Church Attendance in the United States 87
church on any given Sunday. Hadaway et al. assessed the accuracy of this number by
comparing it to mass attendance data collected in many Catholic dioceses. In these
dioceses, parishes conduct a systematic count of attendees at every mass on a desig-
nated weekend. Sometimes counts are done several weekends in a row, in which case
the numbers from each weekend are averaged to estimate the number attending on
any given weekend. Hadaway et al. inflated the attendance numbers reported by dio-
ceses in order to account for the very few parishes whose attendance numbers were not
included in the diocese-wide counts. These adjusted counts became the numerator of
a count-based attendance rate for each diocese.
The denominator was an estimate of the number of Catholics living in the geo-
graphical area covered by each diocese. Hadaway et al. used a nationally representative
survey of religions affiliation that had a large enough sample to reliably estimate the
proportion of self-identified Catholics within each diocese (Kosmin 1991). The total
population of each diocese, drawn from the 1990 U.S. census, was multiplied by the
proportion of Catholics in each location to produce an estimate of the number of self-
identified Catholics in each diocese. At this point, a count-based church attendance rate
was calculated by dividing the adjusted attendance figures by the number of Catholics
in each diocese.
Again the results were clear. Catholic attendance at mass is substantially lower than
the 50 percent figure suggested by research based on self-reported attendance rates. Al-
though there was significant variation across dioceses, when the count data were aggre-
gated only about 28 percent of Catholics attended church on a weekly basis, again lead-
ing to the conclusion that church attendance rates are only about half what previously
existing data would lead one to believe. Chaves and Cavendish (1994) supplemented
this study by gathering data on a total of forty-eight Catholic dioceses, representing
approximately 38 percent of Catholics in the United States. The result was unchanged.
This conclusion – that weekly church attendance in the United States is about half
what the conventional wisdom held it to be, about 20 percent for Protestants and about
25 percent for Catholics – was criticized in several ways, none of which, in our view,
quite hit the target. Consider four of the criticisms, and the responses to them. All of
the responses described below are drawn from Hadaway et al. (1998).
One line of criticism takes issue with the construction of the denominator in
Hadaway et al.’s Catholic estimates (Caplow 1998). This criticism begins with the obser-
vation that more people identify as Catholics than are actively involved in parish life. As
described above, Hadaway et al. used the number of people identifying as Catholics as
the denominator in their calculation of the count-based church attendance rate within
each Catholic diocese. Since the number of people who are active enough to be on
the official rolls of Catholic parishes is smaller than the number of people who simply
identify themselves as Catholic, dividing the number of attenders by the number of
people actually registered at Catholic parishes rather than the number of people who
self-identify as Catholic would produce a higher weekly attendance rate – and a smaller
gap between self-reported and actual attendance rates.
However, this reduction in the gap between self-reported and actual attendance rates
ignores the fact that the high attendance rates from conventional surveys also are based
on the number of self-identified Catholics who respond to the survey. To use a different
denominator in a count-based rate would lead to comparing apples and oranges. It is
difficult to see what the point would be of using the number of registered Catholics