The problem with this chart is it doesnt state how many muslims and hindus are in the population for the polling that was done. I doubt many use OKCupid, so if they only have had a few, the error would be huge and essentially prove nothing. Yet another example of using statistics wrong.