In a comment @ TBogg’s on public displays of religiosity, this gem was revealed:
Knowing that many people will lie and claim that they are regular church attendees, certain researchers — rather than asking people if they attended — instead looked to studies done by the NIH (or CDC or someone) about allergies. The people in these studies had to record where they were every hour of every day of the week. When looking at *these* data, church attendance was determined to be at something like 15% or less (I don’t quite remember).
I’d love to see that study.
More on over-reporting here, here, and here. Any way you slice it, it’s no majority.