There are some polls that are guaranteed to produce inaccurate results no matter how well-conducted they are. For example, there's an established phenomenon whereby if you ask respondents how many sexual partners they've had, men will on average overstate the true number, and women will understate it. People will also claim to have given far more money to charity in the last month or year than they actually have.
There's a similar problem with today's YouGov poll that seeks to test public attitudes to David Cameron's alleged debauchery during his highly privileged university education. 62% of respondents agreed with the proposition that the incident was "years ago" and thus "couldn't matter less". I guarantee you there are people among that 62% who think less of Cameron as a result of the pig story, but who nevertheless gave the answer they felt reflected best on themselves.
You may have seen the famous Yes Minister scene that demonstrates how it's possible for pollsters to get almost any answer they want. The question sequence that would have produced damning results for Cameron goes something like this...
1) Do you agree that sexual acts involving animals should be regarded as criminal offences?
2) At present, sexual acts involving live animals are criminalised, but those involving dead animals are not. Do you believe this is a loophole that should be closed?
3) Do you believe that people should be held responsible for their actions as adults, even if they took place twenty or thirty years ago?
4) If it is true that the current Prime Minister performed a sexual act on a dead pig as a student, do you believe this is a matter of legitimate public interest?