Depends. You can't be objective over attendance unless there's an exact threshold, but there isn't. However I believe majors are weighted slightly more over regular locals.
Again though, you're being really fickle with the details, you're leaving them out when it suits you and including them when it suits you. You can't generalise what I am saying by saying 'ignoring common sense'. Common sense isn't valid here in the usual way because you're assuming YOUR method is common sense and then using the fact you believe you're right as evidence. You can't do that lol. It's not based strictly on the outcome, other factors could be involved, which is why people make the PR's, not statistics, but here I do not believe that how many 'games' you won is relevant if you lost the match. Considerable factors could be that they're not altogether there because of some sort of emotional situation. The thing is though, it's based off of consistent tournament results, so everybody has that leeway of not doing good once in a while, no matter what the reason.
The method we are talking about is whether to consider how many games you won in a set to be relevant, and maybe in some way it is, but not to PR. Rankings will be more accurate if you exclude those factors because there are even more reasons why games can be lost in a set, look at what Ricky said for example. How those players do (win/loss) against other players will provide them with an accurate placement regardless.
Though I also think that including games rather than sets adds for this much higher level of bias, especially when there could be reasons to dropping games to you, even though they are never losing the set. And not something as simple as "well it's because they're close in skill level" because if they are, other results vs other players will show that.