86 Mark Chaves and Laura Stephens
United States attends religious services on a weekly basis had, until recently, enjoyed
this status. This fact had been freely reported by historians and journalists. For ex-
ample, the religion column in a 1991 issue ofThe New York Timesbegan by stating,
“Nearly all surveys of American churchgoing habits show that roughly 40 percent of
Americans attend church once a week” (Goldman 1991). Additional evidence of the
wide acceptance of this statistic is found in introductory sociology and methods text-
books, which almost uniformly report the 40 percent figure in their chapters on religion
or survey research (see, for example, Babbie 1992: 398; Johnson 1992: 548; Kornblum
1991: 514; Luhman 1992: 414; Thio 1992: 393). This example, from a 1992 textbook,
is typical: “Forty-two percent [of Americans] state that they attended a church or syna-
gogue during the preceding seven days. During the last half century, these figures have
shown some consistency....[T]here has been virtually no change in the percentage
of Americans who attended services during the week before they were interviewed”
(Luhman 1992: 414).
This claim – that 40 percent of Americans attend religious services in any given
week – was based on remarkably stable results from surveys in which respondents are
asked to report on their own church attendance practices. The Gallup Organization, for
example, asks people: “Did you, yourself, happen to attend church or synagogue in the
last seven days?” In 1998, 40 percent of Americans answered yes to this question, with
Catholics showing higher rates of attendance (46 percent) than Protestants (42 percent)
(Gallup and Lindsay 1999). Similarly, the General Social Survey asks respondents “How
often do you attend religious services?,” coding their responses into a set of categories
ranging from “never” to “several times a week.” In 1998, the weekly attendance rate
implied by the distribution of responses to this question was 38 percent (Davis et al.
1998).
It now appears, however, that taking at face value the accuracy of individuals’ reports
of their own religious behavior gave us a misleading picture about levels of religious
participation. Hadaway et al. (1993) opened debate on this question by comparing the
rates of church attendance based on the self-reports of respondents with rates based
on observing and counting the number of people actually present at religious services.
They did two things.
First, they examined weekly attendance among Protestants in Ashtabula County,
Ohio. In response to a telephone survey of 602 randomly selected county residents,
35.8 percent of self-identified Protestants said they had attended religious services in the
past seven days, a number nearly identical to the weekly church attendance rate found
in a 1991 sample of all Ohio residents (Bishop 1992), and very similar to rates obtained
in national surveys. After using telephone books and newspapers, and driving every
road in the county to identify churches appearing in neither of those sources, Hadaway
et al. found 159 Protestant churches in Ashtabula county. Attendance rates from each of
the churches were obtained through denominational yearbooks, telephone interviews,
letters, and church visits. The result: Although 35.8 percent of Ashtabula Protestants
claimed to have attended church in the past seven days, only about 20 percent of
Protestants actually attend church on an average Sunday.
This pattern of substantial overreporting of church attendance is not peculiar ei-
ther to Protestants or to Ashtabula county. The second piece of research in this article
was an examination of weekly attendance rates among Catholics in eighteen dioceses
around the country. In national polls, about 50 percent of Catholics say they attend