Quote:
Originally Posted by ryberg
When have we ever been so exacting in terms of the figures and semantics in such cases? Indeed we frequently don't even get reference in any such articles as to how many were polled, just as we frequently ignore the fact that some poll results are referring to subgroups polled rather than the entire body of respondents, just as we don't distinguish between registered voters and so-called "likely voters" or whatever else where the focus is on other groups in society. And many polls we discuss don't have as many respondents overall as that 892 figure for that subgroup in this poll.
|
Which is part of the reason the polls are of such little value, especially this far out from the election. Nevertheless, an issue was raised and response given as to the composition of the poll, I just felt it would be a service to describe that poll composition. The fact that other polls may have had fewer than 892 respondents is not the issue, although I would be hard pressed to call such polls "national" in scope with so few respondents; it is the way the results were presented which originally piqued my interest.
I do remember having a similar recitation of polling methodology not that long ago, so the issue has arisen, at least once, in the recent past. As to your point that you do not distinguish between registered, likely, or "whatever else" voters in the polls you quote, that is one of the reasons the polls are often so widely dissimilar from actual voting results. It seems as if some people get real heartburn just after an election where "their" candidate was shown with a big lead in the polls heading into the election, only to have said candidate lose in the end; paying closer attention to the polling methodology may eliminate the need for Brioschi.