In assuming all respondents have a religion, the framing of the question produces acquiescence bias that inflates data — by as much as 11 points, according to a number of surveys — in favour of religious affiliation.
In assuming all respondents have a religion, the framing of the question produces acquiescence bias that inflates data — by as much as 11 points, according to a number of surveys — in favour of religious affiliation.
On the one hand, changing the way the question is asked might make it more accurate as one snapshot in time. On the other, it makes comparisons between years harder, and the change could mask other religious changes currently happening in the community.
I’m also not sure if asking if they’re religious first and only asking for which if they say yes, won’t have no bias in a different way. I used to know people who would say: “The Bible is truth and not religion”, and those people would be counted as not being religious if the changes were to occur. Then again, those people are rare, and might feel compelled to answer that they are religious anyway, even if they don’t think of themselves as such.