Vol. 1, No. 2
to other articles
in this issue:
A New Establishment?
Race and Disgrace
Submission in Salt Lake
Catholic Controversy I: Jesus Off Broadway
Catholic Controversy II: Handling
Catholic Controversy III: Philadelphia
On the Beat: "Irreligion" in Denmark
The Oklahoman's Bible Belt
Lies, and Polling Data
by Andrew Walsh
Americans are a religious people who attend worship services two, three, and even
four times more frequently than Western Europeans. Thats been an article of
sociological and journalistic faith for 50 years.
It has been built on a mighty foundation of polling data. Since the late
1930s, the Gallup Organization has been asking pollees if they "happened to
attend" church or synagogue in the past seven days. Invariably, about 40 percent
respond that they have done so. Long running surveys like the General Social Survey of the
National Opinion Research Center at the University of Chicago, the Harris polls, and the
polling of the Barna Research Group in California have tended to support the 40 percent
This number is so commonplace that when polling data on church
attendance is released, as it is several times each year, American journalists usually
relegate it to news notes or use it as a springboard to other stories. "Nearly all
surveys of American churchgoing habits show that roughly 40 percent of Americans attend
church once a week," Ari Goldman reported in a typical piece, a 1991 New York
Times story that went on to show that the growth of some churches occurs at the cost
But theres a problem with the 40 percent mantra: It may grossly
exaggerate how often Americans actually attend religious worship. For much of the 1990s,
contending groups of sociologists have competed, usually outside the view of journalists,
to revise and improve their measurements. And many of them have concluded that the
"fact" of high church attendance is built on a foundation of lies-or, more
The underlying issue is that for the past half-century Gallup and other
polling enterprises have relied on self-reporting to determine attendance at worship.
Unfortunately, "theres a well-known tendency for individuals in self-report
surveys to exaggerate what they perceive to be socially desirable behavior," Mark
Chaves, now at the University of Arizona, told Richard Chapman of the Chicago
Sun-Times in December of 1994. No behavior is more "socially desirable"
than church attendance, but pollsters have rarely addressed this shortcoming in their
In 1993, Chaves, Kirk Hadaway of the United Church of Christ, and Penny
Marler of Samford University ignited the debate about "overrepresentation" by
reporting the results of a study of church attendance by Protestants in Ashtabula County,
Ohio, and in 18 Roman Catholic dioceses around the country. Instead of using telephone
polling, the researchers counted heads at services and in parking lots, and checked with
pastors. They then estimated that 20 percent rather than 40 percent of Protestants, and 28
percent rather than 50 percent of Catholics, attend church weekly. The study, "What
the Polls Dont Show: A Closer Look at U.S. Church Attendance," appeared in the American
Scholars whose work has relied on the polling data were particularly
unhappy with the challenge. The Rev. Andrew Greeley-Catholic priest, novelist, columnist,
and sociologist-was incensed by the study, denouncing it in a Religion News Service column
as "a sloppy piece of work" that naively extrapolated regional findings into
national ones. Hadaway, Chaves, and Marler responded that they werent attacking the
validity of poll data, merely pointing out its sharp limitations. "Americans
misreport how often they vote, how much they give to charity, and how frequently they use
illegal drugs. People are not entirely accurate in their self-reports about other areas as
well," Hadaway wrote in the magazine Christian Century. "Males
exaggerate their number of sexual partners, university workers are not very honest about
reporting how many photocopies they make. Actual attendance at museums, symphonies and
operas does not match survey results. We should not expect religious behavior to be immune
to such misreporting."
After the first wave of scholarly discussion, Hadaway et al.
returned to Ashtabula County to measure Catholic attendance. They counted heads at all of
the regularly scheduled masses in the county-38 in 13 parishes-over a several-month
period. Based on the count, they projected an average weekly attendance of 24 percent of
the Catholic population (a figure not far out of line with numbers reported by many
Midwestern Catholic dioceses based on their own head counts). They then polled a
scientifically valid sample of Ashtabula County residents by telephone. Fifty-one percent
of Roman Catholic respondents said they had attended church during the past week.
Thus the "overstatement gap" snapped into focus. In the United
States, the difference between attendance levels of 20 and 40 percent is immense-a swing
of at least 50 million people. Institutional religion, far from being stable and vital in
the United States, might well be weakening under the cover of misleading poll data. Any
way one looks at it, there was a substantial religion news story to cover.
And in the fall of 1993, the academic dispute over levels of church
attendance did gain some coverage in the American newspapers. "A high-level debate,
spiced with a dash of bare-knuckle language, has erupted in some academic circles over
claims by three scholars that church attendance in the United States, as established by
decades of telephone polls, has been heavily inflated," reported a September 18, 1993
story in the Los Angeles Times. "If the revised rate proves true, it would
have significant implications for assumptions about American religious practice that have
long undergirded articles in popular and academic journals."
Reporting that "academic sniping" had broken out, the Atlanta
Constitution quoted Greeley attacking the study and another sociologist, Jay Demerath
of the University of Massachusetts, praising it for bringing out the fact that Americans
inflate their religiosity in traditional polls. "Gallup and other pollsters are aware
of this," Demerath said. "Its kind of a dirty little secret."
But despite the potential significance of the dispute over church
attendance, few news organizations covered the story vigorously. Most of the published
reports were drawn from wire services. Only a few bylined stories discussed the issue in
1993 and 1994.
During the middle of the decade, journalists did give increasing
attention to pollsters whose analyses of church attendance were more complex than
Gallups-especially to the work of George Barna, an outspoken evangelical Protestant.
Coverage of Barnas more volatile and nuanced surveys increased dramatically in this
period, although journalists still ignored the overrepresentation issue. The Barna polls
tended to portray falling rates of church attendance (although from levels closer to 50
percent in the early 1980s to the middle-high 30 percent range).
More concretely, Barna suggested that significant changes in belief and
practice had been obscured by "macro-level" statistics such as Gallups. In
1996 and 1997, the Orlando Sentinel, the Los Angeles Times, the Kansas
City Star, the Washington Post, the San Diego Union, the Chicago
Sun-Times, the Denver Post, and the Houston Chronicle all carried
substantial Barna-based stories suggesting declines in religious practice, significant
regional variations in religious practice, a slowdown in the growth of evangelical
Protestantism, and shifting popular understanding of the religious content of many
questions pollsters ask.
In 1997, Bill Broadway of the Washington Post quoted Barna as
saying that Gallups statistics are valid but fail to "show the subcurrents of
change." For example, Barna said he believed that in 1947, the date of a landmark
Gallup poll, "the vast majority of people believed in a God described in the
Bible." By the 1990s, he claimed, deeper probing revealed that one third of those who
tell pollsters they believe in God do not believe in the Biblical God, embracing instead a
higher consciousness, or a sense of the divine derived from Eastern religions,
or simply many gods."
Meanwhile, beneath the radar of most journalists, the sociological mill
was grinding away. Researchers at the University of Michigans Institute for Social
Research and others began to explore the claims made by Chaves, Hadaway, and Marler. In
general, these studies found strong support for the "overrepresentation thesis."
Follow-up studies have usually focused on who is overrepresenting their
church attendance to pollsters. Perhaps counter-intuitively, scholars now suggest that it
is the most committed believers who overstate their attendance, not those who
seldom or never attend services.
Some poll respondents stretch the boundaries of truth because they feel
that they usually attend weekly worship, even if they didnt happen to do so the week
the pollster called. "Follow-up questions about what people meant by attending
church revealed that a few were counting things other than attending worship-such as
going to weddings, funerals, committee meetings, Sunday school and choir practice,"
Hadaway wrote in the Christian Century. "One individual in Ashtabula County
even said his attendance consisted of mowing the church lawn on the previous
Many of those who overreport, it seems, attend church once or twice a
month but dont rate themselves as less committed. "Regular church attendance is
increasingly difficult, even for those committed to it," Hadaway wrote. "Sunday
morning is no longer sacred time: job responsibilities, sports leagues, family
outings, housework, and many other things get in the way of traveling to a church building
for worship at a scheduled time."
After several quiet years in the media, the debate about church
attendance resurfaced this year. In February, Marler, Hadaway, and Chaves published an
aggressive article, "Overreporting Church Attendance in America: Evidence that
Demands the Same Verdict," in the American Sociological Review. It asserted
the nationwide validity of their estimate of average weekly church attendance at 20
percent. "We believe that too much trust has been placed in survey data and not
enough attention given to membership records, patterns of giving, and even the incredulity
of local church pastors when they hear that 40 percent of Americans attend church during
an average week," Hadaway told the Minneapolis Star Tribune.
Other scholars are still not persuaded. "Theres a claim that
surveys lead to overreporting of church attendance, which seems to be correct. The
question is by how much," Thomas Smith of the National Opinion Research Center told
David Briggs of the Cleveland Plain Dealer in May. "We havent nailed
down how much Americans exaggerate."
Februarys forum in the American Sociological Review
included reactions from many prominent sociologists of religion, most of whom expressed
skepticism about the 20 percent figure and the research methods used. Theodore Caplow of
the University of Virginia, for example, said he thinks the "overrepresentation"
thesis is "not proven."
As estimates proliferate, there does seem to be "more consensus
around a figure of 30 percent than there is on a figure of 20 percent," Smith told
the Plain Dealer. (This would project weekly attendance at worship at about 75
In May, two new voices entered the debate with new data and a new
methodology. A Washington Post article by Richard Morin reported that research by
Stanley Presser of the University of Maryland and Linda Stinson of the U.S. Bureau of
Labor Statistics indicated that church attendance has been falling significantly and that
the number of Americans who lie about their church-going habits is increasing.
Working with three sets of time-use diaries produced by Americans
participating in social scientific research projects in the mid-1960s, 1970s, and 1990s,
Stimson and Presser "determined that the percentage of Americans who attended church
the previous week plummeted from 42 percent in 1965 to 26 percent in 1994." They
argue that their study of time-use diaries avoids many of the problems of respondent
"social desirability bias" because "those in the diary study were asked
only to account for how they spent their time, and not whether they went to church."
Indeed, the 1992-94 diarieswere produced in a U.S. Environmental Protection
Agency-sponsored study designed to determine how often individuals were exposed to harmful
The Spring 1998 debate triggered substantial articles in, among others,
the Plain Dealer, the Dallas Morning News, the Hartford Courant,
and the Christian Science Monitor, but few journalists explored the local angles
of the story or considered the implications of a substantially diminished and diminishing
sector of institutional religion.
Despite the complexities of the discussion, it does matter whether
weekly attendance at worship is 20 or 30 or 40 percent of the population. Here are several
of the questions that need to be sorted out.
- If attendance is at the 20 percent level, we are due for a thorough
reassessment of long held views about the strength of organized religion in the United
States. If the emperor has no clothes, or is more scantily clad than thought heretofore,
thats not only important to know for its own sake but may also be a sign that many
other perceptions are distorted by an exaggerated emphasis on polling data.
- If attendance turns out to be 30 percent or more, the entire debate may
be a tempest in an academic teapot. But even such a figure would call for vigorous efforts
to determine whether participation rates are changing or whether we have simply developed
more accurate measures.
- One emerging school of thought holds that much of the aggregate decrease
in participation rates may be caused largely by changes in the practices of Catholics, who
make up the largest American religious group. Theres considerable evidence that
Catholic rates of participation have been moving rapidly downward toward Protestant norms.
If so, the "Protestantization" of American Catholicism since the mid-1960s could
be the most undercovered religion story around.
- Stories about church attendance ultimately relate to how ordinary
Americans spend their time and energy-a major preoccupation of American journalists in
recent years. If the new studies prove correct, they might indicate that committed church
members feel that its impossible to attend services weekly even though they would
like to. That would suggest an attenuation of institutional commitment that aligns in an
interesting way with patterns of declining voter participation and less activity in other
realms of the voluntary sector.
- The persistence of claims of 40 percent or higher church-attendance rates
may suggest that even committed believers are drifting off into a private or detached
realm of religious observance. But if so, why do Americans continue to tell pollsters
their attendance levels are high when direct social pressures to conform and attend
worship are far weaker in most parts of the country than they used to be? Is this
insistence on exaggerating participation a sign of the strength or weakness of religion in
However and whenever these and related questions are resolved,
journalists would do well to break one habit right away. To this day, many news outlets
continue to report the latest release of polling data with no reference to the dispute
over its value. Its almost as if reporters arent reading their own clip files.
On May 31, 1997, for example, the Washington Post carried a
story by religion writer Bill Broadway under the headline: "Poll Finds America
as Churched as Ever; Belief in God, Afterlife Have Changed Little Since 1947,
but Faithful Sample More Forms of Spirituality." The story quoted George H. Gallup,
Jr. of the Princeton Religion Research Center saying that church attendance was stable
"despite the church-hopping and spiritual experimentation common among American
Unqualified Gallup poll data was still cited high up in arguments about
the persistent and even increasing role of religion in American life. "Religion is
breaking out all over," Bill Moyers wrote in USA TODAY on October 13, 1996.
"Public confidence in both organized religion and the clergy has been renewed. Church
attendance has held steady and in some instances-among teens for example-is up
On June 28, 1997, a story reported by Dan Lothian on NBCs Nightly
News carried the headline: "Nationwide Increase in Church Attendance Creates a
Building Boom for Contractors and Banks." In this case, the significant indicators of
growth were bank loans and construction contracts for new synagogues, Catholic churches,
and (especially) suburban megachurches. Charles Arn, identified simply as "a church
growth researcher," asserted: "Across the country, we are seeing an increase in
Dont bet on it.