[tptacek]

I'm sorry but I'm not sure I understand what you're saying. Which specific poll are you referring to and what specific problem do you have with its methodology? What poll more accurately describes public opinion, or is public opinion on this subject either "unknowable" or in some other way clear?

[coldtea]

How are these polls unlike all the other polls that e.g. predicted Trump will lose by a large margin?

Most polls are quite vague approximations, when they aren't just lying with statistics. And you can easily frame them in all kinds of ways by how one asks the questions and how they present an issue.

[tptacek]

Those polls were broadly accurate, as has been heavily reported. In particular, the outcome tracked the national polls, and upsets were targeted and involved very narrow margins.

But, of course, none of your comment answers the questions I posed. I'm less interested in litigating abstractions and more in the specifics. Care to take a whack at addressing them?

[coldtea]

>Those polls were broadly accurate, as has been heavily reported.

The polls about The Donald losing were also heavily reported: http://www.politicususa.com/2016/10/26/polls-shatter-myth-pr...

In any case, unlike you, I'm more interested in abstractions based on large historical body of evidence (for not trusting pollsters in general) than in specifics (evaluating a particular poll, especially in hot political and partisan issues, which presupposes that I can trust what those that created it say, that is not only trusting the methodology as they describe it, but also that they followed it -- and we obviously have no way of knowing the latter).

How did you verify the poll? Because this is not some scientific paper where one can retract the steps and redo the numbers. Just trusting the methodology reported (a meagre page with a 10-mile high view of the procedure and no specifics in the case of Pew Research Center, for example?) and having faith that no poll and/or research center would ever misreport or falsify?

Maybe my mistrust is wrong, but I don't see the trust as being any better, especially with so many off the mark polls that been shown in practice to be so (the actually verifiable and falsifiable ones, like the ones on voting preferences).

[tptacek]

I'm really sorry, I'm not sure I follow what you're saying here, but I am confident that you're pursuing a different argument than the one I was, and I'm just not interested in it.

[nl]

Which polls were so far in error? The presidential vote polls were ~2% out and I think most in this thread on this topic will concede a 2% error either way.