Attitudes and Behavior 315since that time, researchers have identified several factors that
influence attitude-behavior consistency, and the appropriate
conclusion seems to be that measures of attitudes and behavior
are closely related in some circumstances but not in others. We
outline these factors in the following sections.
Compatibility of Attitude and Behavior Measures
An important conceptual advance came from Fishbein and
Ajzen’s (1975) theory of reasoned action. These theorists dis-
tinguished between attitudes toward objects and attitudes to-
ward behaviors—a distinction that can also be viewed as
general attitudes versus specific attitudes. Hypothetically,
attitudes toward objects should influence the favorability of
theclassof behaviors related to the object, whereas attitudes
toward behaviors should influence the favorability of those
specific behaviors.
The point made by Fishbein and Ajzen was that for there
to be a strong relation between measures of attitudes and be-
havior, the measures must becompatible(or congruent) in
terms of theirspecificity: Measures of general attitudes (to-
ward objects) predict general or broad behavior measures
(encompassing the class of relevant behaviors, also called
multiple act behavioral criteria), whereas measures of spe-
cific attitudes (toward behaviors) predict specific behavior
measures (the single, focal behavior). Single behaviors can
be specified along four dimensions: action (e.g., giving
money), target (e.g., to a homeless person), context (e.g., on
the street), and time (e.g., at lunchtime today). To predict
single behaviors maximally, the measure of attitude should
correspond on as many dimensions of specification as possi-
ble. For example, a measure of the individual’s attitude to-
wardgiving money to a homeless person on the street at
lunchtime todaywould be the best predictor of this specific
behavior, whereas measures of attitudes that corresponded
only on the action dimension (attitudes toward giving
money) or only on the target dimension (attitudes toward
homeless people) would rarely yield strong correlations.
Many early researchers inappropriately used general attitude
measures (e.g., participants’ attitudes toward an ethnic
group) to try to predict specific behavior measures (e.g., how
participants behaved toward a particular member of the eth-
nic group in a particular setting at a particular time). When
measures of attitudes and behavior have been highly com-
patible in terms of their specificity, attitude-behavior correla-
tions have been substantial (see Ajzen & Fishbein, 1977;
Kraus, 1995).
The impact of another kind of compatibility on attitude-
behavior consistency was investigated by Lord, Lepper, and
Mackie (1984). These researchers proposed a “typicality
effect,” such that attitudes toward a social group would pre-
dict individuals’ behavior toward typical members of the
group better than the same attitudes would predict behavior
toward atypical members of the group. Results showed that
individuals’ attitudes toward gay men predicted how they be-
haved toward a gay man who closely matched the stereotype
better than the same attitudes predicted how they behaved
toward a gay man who differed substantially from the stereo-
typical image of gay men. Thus, compatibility between group
stereotypes and individual group members influences
whether attitudes toward the group predict behavior toward
those individual members (see also Blessum, Lord, & Sia,
1998).Nature of the BehaviorCertain kinds of behavior are more predictable from attitudes
than are other kinds of behavior. In particular, attitudes are
hypothesized to guide onlyvolitionalactions—behaviors that
individuals are free to perform or to not perform. When
strong external incentives or constraints exist regarding an
action, attitudes may not play much role in determining be-
havior. For example, politeness norms may cause people to
say hello to coworkers whom they dislike. This conceptual
point—that social pressures often guide behavior—was rec-
ognized in the theory of reasoned action (Fishbein & Ajzen,
1975) by including subjective norms as a determinant of be-
havioral intentions that was distinct from attitudes.Subjective
normsrefer to individuals’ perceptions that other people who
are important to them want them to act in certain ways.
Researchers have identified several factors, including the
nature of the behavior, that influence the degree of impact
that attitudes and norms exert on behavior. For example,
Ybarra and Trafimow (1998) showed that increasing the ac-
cessibility of individuals’private selfcognitions (i.e., assess-
ments of the self by the self) led participants to place more
weight on attitudes than on perceived norms in behavioral
choices, whereas increasing the accessibility of individuals’
collective selfcognitions (assessments of the self by other
people and reference groups) led participants to place more
weight on perceived norms than on attitudes in behavioral
choices. Presumably, these findings reflected that attitudes
derive from personal preferences, whereas norms derive
from other people.
Ajzen (1985, 1991) proposed a revision to the theory of
reasoned action, which he labeled the theory of planned
behavior(for a review, see Conner & Armitage, 1998). This
model includes perceived behavioral control as another
determinant of intentions and behavior, distinct from both
attitudes and subjective norms. The construct of perceived