In a more recent study, Hadaway estimated that if the number of Americans who told Gallup pollsters that they attended church in the last week were accurate, about 118 million Americans would be at houses of worship each week. By calculating the number of congregations (including non-Christian congregations) and their average attendance, Hadaway estimated that in reality about 21 percent of Americans attended religious services weekly--exactly half the number who told pollsters they did.

Finally, in a brand new paper, Philip Brenner at the University of Michigan's Institute for Social Research compared self-reported attendance at religious services with "time-use" interviews in the United States, Canada, the Netherlands, Germany, France, Norway, Finland, Slovenia, Italy, Spain, Austria, Ireland, and Great Britain. Brenner looked at nearly 500 studies over four decades, involving nearly a million respondents.

Brenner found that the United States and Canada were outliers--not in religious attendance, but in overreporting religious attendance. Americans attended services about as often as Italians and Slovenians and slightly more than Brits and Germans. The significant difference between the two North American countries and other industrialized nations was the enormous gap between poll responses and time-use studies in those two countries.

Why do Americans and Canadians feel the need to overreport their religious attendance? You could say that religiosity for Americans is tied to their identity in a way that it is not for the Germans, the French, and the British. But that only restates the mystery. Why is religiosity tied to American identity?