Fmr President & Senator Polnut
polnut
Atlas Icon
Posts: 19,489
Political Matrix E: -2.71, S: -5.22
|
|
« on: July 11, 2016, 01:43:29 AM » |
|
Actually men are pretty oversampled in this poll (55.4% of this sample) to women (44.6% of the sample), because women are more likely to turn out to vote (53% of the electorate in 2012 were women). I only see very bad things for Trump/very good things for Clinton in this poll.
Wanted to bump this, because Pew buried it in most of the reports I read. I think 1960 was the last election where men outnumbered women, and most cycles it's not close.
Where does it say what fraction of the sample is male and what fraction is female?
http://www.people-press.org/2016/07/07/vote-preference-over-time/
1,655 is the registered voter subsample, who were the only ones asked horse race questions according to the questionnaire. 917 were men, and 738 were women.
But those are unweighted numbers. That’s just the raw number of men who picked up the phone vs. the raw number of women who picked up the phone. Once they’ve done the survey, they do demographic weighting to make the demographics match the demographics of US registered voters.
No, they're not. Numbers all match the weighted results from the press release. Topline is 51-42, Hipanics 66-24, youth 60-30.
http://www.people-press.org/2016/07/07/2-voter-general-election-preferences/
Huh? When they say that men are Trump 49% Clinton 43% from a 917 person sample and women are Clinton 59% Trump 35% from a 738 person sample, that doesn't mean that every single one of those men is weighted the same or every single one of those women is weighted the same, or that the men as a group are weighted the same as the women. It just means that 917 of the people they polled were men and 738 were women. There's still weighting done after the fact.
Those are the final weighted subsamples. Seriously, do the arithmetic, or maybe have a grownup do it for you.
|