308 Attitudes in Social Behavior
change is the selected route to imbalance reduction, Heider
did not indicate whether the attitude toward O or the attitude
toward X is more likely to change.
Osgood and Tannenbaum’s (1955) congruity theory ad-
dressed this latter issue by proposing that attitudes toward both
O and X would change in the face of imbalance. In addition,
these researchers predicted that the amount of attitude change
would depend on the extremity of each attitude, such that the
more extreme attitude would change the least. (Balance the-
ory did not consider the role of attitude extremity.) These
predictions have received some support (e.g., Tannenbaum,
1966), with important exceptions (e.g., Tannenbaum &
Gengel, 1966).
Relations Between Attitudes, Values, and Ideologies
Attitudes and Values
Not only are different attitudes interconnected, but they may
also be related to other, higher-order constructs such as val-
ues. The capacity of attitudes to express values is highlighted
by theories describing the value-expressive function of atti-
tudes (e.g., Herek, 1986; D. Katz, 1960) and by measures that
specifically include value-relevant beliefs in the assessment
of attitude components (e.g., Esses et al., 1993). In addition,
Rokeach’s (1973) seminal theory of values emphasized the
role of values in driving attitudes. He suggested that a rela-
tively small set of social values underlie most attitudes. Con-
sistent with this reasoning, rankings of the importance of
values have been shown to predict a large variety of attitudes
and behavior (e.g., Maio, Roese, Seligman, & Katz, 1996).
Moreover, priming a value makes accessible a variety of
value-relevant attitudes, but priming value-relevant attitudes
does not make accessible a variety of values (Gold &
Robbins, 1979; Thomsen, Lavine, & Kounios, 1996), sug-
gesting that values are above attitudes in the hierarchical
network of attitudes, beliefs, and values.
The potential centrality of values is also reflected in
Rosenberg’s (1960, 1968) evaluative-cognitive consistency
theory. According to this theory, people strive for consis-
tency between their attitudes and social values. This pursuit
of consistency is similar to the pursuit of balance in P-O-X
triads. Specifically, people seek consistency across a series of
person-value-object (P-V-X) triads or bands. Each band
contains the person’s attitude toward the attitude object
(e.g., censorship), the person’s belief in the importance of a
particular value (e.g., freedom), and the perceived relation
between the attitude object and the value (e.g., censorship
threatens freedom). Moreover, for any given attitude object,
the number of bands equals the number of relevant values,
such that the bands differ only in their referent values (e.g.,
P-V 1 -X, P-V 2 -X, P-V 3 -X). Rosenberg (1960) suggested that
people are unlikely to restore consistency by changing per-
sonal values because each value can be relevant to many
attitudes. Thus, changing a value may balance triads for one
attitude object, but could also create imbalance in other
triads. Consistent with this reasoning, Rosenberg observed
that people were more likely to change their beliefs about the
relations between an attitude object and relevant values than
to change the values themselves.Attitudes and IdeologiesAttitudes may also express ideologies, which are clusters of
thematically related values and attitudes (Converse, 1964;
McGuire, 1985). Liberalism and conservatism are well-
known ideologies. Liberal ideologies encompass attitudes
and values that promote universal rights and benevolence,
whereas conservative ideologies encompass attitudes and
values that promote freedom and self-enhancement (e.g.,
Kerlinger, 1984).
If the liberal-conservative dimension is a valid means for
sorting political attitudes, then people should tend to endorse
either conservative attitudes or liberal attitudes, but not both.
Yet people’s actual endorsement of liberal and conservative
attitudes does not follow this simple pattern (Converse, 1964;
Fleishman, 1986). Multidimensionality is most evident
among people who lack expertise in political issues (Lavine,
Thomsen, & Gonzales, 1997; Lusk & Judd, 1988). Re-
searchers have found at least two distinct ideological dimen-
sions within political attitudes: attitudes toward moral
regulation versus individual freedom, and attitudes toward
compassion versus competition (e.g., Ashton, Esses, & Maio,
2001; Boski, 1993).
There has been recent interest in ideologies from re-
searchers examining nonpolitical attitudes as well. For exam-
ple, researchers in the area of intergroup attitudes have
examined several ideological dimensions, including multi-
culturalism versus color blindness (Wolsko et al., 2000) and
individualism versus communalism (I. Katz & Hass, 1988).
Diverse ideologies have also been examined in studies of at-
titudes toward gender roles (Spence, 1993), body weight and
obesity (Quinn & Crocker, 1999), ways of life (de St. Aubin,
1996), and violence (Cohen & Nisbett, 1994).
At present, there is little evidence documenting precisely
how attitudes express broad values and ideologies. For exam-
ple, values may occasionally function as post hoc justifica-
tions for attitudes, rather than as their psychological basis
(Kristiansen & Zanna, 1988). When causal influences of val-
ues and ideologies do occur, the effects may be indirect or
direct. In an indirect effect, values and ideologies influence a
specific attitude indirectly through other attitudes, whereas a