Structure of Attitudes 301Nonetheless, such measures do not utilize the three-
component model as rigorously as do measures that assess
directly the attitude-relevant beliefs, feelings, and behaviors.
Breckler (1984) provided an excellent example of the direct
assessment of attitude-relevant beliefs, feelings, and behav-
iors. His research measured attitudes toward snakes and used
a variety of verbal and nonverbal indicators. The verbal mea-
sures asked participants to rate (using self-report scales) their
beliefs, feelings, and past behaviors toward snakes. The non-
verbal measures assessed attitude-relevant affect and behav-
ior, using recordings of participants’ heart rate and behavior
in the presence of a live snake. The verbal and nonverbal
measures for each component were then aggregated to form
overall indices for each attitude component.
Open-ended measures offer another method for assess-
ing the three components of attitudes. These measures ask
participants to list their beliefs, feelings, and behaviors re-
garding the attitude object (Esses & Maio, 2002; Haddock &
Zanna, 1998). Participants then rate the valence of each
response by using a semantic-differential scale. This approach
makes it necessary for respondents to indicate responses
that are accessible to them, rather than simply rate agree-
ment with responses that the researcher presents (Esses &
Maio, 2002).
Evidence In support of the three-component model, re-
search has found that people’s beliefs, feelings, and behav-
iors toward an attitude object are correlated but distinct. For
example, Breckler (1984) found that people’s beliefs, feel-
ings, and behaviors toward snakes were moderately corre-
lated when the components were assessed using verbal and
nonverbal measures in a context in which a snake was pre-
sent. His use of verbal andnonverbal measures provides a
good test of the three-component model, because this tech-
nique corrects for the systematic measurement error that
would occur if either technique were used alone. (In fact, the
components were highly intercorrelated when verbal items
alone were used in the absence of a snake.)
Using primarily pen-and-paper measures, additional re-
search has examined the distinction between the cognitive
and affective components, and such research has found mod-
erate correlations for attitudes toward a large variety of ob-
jects (e.g., birth control, blood donation, microwaves;
Breckler & Wiggins, 1989; Crites, Fabrigar, & Petty, 1994;
Haddock & Zanna, 1998). Further, Trafimow and Sheeran
(1998) found that attitude-relevant feelings and beliefs were
clustered separately in memory.
Given the evidence that the cognitive and affective com-
ponents are distinct, attitudes in different domains may be
uniquely related to one or the other component. Consistent
with this prediction, cognitive responses are strong predictors
of attitudes toward a variety of controversial issues (e.g.,
capital punishment, legalized abortion, nuclear weapons;
Breckler & Wiggins, 1991; Crites et al., 1994), whereas
affective responses are strong predictors of attitudes toward
blood donation (Breckler & Wiggins, 1989), intellectual
pursuits (e.g., literature, math; Crites et al., 1994), smoking
(Trafimow & Sheeran, 1998), and politicians (Glaser &
Salovey, 1998).Belief-Based AttitudesGuiding Assumptions It is also possible to view attitudes
as evaluative responses to an object that are influenced by be-
liefs alone (e.g., McGuire, 1960; Wyer, 1970). From this per-
spective, it is important to understand exactly how beliefs are
interrelated and how beliefs are linked to affective responses.
For example, a message might argue that it is good to reduce
waste, and therefore that people should recycle waste. The
message is persuasive if the message recipient accepts both
the premise of the argument (i.e., reducing waste is good)
and the implied link between the premise and the conclusion
(i.e., recycling will reduce waste). Notice that the evaluative
nature of the premise (reducing waste is good) introduces an
evaluative bias into the conclusion—that is, people should
become more favorable toward recycling because of its
desirable implications for reducing waste. In this manner,
attitudes can be evoked by beliefs (i.e., premises) that are
evaluative in nature.
The notion that attitudes reflect the acceptance or rejec-
tion of evaluative premises is central to the well-known
expectancy-value perspective on attitudes (e.g., the theory of
reasoned action;Fishbein & Ajzen, 1975). According to this
approach, an attitude is the sum of all of the evaluative beliefs
regarding the attitude object. For instance, if people believe
that recycling is easy and that recycling helps the environ-
ment, people should hold a positive attitude toward recy-
cling. This attitude is positive because both beliefs link
positively valued attributes to the behavior. Of course, beliefs
are rarely held with absolute certainty. For example, a person
may be only 70% certain that recycling is easy, but also be
100% certain that recycling helps the environment. Accord-
ing to the expectancy-value model, beliefs have less impact
on attitudes when they are less certain. This reasoning is fre-
quently summarized in a well-known equation: Abiei,
whereAis the total attitude toward the attitude object, biis
the subjective belief that the object possesses attribute i(e.g.,
the probability that recycling helps the environment), and ei
is the evaluation of attribute i(e.g., the positive value at-
tached to the environment).