocognitive networks and
social inﬂuence: an illustration in a group
decision-making context. J. Pers. Soc. Psychol.
73:296–309
Kassin SM, Kiechel KL. 1996. The social psychology
of false confessions: compliance,
internalization, and confabulation. Psychol.
Sci. 7:125–28
Kassin SM, Sommers SR. 1997. Inadmissible
testimony, instructions to disregard, and the
jury: substantive versus procedural considerations.
Pers. Soc. Psychol. Bull. 23:1046–
54
Katz D. 1960. The functional approach to the
study of attitudes. Public Opin. Q. 24:163–
204
Kelly JR, Jackson JW, Hutson-Comeaux SL.
1997. The effects of time pressure and task
differences on inﬂuence modes and accuracy
in decision-making groups. Pers. Soc.
Psychol. Bull. 23:10–22
Kelman HC. 1958. Compliance, identiﬁcation,
and internalization: three processes of attitude
change. J. Conﬂict Resolut. 2:51–60
Killeya LA, Johnson BT. 1998. Experimental
induction of biased systematic processing:
the directed thought technique. Pers. Soc.
Psychol. Bull. 24:17–33
Lambert AJ, Cronen S, Chasteen AL, Lickel
B. 1996. Private vs public expressions of
racial prejudice. J. Exp. Soc. Psychol.
32:437–59
LaTour MS, Rotfeld HJ. 1997. There are
threats and (maybe) fear-caused arousal:
theory and confusions of appeals to fear and
fear arousal itself. J. Advert. 26:45–59
Lavine H, Snyder M. 1996. Cognitive processing
and the functional matching effect in
persuasion: the mediating role of subjective
perceptions of message quality. J. Exp. Soc.
Psychol. 32:580–604
Levine J. 1996. Solomon Asch’s legacy for
group research. Presented at Annu. Meet.
Soc. Pers. Soc. Psychol., Sturbridge, MA
Liberman A, Chaiken S. 1996. The direct effect
of personal relevance on attitudes. Pers.
Soc. Psychol. Bull. 22:269–79
Lindzey G, Aronson E, eds. 1985. Handbook
of Social Psychology, Vols. 1, 2. New York:
Random House
Lundgren SR, Prislin R. 1998. Motivated cognitive
processing and attitude change. Pers.
Soc. Psychol. Bull. 24:715–26
Mackie DM, Gastardo-Conaco MC, Skelly JJ.
1992. Knowledge of the advocated position
and the processing of in-group and outgroup
persuasive messages. Pers. Soc. Psychol.
Bull. 18:145–51
Mackie DM, Skelly JJ. 1994. The social cognition
analysis of social inﬂuence: contributions
to the understanding of persuasion
and conformity. In Social Cognition:
Impact on Social Psychology, ed. P Devine,
D Hamilton, T Ostrom, pp. 259–89. New
York: Academic
Mackie DM, Smith ER. 1998. Intergroup relations:
insights from a theoretically integrative
approach. Psychol. Rev. 105:499–529
ATTITUDE CHANGE 567
Mackie DM, Worth LT, Asuncion AG. 1990.
Processing of persuasive ingroup messages.
J. Pers. Soc. Psychol. 58:812–22
Maio GR, Bell DW, Esses VM. 1996. Ambivalence
in persuasion: the processing of messages
about immigrant groups. J. Exp. Soc.
Psychol. 32:513–36
Maio GR, Olson JM. 1998. Attitude dissimulation
and persuasion. J. Exp. Soc. Psychol.
34:182–201
Marsh KL, Hart-O’Rourke DM, Julka DL.
1997. The persuasive effects of verbal and
nonverbal information in a context of value
relevance. Pers. Soc. Psychol. Bull. 23:563–
79
Martin R. 1998. Majority and minority inﬂuence
using the afterimage paradigm: a
series of attempted replications. J. Exp. Soc.
Psychol. 34:1–26
McConnell AR, Leibold JM, Sherman SJ.
1997. Within-target illusory correlations
and the formation of context-dependent
attitudes. J. Pers. Soc. Psychol. 73:675–86
McGarty C, Haslam SA, Hutchinson KJ,
Turner JC. 1994. The effects of salient
group memberships on persuasion. Small
Group Res. 25:267–93
McLeod PL, Baron RS, Marti MW, Yoon K.
1997. The eyes have it: minority inﬂuence
in face-to-face and computer-mediated
group discussion. J. Appl. Psychol. 82:706–
18
McMahan S, Witte K, Meyer J. 1998. The perception
of risk messages regarding electromagnetic
ﬁelds: extending the extended
parallel process model to an unknown risk.
Health Commun. 10:247–59
Millar MG, Millar KU. 1996. Effects of message
anxiety on disease detection and health
promotion behaviors. Basic Appl. Soc. Psychol.
18:61–74
Moscovici S. 1976. Social Inﬂuence and Social
Change. London: Academic
Moscovici S. 1985. Social inﬂuence and conformity.
See Lindzey & Aronson 1985,
2:347–412
Moskowitz GB. 1996. The mediational effects
of attributions and information processing
in minority social inﬂuence. Br. J. Soc. Psychol.
35:47–66
Mugny G. 1980. The Power of Minorities. London:
Academic
Munro GD, Ditto PH. 1997. Biased assimilation,
attitude polarization, and affect in
reactions to stereotype-relevant scientiﬁc
information. Pers. Soc. Psychol. Bull.
23:636–53
Murray SL, Haddock G, Zanna MP. 1996. On
creating value-expressive attitudes: an
experimental approach. See Seligman et al
1996, pp. 107–33
Nelson TE, Kinder DR. 1996. Issue frames and
group-centrism in American public opinion.
J. Polit. 58:1055–78
Nelson TE, Oxley ZM, Clawson RA. 1997.
Toward a psychology of framing effects.
Polit. Behav. 19:221–46
Nemeth CJ. 1986. Differential contributions of
majority and minority inﬂuence. Psychol.
Rev. 93:23–32
Nemeth CJ, Rogers J. 1996. Dissent and the
search for information. Br. J. Soc. Psychol.
35:67–76
Ottati V, Terkildsen N, Hubbard C. 1997.
Happy faces elicit heuristic processing in a
televised impression formation task: a cognitive
tuning account. Pers. Soc. Psychol.
Bull. 23:1144–56
Peterson RS, Nemeth C. 1996. Focus versus
ﬂexibility: Majority and minority inﬂuence
can both improve performance. Pers. Soc.
Psychol. Bull. 22:14–23
Petty RE, Krosnick JA. 1995. Attitude
Strength: Antecedents and Consequences.
Mahwah, NJ: Erlbaum
Petty RE, Wegener DT. 1998a. Attitude
change: multiple roles for persuasion variables.
See Gilbert et al 1998, 1:323–90
Petty RE, Wegener DT. 1998b. Matching versus
mismatching attitude functions: implications
for scrutiny of persuasive messages.
Pers. Soc. Psychol. Bull. 24:227–40
Petty RE, Wegener DT, Fabrigar LR. 1997.
Attitudes and attitude change. Annu. Rev.
Psychol. 48:609–47
568 WOOD
Petty RE, Wegener DT, White PH. 1998. Flexible
correction processes in social judgment:
implications for persuasion. Soc.
Cogn. 16:93–113
Pfau M. 1997. The inoculation model of resistance
to inﬂuence. Prog. Commun. Sci.
13:133–71
Pool GJ, Wood W, Leck K. 1998. The selfesteem
motive in social inﬂuence: agreement
with valued majorities and
disagreement with derogated minorities. J.
Pers. Soc. Psychol. 75:967–75
Pratkanis AR, Turner ME. 1996. Persuasion
and democracy: strategies for increasing
deliberative participation and enacting
social change. J. Soc. Issues. 52:187–205
Prentice-Dunn S, Jones JL, Floyd DL. 1997.
Persuasive appeals and the reduction of
skin cancer risk: the roles of appearance
concern, perceived beneﬁts of a tan, and
efﬁcacy information. J. Appl. Soc. Psychol.
27:1041–47
Price V, Tewksbury D. 1997. News values and
public opinion: a theoretical account of
media priming and framing. Prog. Commun.
Sci. 13:173–212
Priester JR, Petty RE. 1996. The gradual
threshold model of ambivalence: relating
the positive and negative bases of attitudes
to subjective ambivalence. J. Pers. Soc.
Psychol. 71:431–49
Prislin R, Pool GJ. 1996. Behavior, consequences,
and the self: Is all well that ends
well? Pers. Soc. Psychol. Bull. 22:933–48
Prislin R, Wood W, Pool GJ. 1998. Structural
consistency and the deduction of novel
from existing attitudes. J. Exp. Soc. Psychol.
34:66–89
Read SJ, Vanman EJ, Miller LC. 1997. Connectionism,
parallel constraint satisfaction
processes, and gestalt principles:
(re)introducing cognitive dynamics to
social psychology. Pers. Soc. Psychol. Rev.
1:26–53
Reicher S, Hopkins N. 1996a. Seeking inﬂuence
through categorizing self-categories:
an analysis of anti-abortionist rhetoric. Br.
J. Soc. Psychol. 35:297–311
Reicher S, Hopkins N. 1996b. Self-category
constructions in political rhetoric; an analysis
of Thatcher’s and Kinnock’s speeches
concerning the British miners’ strike
(1984–5). Eur. J. Soc. Psychol. 26:353–71
Richard R, Van der Pligt J, De Vries NK.
1996a. Anticipated affect and behavioral
choice. Basic Appl. Soc. Psychol. 18:111–
29
Richard R, Van der Pligt J, De Vries NK.
1996b. Anticipated regret and time perspective:
changing sexual risk-taking
behavior. J. Behav. Dec. Mak. 9:185–99
Rogers RW. 1983. Cognitive and physiological
processes in fear appeals and attitude
change: a revised theory of protection motivation.
In Social Psychophysiology: A
Sourcebook, ed. JT Cacioppo, RE Petty, pp.
153–77. New York: Guilford
Rogers RW, Prentice-Dunn S. 1997. Protection
motivation theory. In Handbook of Health
Behavior Research, ed. D Gochman,
1:113–32. New York: Plenum
Romero AA, Agnew CR, Insko CA. 1996. The
cognitive mediation hypothesis revisited:
an empirical response to methodological
and theoretical criticism. Pers. Soc. Psychol.
Bull. 22:651–65
Rothman AJ, Salovey P. 1997. Shaping perceptions
to motivate healthy behavior: the
role of message framing. Psychol. Bull.
121:3–19
Rothman AJ, Schwarz N. 1998. Constructing
perceptions of vulnerability: personal relevance
and the use of experiential information
in health judgments. Pers. Soc.
Psychol. Bull. 24:1053–64
Ruscher JB, Duval LL. 1998. Multiple communicators
with unique target information
transmit less stereotypical impressions. J.
Pers. Soc. Psychol. 74:329–44
Rusting CL. 1998. Personality, mood, and the
cognitive processing of emotional information:
three conceptual frameworks. Psychol.
Bull. 124:165–96
Schul Y, Goren H. 1997. When strong evidence
has less impact than weak evidence: bias,
ATTITUDE CHANGE 569
adjustment, and instructions to ignore. Soc.
Cogn. 15:133–55
Schwarz N. 1997. Moods and attitude judgments:
a comment on Fishbein and Middlestadt.
J. Consum. Psychol. 6:93–98
Schwarz N, Clore GL. 1996. Feelings and phenomenal
experiences. See Higgins & Kruglanski
1996, pp. 433–65
Seligman C, Olson JM, Zanna MP, eds. 1996.
The Psychology of Values: The Ontario
Symposium, Vol. 8. Mahwah, NJ: Erlbaum
Shestowsky D, Wegener DT, Fabrigar LR.
1998. Need for cognition and interpersonal
inﬂuence: individual differences in impact
on dyadic decision. J. Pers. Soc. Psychol.
74:1317–28
Shultz TR, Lepper MR. 1996. Cognitive dissonance
reduction as constraint satisfaction.
Psychol. Rev. 103:219–40
Slater MD, Rouner D. 1996. Value-afﬁrmative
and value-protective processing of alcohol
education messages that include statistical
evidence or anecdotes. Commun. Res.
23:210–35
Smith ER, Fazio RH, Cejka MA. 1996. Accessible
attitudes inﬂuence categorization of
multiply categorizable objects. J. Pers. Soc.
Psychol. 71:888–98
Smith MB, Bruner JS, White RW. 1956. Opinions
and Personality. New York: Wiley
Smith SM, Petty RE. 1996. Message framing
and persuasion: a message processing analysis.
Pers. Soc. Psychol. Bull. 22:257–68
Stadler DR, Baron RS. 1998. Attributional
complexity as a moderator of dissonanceproduced
attitude change. J. Pers. Soc. Psychol.
75:449–55
Steele CM. 1988. The psychology of selfafﬁrmation.
Adv. Exp. Soc. Psychol.
21:261–302
Stone J, Wiegand AW, Cooper J, Aronson E.
1997. When exempliﬁcation fails: hypocrisy
and the motive for self-integrity. J.
Pers. Soc. Psychol. 72:54–65
Sturges JW, Rogers RW. 1996. Preventive
health psychology from a developmental
perspective: an extension of protection
motivation theory. Health Psychol. 15:158–
66
Tajfel H. 1981. Human Groups and Social
Categories: Studies in Social Psychology.
Cambridge, UK: Cambridge Univ. Press
Tajfel H. 1982. Social psychology of intergroup
relations. Annu. Rev. Psychol. 33:1–
39
Thompson MM, Zanna MP, Grifﬁn DW. 1995.
Let’s not be indifferent about (attitudinal)
ambivalence. See Petty & Krosnick 1995,
pp. 361–86
Turco RM. 1996. Self-referencing, quality of
argument, and persuasion. Curr. Psychol.
Dev. Learn. Pers. Soc. 15:258–76
Turner JC. 1982. Towards a cognitive redeﬁnition
of the social group. In Social Identity
and Intergroup Relations, ed. H Tajfel, pp.
15–40. Cambridge, UK: Cambridge Univ.
Press
Turner JC. 1991. Social Inﬂuence. Paciﬁc
Grove, CA: Brooks/Cole
Turner JC, Oakes PJ. 1997. The socially structured
mind. In The Message of Social Psychology:
Perspectives on Mind in Society,
ed. C McGarty, SA Haslam, pp. 355–73.
Cambridge, MA: Blackwell
Tykocinski O, Higgins ET, Chaiken S. 1994.
Message framing, self-discrepancies, and
yielding to persuasive messages: the motivational
signiﬁcance of psychological situations.
Pers. Soc. Psychol. Bull. 20:107–15
Van der Pligt J, Zeelenberg M, Van Dijk WW,
De Vries NK, Richard R. 1998. Affect, attitudes,
and decisions: Let’s be more speciﬁc.
Eur. Rev. Soc. Psychol. 8:33–66
van Knippenberg D, Wilke H. 1992. Prototypicality
of arguments and conformity to
ingroup norms. Eur. J. Soc. Psychol.
22:141–55
Visser PS, Krosnick JA. 1998. Development of
attitude strength over the life cycle: surge
and decline. J. Pers. Soc. Psychol. 75:1389–
410
Wegener DT, Petty RE. 1996. Effects of mood
on persuasion processes: enhancing, reducing,
and biasing scrutiny of attituderelevant
information. In Striving and
570 WOOD
Feeling: Interactions Between Goals,
Affect, and Self-Regulation, ed. LL Martin,
A Tesser, pp. 329–62. Mahwah, NJ:
Erlbaum
Wegener DT, Petty RE. 1997. The ﬂexible correction
model: the role of naive theories in
bias correction. Adv. Exp. Soc. Psychol.
29:141–208
Wilson TD, Hodges SD. 1992. Attitudes as
temporary constructions. In The Construction
of Social Judgment, ed. LL Martin, A
Tesser, pp. 37–65. Mahwah, NJ: Erlbaum
Witte K. 1992. Putting the fear back into fear
appeals: the extended parallel process
model. Commun. Monogr. 59:329–49
Witte K. 1998. Fear as motivator, fear as inhibitor:
using the extended parallel process
model to explain fear appeal successes and
failures. In Handbook of Communication
and Emotion: Research, Theory, Applications,
and Contexts, ed. PA Andersen, LK
Guerro, et al, pp. 423–50. San Diego:
Academic
Witte K, Berkowitz JM, Cameron KA, McKeon
JK. 1998. Preventing the spread of
genital warts: using fear appeals to promote
self-protective behaviors. Health Educ.
Behav. 25:571–85
Wittenbrink B, Judd CM, Park B. 1997. Evidence
for racial prejudice at the implicit
level and its relationship with questionnaire
measures. J. Pers. Soc. Psychol. 72:262–74
Wood W. 1999. Motives and modes of processing
in the social inﬂuence of groups. In
Dual-Process Theories in Social Psychology,
ed. S Chaiken, Y Trope, pp. 547–70.
New York: Guilford
Wood W, Lundgren S, Ouellette J, Busceme S,
Blackstone T. 1994. Minority inﬂuence: a
meta-analytic review of social inﬂuence
processes. Psychol. Bull. 115:323–45
Wood W, Matz. 1999. Attitudes vary with
salient group identity. Unpublished data,
Texas A & M Univ.
Wood W, Pool GJ, Leck K, Purvis D. 1996.
Self-deﬁnition, defensive processing, and
inﬂuence: the normative impact of majority
and minority groups. J. Pers. Soc. Psychol.
71:1181–93
Zarnoth P, Sniezek JA. 1997. The social inﬂuence
of conﬁdence in group decision making.
J. Exp. Soc. Psychol. 33:345–66
Zuwerink JR, Devine PG. 1996. Attitude
importance and resistance to persuasion:
It’s not just the thought that counts. J. Pers.
Soc. Psychol. 70:931–44