Religion and Political Behavior 307
the Democratic Party. In other work, Brooks (1999) shows that family values have
become an increasingly important social problem, but that it is primarily religious
conservatives who express concern about it.
Other analysts have explicitly challenged the model. DiMaggio, Evans, and Bryson
(1996) examined changes in public attitudes toward a wide array of social issues and
found little support for the view that any significant polarization has occurred since
the 1970s. Davis and Robinson (1996) found that the gap between religious conserva-
tives and liberals is much smaller than often thought, limited to a handful of social
issues, and on economic issues religious conservatives are actually somewhat more
supportive of governmental action to secure greater equality than religious liberals.
New Evidence Using Relative Measures of Religious Cleavages
The recent investigations of the first author, in collaboration with Clem Brooks, explic-
itly sought to reconsider these five issues, as well as to develop some overall estimates of
the changing impact of religious groups on U.S. party coalitions (Manza and Brooks
1997, 1999, 2001). We briefly summarize this line of research here. Three advances
over earlier research on religion and politics defined the methodological contributions
of our research. First, analyses of the relationship between social groups and politi-
cal behavior that fail to employ statistical models that allow for distinctions between
trends influencingallgroups from those influencing onlysomegroups neglect impor-
tant information. Second, research on the social group foundations of political behavior
should include analyses of (a) group size and (b) group turnout, alongside group voting
patterns. The size of groups and their turnout rates will shape theimpactof group-based
alignments on major party electoral coalitions, a crucial way in which the interaction
between religious groups (who seek influence) and political parties (who seek votes)
takes place (see Manza and Brooks [1999: Chapter 7] for further discussion). Finally,
research on religious cleavages and political behavior in the United States should em-
ploy adequate measures of the cleavage itself. Although considerably less common than
twenty years ago, some analysts of religion and politics have persisted in failing to take
into account the divisionsamongProtestants as well asbetweenProtestants, Catholics,
Jews, and others.
Employing models embodying these principles, our investigations of the changing
contours of religion and political behavior in the United States suggested a number of
conclusions, some of which are consistent with the thrust of previous findings, and
others that challenge the conventional wisdom:
The religious cleavage as a whole has declined very modestly since 1960. The decline
is due solely to the shift toward the center of one group – liberal Protestants – and
thus does not reflect any societal-wide trend toward dealignment.
Liberal Protestants have moved from being the most Republican religious group in
the 1960s, to an essentially centrist position by the 1990s. This transformation has
overwhelmingly been driven by their increased liberalism on social issues.
Conservative Protestants havenotrealigned toward the Republican Party, in large
measure because they have always been Republican partisans in the period (since
1960) for which we have adequate measures. Much of the confusion about the
political preferences of conservative Protestants reflects a one-time shift toward