Researchers still puzzled over
discrepancies in attendance polls
___By Mark Wingfield
___Managing Editor
___HOUSTON--If you want to start a polite fight, put a group of religious researchers in a room together and ask them to discuss why traditional polling methods consistently overstate Americans' church attendance.
___Just such a conversation occurred at the annual meeting of the Religious Research Association in Houston this fall. Three panelists presented their theories on the subject, and then the audience got to chime in.
___In the audience, of course, were other researchers who had published different theories on the same problem.
___While some see the issue cutting to the very core of social research of all kinds, not everyone even agrees there is a problem.
___Hardly anyone had even suggested there was a problem until 1993, when a trio of two Southern Baptists and a Catholic took the unprecedented step of actually counting heads in every congregation in a small Alabama county one Sunday.
___That research by Kirk Hadaway, Penny Long Marler and Mark Chaves rocked the world of religious research because it called into question the results of Gallup telephone polls that have been taken for decades.
___Gallup studies have shown a consistent pattern of church attendance in the United States across the 20th century, hovering above 40 percent for both Protestants and Catholics. Those results were obtained through telephone surveys in which participants were asked if they had personally attended a church or religious service in the preceding seven days.
___The research by Hadaway, Marler and Chaves reported much lower rates of church attendance--20 percent for Protestants and 28 percent for Catholics--based on the actual head counts.
___Their explanation: People interviewed on the telephone about church attendance tend to paint a rosier picture than reality because it is socially desirable to say you attend church regularly.
___In research terms, this is called a response bias.
___And in this case, if response bias is the culprit, it is so big that it indicates a problem for all kinds of surveys, said Larry Iannaccone, a professor at Santa Clara University in California. "We might as well give up on all survey-based research," he told the group.
___Like many others, Iannaccone has his own theories of what's gone wrong: "This isn't a sign of response bias. It's just a sign of us researchers being stupid."
___He contends the basic Gallup question and others like it are flawed or the results gathered by such questions are misunderstood.
___"If I claim to attend church every week, what do I mean?" he asked. "That I never get sick? That I never go on vacation? That I never oversleep? ... No, I mean I go whenever I reasonably can."
___Researchers have applied a level of literal interpretation to church attendance data that they never would apply to other data, Iannaccone said.
___For example, if researchers asked people how often they go to work, respondents might answer, "Every day." But few people actually go to work seven days a week. What the answer really means is that person normally goes to work on every scheduled work day, he said.
___It is possible, Iannaccone suggested, that people give answers to the Gallup question based on their customary habit, realizing there are "excused absences" for things such as vacations, church mission trips, illness and business trips. "Dedicated attenders probably miss 15 to 20 weeks for reasonable reasons."
___A second problem, he suggested, could be sample bias. The people most likely to answer the phone and talk with pollsters also are the most likely to have attended church, he said. Likewise, those least likely to attend church may be least likely to be home to talk with pollsters.
___Yet another warning was sounded by Sean Everton, a graduate student in sociology at Stanford University. He presented an analysis of how church attendance varies dramatically according to seasonal patterns.
___His analysis shows Easter Sunday is the highest day of church attendance nationwide, with three other special days nearly tying for second place--Palm Sunday, Mother's Day and Thanksgiving week.
___Church attendance also gets a boost during Advent--the four weeks leading up to Christmas--but the impact is spread out more because the focus isn't on a single Sunday.
___Everton also documented a pattern of low-attendance Sundays generally following high-attendance Sundays.
___The basic cycle of church life in America shows attendance dropping off each spring, bottoming out in summer, then picking up again by September, leading to the peak days of Thanksgiving, Advent, Palm Sunday, Easter and Mother's Day.
___"The seasonal effects of church attendance are very clear," he reported. "Religious holidays and special-event Sundays attract more people."
___Not all congregations experience the same degree of variation in attendance, though. The relative strictness of a church's doctrine has a direct bearing on this, he said, noting that the more strict a congregation is the less variation is found in attendance.
___"There is far less variation among evangelical congregations" than among mainline Protestant congregations, he explained.
___This understanding of the seasonal cycle of church attendance sheds light on the polling problem, Everton said, by demonstrating that it makes a difference when the polling is done.
___That difference was highlighted by Bob Woodberry, a graduate student in sociology at the University of North Carolina. He compared the results of research based on time-use surveys from two periods of time.
___Other researchers have published data suggesting church attendance has declined precipitously, based on records individuals have been asked to keep of how they use their time.
___The problem with this research, he said, is that one of the most prominent case studies uses data from one year that falls over the Palm Sunday/Easter Sunday period, while data for the second year does not include those high-attendance Sundays in the sample.
___The bottom line, he said: "If you look at comparable data, you find no decline in church attendance on time-use surveys."
___The problem, though, is that the level of church attendance indicated on these time-use surveys turns out to be around 30 percent--still 10 points below the results obtained by the traditional Gallup question.
___That leaves unanswered the question of why.
___Meanwhile, it doesn't take a rocket scientist to see there's a problem with traditional polling results on church attendance, one participant in the discussion noted.
___"Where are the cars? If 40 percent of the population really was attending church, you'd have more traffic on Sunday mornings."

©2000 The Baptist Standard