I know enough about statistics to know that an all-online poll has a significant sampling error, which is one reason why I read the Methodology section. Quote from Methodology notes: "This online sample was not a probability sample." In other words, despite doing their best to weight the sample, they can't be sure that it was representative. And I quite obviously did read past Table 2-A, otherwise I would not have seen the "very religious" category in Table 2-B that I pointed out earlier.