Green and Mark Silk
In the current election cycle, the “religion gap” has
emerged as a new factor in American politics. In December and January, there
was widespread news coverage of survey data, published in these pages and
elsewhere, showing that citizens who claim to attend religious services
regularly are now 50 percent more likely to vote Republican than those who
say they don’t.#1
The data suggest that electoral politics in the United
States has turned into something of a religious conflict, pitting—at least
at the extremes—devout Republicans against secularist Democrats. In order to
understand the nature of this conflict, however, it is important to view the
religion gap through the lens of gender.
When gender as well as religion is taken into account,
the critical gap in partisan preference turns out to be between men who
report attending services once a week or more often (“regular attenders”)
and women who report attending less than once a week (“less attending”). The
rest of the adult population—regular attending women and less attending
men—are evenly divided, and could swing either way in 2004.
The religion gap and the better-known gender gap both
appear to be products of value conflicts that arose in the 1960s. Evidence
of this can be seen in Figure 1, which uses a variety of survey data to
trace the two gaps from 1936 to 2000.
In the New Deal era of the 1930s and the 1940s, neither
worship attendance nor gender was a particularly important factor at the
ballot box, but to the extent they mattered they showed the opposite of the
situation today: Women were moderately more Republican than men, while those
who attended religious services regularly slightly favored the Democrats.
Women’s partisan preference switched dramatically in
the 1964 election, going from six percent pro-Republican to four percent
pro-Democratic. This presumably reflected concerns about GOP presidential
candidate Barry Goldwater’s bellicose posture on the war in Vietnam and
perhaps, in addition, support for the Great Society programs of President
Over the next several elections women’s preferences
bounced around from moderately to marginally pro-Democratic. But since the
1980 Carter-Reagan election, when they favored Carter over Ronald Reagan by
seven percent, women have shown a solid and significant preference for
Democrats, reaching a high point of 13 percent for Bill Clinton over Bob
Dole in 1996. In 2000, the gender gap was about 12 percent.
As for the religion gap, its watershed political moment
was the 1972 McGovern-Nixon election, when regular attenders’ preference for
the Republican candidate leaped from one percent to 10 percent. This is
perhaps best understood as a choice by the traditionally religious in favor
of Nixon’s “silent majority” and against the countercultural McGovernites.
This pro-Republican religion gap shrank progressively
during the two Carter campaigns, stabilizing through the 1980s in the
mid-single digits. But then, in the 1992 election, it shot up from five
percent to 17 percent, and has grown slowly and steadily ever since. Between
the early and ongoing revelations of Bill Clinton’s personal immorality and
the GOP’s steadfast embrace of the “traditional family values” agenda of the
religious right, it is not hard to explain this persistent and growing gap.
Table 1 looks at the combination of worship attendance
and gender, revealing a strikingly symmetrical pattern.
Three-quarters of regular attending men voted for Bush, while
three-quarters of the less attending women voted for Gore. Just as
fascinating are the two groups in the middle. The regular attending women
and less attending men divided their votes almost evenly between Bush and
Each category represented a hefty slice of the
electorate in the 2000 election. The Bush-backing regular attending men made
up 21 percent of voters, while the Gore-supporting less attending women were
24 percent. The evenly divided categories were somewhat larger, with less
attending men at 26 percent and regular attending women—the plurality
group—at 28 percent. Given how narrowly divided the electorate now appears
to be, a small swing in just one of these large groups could determine the
outcome of the 2004 election.
What social realities lie behind these categories?
Regular attenders tend to be older than their less regularly attending
counterparts. They are also more likely to be married and live in the South.
Regular attending women are more likely to be homemakers. But the regulars
and less regulars differ little in education or income.
Religious identity matters more than any of the above
demographic differences. White evangelical Protestants and Mormons vote
strongly Republican; the default setting for black Protestants and Jews is
Within some groups there are no gaps: Black Protestants
and Jews vote Democratic regardless of attendance rates or gender.
Elsewhere, a single gap applies. In the 2000 election, women who described
themselves as “secular” were markedly more Democratic than men who did,
though both were together at the bottom of the worship-attendance scale.
The largest religious identity groupings do, however,
contain both gaps. No religious group supported Bush over Gore more strongly
than white evangelicals, yet while nearly 90 percent of regularly attending
male evangelicals voted for Bush, only 77 percent of their female
counterparts did. Regular attending mainline Protestants, the next most
favorable group for Bush, showed an astonishing gender gap of 36 percentage
points (92 percent for Bush among men versus 56 percent among women). Among
regular attending Catholics, who were less pro-Bush overall, the gender gap
was a much smaller 11 percentage points (63 percent for Bush among men
versus 52 percent among women).
Less attending men were substantially more supportive
of Gore, with Catholics the most Democratic (48 percent), followed by
evangelicals (39 percent) and then mainliners (26 percent). Among the less
frequent attenders the gender gap was also substantial: 21 percentage points
for Catholics, 20 for evangelicals, and 34 for mainliners.
Indeed, the most consistent voting pattern across these
denominational lines occurred among less attending women. Among them, 59
percent of the evangelicals, 60 percent of the mainliners, and 69 percent of
the Catholics gave their votes to Gore.
So much for the categories per se. In order to
understand the political dynamics of this complex system it is necessary to
look at cultural attitudes that cut across differences of gender,
attendance, and religious identity—and push voters in one partisan direction
or the other. These attitudes can be discerned from the survey data showing
voters’ views of prominent interest groups.
Political groups dedicated to promoting “traditional
family values” united men and women according to level of attendance. Thus,
58 percent of regular attending men and 52 percent of regular attending
women had a favorable view of the Christian Right. That contrasts with only
about one-fifth of the less regular attending men and women.
The same pattern held for views on pro-life and
feminist groups, with some gender-based nuance: Regular attending women were
modestly less “pro-life” than their male counterparts, while less regular
attending men were less “feminist” than the comparable women.
On the other hand, roughly 40 percent of both regular
and less attending men felt positively about the National Rifle Association.
But for both regular and less attending women the level of support was some
20 percentage points less favorable. Similarly, more than half of each
category of women held a favorable view of teachers’ organizations, while
both groups of men were some 15 percentage points less favorable.
These competing axes of cultural conflict help explain
the partisan breakdown of the four critical gender-and-religion categories.
Regular attending men voted strongly Republican in large part because of
their consistently conservative views on sexual morality, guns, and
education. Their political opposites, the less attending women, were just as
strongly Democratic because of their consistently liberal views on abortion,
guns, and public education.
The other categories were evenly divided at the ballot
box precisely because of their crosscutting values. Regular attending women
were pulled in a Republican direction by traditional morality, but their
worries about guns and education pushed them in a Democratic direction. For
less attending men guns and schools led toward the GOP—but hostility to
morals regulation pointed to the Democrats.
Because value conflicts create swing voters, regular
attending white Catholic and mainline Protestant women and less attending
white Catholic and evangelical men are likely to be up for grabs in 2004.
Journalists interested in gauging how well the respective campaigns are
doing with these swing groups might try interviewing at a few selected
locations between, say, 10 o’clock and 12:30 on Sunday mornings.
Check out The Home Depot and the local links, where
many Catholic and evangelical men are rumored to be found bright and early
on the Lord’s Day. If these weekend warriors like Kerry, look for a new
administration in 2005.
Then buttonhole women coming out of worship at Catholic
and mainline Protestant churches. If these church ladies are tilting toward
Bush, then the president will likely get a new lease on the White House.•
Once a Week?
Table 2 illustrates the relationships between
self-reported worship attendance and the 2000 presidential vote that
underlie the religion gap. It shows, first, that the relationship between
attendance and the vote varies almost in lockstep: More than two-thirds of
voters who claimed to attend worship more than once a week voted for Bush,
while those who reported never attending worship voted nearly as heavily for
The table also shows that the break between Bush
supporters and Gore supporters comes between “once a week” and “once or
twice a month.” This once-a-week break point is consistent with other
Why should there be such a difference in political
behavior between those who say they attend religious services once a week
and those fairly observant folks who say they attend once or twice a month?
The simple answer is that for most Americans weekly worship attendance is
normative: Telling a pollster that you attend weekly is a widely accepted
way of saying that you are a religious person.
In fact, recent sociological research has proven that
many Americans “over report” their worship attendance (just as they “over
report” their voting in presidential elections). For more than half a
century roughly 40 percent of Americans have told the Gallup poll that they
had been to worship services in the past week, but the real number now seems
to be in the mid-twenty percent range. So the religion gap is as much about
citizens’ religious self-perception, and self-presentation, as it is about
their actual participation in congregational life.
1See John Green and Mark
Silk, “The New Religion Gap,” Religion in the News (Fall 2003) <http://www.trincoll.edu/depts/csrpl/RINVol6No3/2004%20Election/religion%20gap.htm>.
2The data from 1936-1948 were
estimated from Gallup surveys made available by the Roper Center at the
University of Connecticut. Data from 1952-1996 were estimated from the
National Elections Study, provided by the Interuniversity Consortium for
Social and Political Research at the University of Michigan. The 2000 and
2004 data points come from surveys conducted at the University of Akron.
(Because the Republicans are in power at this writing, the gaps are
portrayed in net Republican advantage, with positive and negative signs, but
the same patterns would obtain if the signs were reversed and the figures
were portrayed as net Democratic advantage.) Further details available from
1 and much of the subsequent analysis uses the Third National Survey of
Religion and Politics, conducted at the University of Akron in 2000.
(Post-election sample weighted N=3000). This survey has detailed measures of