I don't put much faith in polls generally, but I put even less faith in polls where people are asked how they feel about porn. I don't think you can come to any reasonable conclusion from data of such low quality as is typical of polling these days.

Even in the absolute best circumstances where enough people are polled to be representative, and those people aren't asked any leading/misleading questions, and the identity of all those people are known, pre-selected without bias, and verified (preventing the same person/group of people voting 50 times or brigading some anonymous internet survey), and all of those people are 100% confident that their answers are private and won't be able to be used against them, you're still left with the fact that people lie. All the time. Especially about anything to do with sex. They also have terrible memories and their beliefs about themselves and their views often don't hold up when their actual behavior is observed. Self-reported data is pretty weak even when sex/shame/morality/fear of punishment don't come into play.

Without really digging into the specifics to try to work out how seriously you can take a given survey's results at all, it's best to just not to treat them seriously.