I'm gratified to see that the lizardman coefficient is consistent. In the chart, 2% of "None/Atheist/Agnostic" apparently attends religious services every week.
My brother and I used to take our grandmother to church for Easter mass, and a friend of mine goes once a month with his wife's family, despite all three of us being atheists. I really don't think that 2% is guaranteed to be polling error, as Lizardman's Constant would suggest.