By DAVID WILLIAMSON
UNC-CH News Services
CHAPEL HILL -- Telephone surveys suggest that during an average week, about 40 percent of U.S. residents attend church, a figure at odds with head counts indicating only about half that many actually attend. Now, a new University of North Carolina at Chapel Hill study shows the truth likely lies somewhere in between.
Almost 30 percent of people in this country are weekly churchgoers, a UNC-CH religion scholar says.
"If actual church attendance is only half the level reported on surveys, as some researchers have maintained, that would have serious implications for our understanding of life in the United States," said Robert Woodberry, a doctoral student at UNC-CH. "If surveys misquote church attendance by such a large margin, they may misreport other behaviors and attitudes by similar amounts.
"Unfortunately, such errors are difficult to detect because we have few reliable measures of behavior to compare with survey estimates," Woodberry said. "The result would be that all kinds of survey-based research would be suspect."
A report on the findings, titled "When Surveys Lie and People Tell the Truth: How Surveys Oversample Church Attenders" appears this week in the American Sociological Review, a professional journal.
A 1993 article in the same journal questioned the traditional 40 percent attendance figure. Authors suggested it was twice as high as it should have been because of "social desirability bias," the tendency of people to over-report -- often unconsciously -- behaviors they think interviewers will consider positive. Or people surveyed interpret the question to mean, "Are you a good Christian"?
Woodberry found that social desirability bias was not a major factor. Telephone polls like Gallup overestimate church attendance because regular churchgoers are easier to contact and more cooperative. He also found methods of counting Protestants going to worship services to be flawed. Rather than using confirmed head counts, they relied on letters from ministers or counting cars in church parking lots.
Catholic attendance data tended to be significantly better, but still left out, for example, such groups as Eastern Rite Catholics, mass attendance at shrines and students attending mass at Newman centers on college campuses.
"My research shows that when you adjust to correct for the people who aren't included in head counts and use more accurate estimates of the Catholic and Protestant population in each region, attendance figures come up substantially," he said. "At the same time, most polls oversample regular church attenders and so overrepresent their opinions.
"Fortunately, survey sampling problems are easier to correct than social desirability bias."
The Religious Research Association and the Society for the Scientific Study of Religion supported the investigation.
- 30 -
Note: Woodberry can be reached at (919) 962-8756 (w) or 969-7583 (h).
Contact: David Williamson, 962-8596.
Journal
American Sociological Review