Religion has long been a critical factor in American elections.
But until recently its impact has been felt in terms of affiliation—membership
in specific denominations and other religious bodies. For example, northern
mainline Protestants have traditionally voted Republican, while Jews have
been solidly in the Democratic camp.
In recent years, however, another aspect of religion has become
politically significant: commitment, or the extent to which citizens
are actively engaged in their faith. Americans who say they attend worship
at least once a week are more likely to vote for Republicans, while less
regular attenders—including those who claim to be nonreligious—gravitate
toward the Democrats.
During the past decade, this new “religion gap” between the two
parties has grown dramatically. It can be measured in various ways, but
perhaps most clearly by looking at the party preferences of frequent worship
According to Voter News Service (VNS) exit polling, in the 1992
congressional election, frequent worship attenders preferred Republican to
Democratic candidates for the House of Representatives by 53 to 47 percent.
By the 2002 congressional election, this six percent gap had ballooned to 20
percentage points, with frequent attenders voting in favor of Republican
House candidates by 60 to 40 percent.
The bulk of this shift occurred between 1994, when the gap was
eight percent, and 2000, when it was 18 percent. Unfortunately, VNS did not
ask a worship attendance question in either 1996 or 1998, so it is not
possible to track the growth of the religion gap more precisely over this
How to account for that growth? One possible explanation is the
1997 Lewinsky affair that culminated in the impeachment and subsequent
acquittal of President Clinton. Although the country as a whole did not
favor removing Clinton from office, a sizable number of frequent worship
attenders may have been sufficiently distressed by the affair to change
their voting habits.
In this regard, it is worth noting that in 1992 frequent
attenders were considerably more likely to vote for Democratic House
candidates than they were for Democratic presidential candidate Clinton, who
at the time was dogged by charges of womanizing. In contrast to the six
percent religion gap in the 1992 congressional vote, frequent attenders
preferred President George H. W. Bush to Clinton by a margin of 14 percent.
However, in the 2000 presidential contest frequent attenders
chose George W. Bush over Vice President Al Gore by 20 percent—just two
percentage points higher than the religion gap in 2000 congressional voting
and identical to the 2002 religion gap. It is conceivable that, in the late
1990s, an increasing number of frequent attenders transferred their moral
disapproval of Clinton onto his party as a whole.
However we explain the religion gap, at the dawn of the 21st
century it is substantially larger than the widely discussed “gender
gap”—the tendency of women to vote Democratic more than Republican. In the
2000 congressional vote, where the religion gap was 18 percent, women
preferred Democrats to Republicans by just 10 percent.
None of this is to say that the old politics of religious
affiliation has ceased to operate. On the contrary, affiliation, which is
closely tied to ethnic and racial identity, continues to have profound
effects on American voting patterns. But these effects are now being shaped,
within many religious voting blocs, by a new politics of commitment. Only by
taking both aspects of religion into account can we assess the full impact
of religion on electoral politics.
In order to suggest how religion may matter in 2004, we first
need to look more closely at how religion has figured in the past two
national elections. The 2000 and 2002 VNS polls, which rely on interviews
with thousands of respondents, asked three questions about religion:
membership in the “Religious Right”; general religious affiliation
(Protestant, Catholic, Other Christian, Jew, Something else, and none); and
frequency of worship attendance. The responses enable us to differentiate 13
major religious categories based on affiliation and commitment.
They are listed, according to the degree of support for the Republican
Party, in Tables 1 and 2.
To be sure, these are crude groupings that
mask important denominational identities. The great bulk of those in the
“Religious Right” category are white evangelical Protestants, including
members of the Southern Baptist Convention, Pentecostal, and Holiness
churches. The “Frequent Attending White Protestants” category covers many of
these evangelicals as well as more committed members of mainline Protestant
“Other White Christians” is a polyglot
category that includes some evangelicals belonging to nondenominational
megachurches as well as Mormons, Christian Scientists, and Eastern Orthodox.
Most of these denominations tend to have relatively high levels of religious
commitment, and like the frequent attending white Protestants tend to vote
As indicated in Table 1, George W. Bush won
all but one of the white Christian groups in 2000. Not surprisingly, he did
exceptionally well with the frequent attending white religious right (87
percent), which now represents the heart of the Republican coalition. With
non-Catholic Christians, Bush’s performance ranged from 61 percent of
frequent attending white Protestants to 51 percent of less attending white
Frequent attending white Catholics were one
of his strongest sources of support, giving him 57 percent of their vote
compared to 40 percent for Gore. Among white Christians, only the less
attending Catholics preferred Gore to Bush, by a margin of four percent. All
in all, differences in commitment were highly significant throughout the
For his part, Gore did especially well among
minority faiths, drawing an astounding 91 percent of the black Protestant
vote, 78 percent of the Jewish vote, and 67 percent of the Latino Catholic
vote. Among Latino Protestants, he eked out a two percent margin over Bush.
In contrast to whites, it appears that frequency of worship attendance has
little effect on the voting behavior of minorities, and especially of
The 2000 congressional vote, in which
Republicans retained control of the House of Representatives by a small
margin and ended up tied for control of the Senate, broke down along much
the same lines as the presidential vote, with a small but telling religious
variation. All the frequent attending Christian groups voted slightly less
for GOP congressional candidates than they did for Bush, while all the less
attending Christian groups did just the opposite. Meanwhile, Jews and
seculars were significantly more likely to vote Republican for Congress than
for President. It seems likely that candidate Bush’s religiosity gained him
some support among committed Christians while losing him votes elsewhere.
The 2002 polls tell much the same story, but
with some intriguing variations.
In a midterm election in which the president’s party, uncharacteristically,
picked up seats in both the House and Senate, non-Catholic Christians and
Jews moved towards the Republicans, while Catholics of every kind shifted in
a Democratic direction. Notably, Latino Protestants, who had shown a small
preference for congressional Democrats in 2000, moved decisively into the
Republican camp in 2002, voting GOP by a margin of 14 percent. And while
Jews continued to be a solid Democratic constituency, they increased their
vote for Republican House candidates from 27 to 35 percent.
On the other side, frequent attending white
Catholics lowered their margin of preference for Republicans from 14 to 8
percent, while Latino Catholics increased their pro-Democratic margin by
five percent. Less attending white Catholics swung from preferring
Republicans by one percentage point in 2000 to preferring Democrats by nine
percent in 2002.
How to make sense of these shifts? Robust
support for Israel and the war in Iraq doubtless helped round up more Jewish
votes than usual for the GOP. Otherwise, the faith-based programs and
rhetoric emanating from Congress and the White House seem to have appealed
largely to Protestants and “other” Christians—but not to Catholics. Why?
The Republican “preferential option for the
well-to-do” embodied in tax policy is certainly at odds with Catholic social
teachings, and in addition, many Catholics are leery of the evangelical
Protestant style increasingly associated with Republicanism. Such concerns
may well have blunted the Republican appeal on moral issues like abortion or
It is by looking at these particular groups
that we can understand why the overall “religion gap” increased by only two
percent during the first two years of the very religiously focused Bush
administration. The Republican gains among frequent attending white
Protestants and several minority Christian faiths were largely offset by the
Democratic shift of Hispanic and frequent attending non-Hispanic Catholics.
Of course, one major difference between the
2002 and 2000 elections is turnout: Far more people in every religious group
vote in presidential elections. Unfortunately, exit polls are of little help
in assessing turnout, since nonvoters are not included. In fact, turnout is
one of the most difficult things to measure accurately with any kind of
However, the available evidence does suggest
enormous variation in turnout by these religious groups. Frequent attending
white Protestants and Catholics, as well as Jews, tend to have the highest
turnout. Latino Catholics, less attending white Protestants, and Catholics
tend to have the lowest turnout. Seculars, black Protestants, and the white
religious right fall somewhere in between.
Moreover, turnout can vary on a regional
basis. We know, for example, that in 2002 frequent attending white
Protestants turned out in significantly higher numbers in the South than in
other regions of the country—and that this fueled impressive GOP victories
south of the Mason-Dixon line (see John Green, “The Undetected Tide,”
Religion in the News, Spring 2003.
What does all this mean for the 2004
election? The basic patterns of affiliation and commitment are unlikely to
change dramatically. The Republican alliance of white Protestants and
committed other Christians will once again face off against the Democratic
coalition of minority faiths and the less committed. The religion gap will
probably not grow much, if at all; almost certainly it will remain large.
The electorate is so closely divided that even modest changes
in the votes and turnout of key constituencies could make a difference, and
neither party can take religious voting blocs of granted.
For Bush and the Republicans, the white religious right is a
critical constituency, and a high level of support must be maintained by
action on moral issues. But the GOP must also maximize support from other
elements of its alliance, which includes groups that may be turned off by
moral issue appeals. Frequent attending white Catholics are very important
to the Republican cause, and here the 2002 congressional vote offers some
reason for concern. At the same time, there may be an opening for the GOP to
increase its support among Latino Protestants and Jews.
What about the Democrats and whoever their presidential
candidate may be? Job One is to maintain high support among the minority
faiths: black Protestants, Latino Catholics, and Seculars. The Democrats
will need to recover their standing among Jews and to maximize support from
less attending Catholics and Protestants so as to neutralize the religion
gap. Anything they could do to make gains among regular worship attenders
would of course be a bonus. Expect a fierce contest for Catholics and
Latinos in 2004.
Accomplishing any of these partisan goals requires politicians
to know how religious communities are connected to politics. In general
terms, the politics of affiliation and commitment revolve around values,
cues, and contacts.
Religious communities are characterized by distinctive values,
and, depending on the issues under consideration, these can be a basis for
choosing between candidates and parties. They are also places where religion
and politics are regularly discussed—by clergy, lay leaders, and congregants
themselves—and such discussion gives cues for political choices.
Politicians target religious communities for mobilization based
on the patterns of values and cues; any such contacts can influence
political choices as well. In general, the more committed members of a
religious community tend to embrace the group’s distinctive values more
fully, are more aware of the available cues, are exposed to more political
contacts, and thus are likely to make different choices than their less
In short, while the religion gap provides a critical index of
recent American voting behavior, the electoral impact of religion is not one
story but many different stories.
What issues push buttons for the various kinds of Protestants,
Catholics, and non-Christians? Which cues are being given in churches,
synagogues, and mosques? What contacts are reaching the frequent worshipers
and missing the infrequent? Who is doing what to mobilize whom?
The answers in 2004 remain to be seen—and reported.
Religion and 2000 Presidential Vote
Frequent attending white “religious right”
87% 11 2
Frequent attending white Protestants 61% 37 2
Frequent attending white Catholics 57% 40 3
Other white Christians 57%
Less attending “religious right”
56% 38 6
Less attending white Protestants 51% 47 2
ENTIRE ELECTORATE 48% 48 4
Latino Protestants 48%
Less attending white Catholics
46% 50 5
Other Non-Christians 36% 54
Latino Catholics 30%
Black Protestants 9%
TABLE 2: Religion and Congressional
Rep Dem Othr
85% 14 1
white ”religious right”
60% 37 3
white “religious right”
Christians 62% 35
54% 42 4
Protestants 56% 42
48% 50 2
attending white Protestants 55%
attending white Catholics 53%
attending white Catholics 44%
49% 48 3
35% 61 4
32% 63 5
Catholics 28% 71
30% 67 3
Protestants 10% 89
12% 88 *
* less than one percent
The 2002 VNS exit poll had 17,872 cases over all, of which 8,188 were
asked the religion questions; the 2000 VNS exit poll had 13,259 cases
overall, and the religion items were available for 9,246 cases. In both
surveys, the margin of error was less than two percent. Details of
coding and estimations are available from John Green at
Because of technical problems, the 2002 VNS polls were not released
until September 2003.