although no method is perfect.

One alternative is to sample heavily in areas where there are many Jews, and then attempt to adjust for that type of sampling.  I've not seen this done for religious denominations, but it was done quite effectively by a guy who was trying to get a random sample of GLBT people