Something Odd About Polls….
Former Democratic worker Dana Houle noted some odd, systematic things about polling this morning…
Whites voters vs results/overall
2008 exits=McCain+12/Obama+7
2012 exits=Romney+20/Obama+4
Marist nat’l poll=Trump+12/Clinton+1Riiiiiight…
— Dana Houle (@DanaHoule) November 6, 2016
Whites voters vs results/overall
2008 exits=McCain+12/Obama+7
2012 exits=Romney+20/Obama+4
Marist nat’l poll=Trump+12/Clinton+1Riiiiiight…
— Dana Houle (@DanaHoule) November 6, 2016
Last 18 months been literally hundreds if not thousands of polls showing Hillary>Obama w whites but worse overall https://t.co/kgA0qWGrXg
— Dana Houle (@DanaHoule) November 6, 2016
This is something folks note in passing, but really haven’t thought about. And it really should be.
Houle thinks there are two factors, one of which we know about:
White share of electorate, per exits:
1992=87%
1996=83%
2000=81%
2004=77%
2008=74%
2012=72%
White share’s been dropping independent of Obama— Dana Houle (@DanaHoule) November 6, 2016
And folks have talked about this a lot. Posting a whiter electorate bucks voting trends since 1992. And it also bucks demographic trends as the non-white population has been growing faster than the white population for a while now.
But there’s another factor:
@DanaHoule @nickgourevitch and the GOP leaning large non white group is???
— JMizzle (@JMizzle36) November 6, 2016
This gets to a problem w RDD media polls: Marist, for example, has white/black/Latino=91%, 9% “other.” Um, no. https://t.co/pJWErc6h6G
— Dana Houle (@DanaHoule) November 6, 2016
Whites are only racial category that’s not reliably Dem. Yet many media polls have 5-10% “other,” & it’s almost always Repub +10 or more
— Dana Houle (@DanaHoule) November 6, 2016
THIS DOES NOT PASS THE SNIFF TEST. Pollsters often do not break out Asians as a category, so they are lumped into the Other category. And we KNOW that Asian Americans as a group are heavily Democratic and Clinton leaning.
There’s something hinky going on here…and few people are thinking about it.