Pay or Pray?

Via Marginal Revolution, we are alerted to a paper by Jonathan Gruber (unfortunately available only for a $5 charge), which finds:

Using data over three decades from the General Social Survey, as well as confirming the impact of such subsidies on religious giving using the Consumer Expenditure Survey, [there is] strong evidence that religious giving and religious attendance are substitutes: larger subsidies to charitable giving lead to more religious giving, but less religious attendance, with an implied elasticity of attendance with respect to religious giving of -0.92.

It would be interesting to know whether Gruber's data could be fine-tuned. In particular, I wonder whether the substitution effect is stronger or weaker in churches adhering to sola fide. I've emailed Gruber to ask.