Please register to participate in our discussions with 2 million other members - it's free and quick! Some forums can only be seen by registered members. After you create your account, you'll be able to customize options and access all our 15,000 new posts/day with fewer ads.
Though the times poll did their polling from registered voters broken down nationally by %. This was at least correct however the total respondents were 1600 people and 1358 were registered voters, so the 248 voters they have in there that didn't choose to respond to anything could throw this thing into the gutter. 15.5% of the respondents didn't supply their party affiliation so this could dramatically skew the results based upon the people polled.
There were also a lot of senate races polled.
Quinnipiac released several races. No polling information was given. Two days ago I tore the crappy polls they released that put Trump in front because of the overpolling of republicans by a bout 5 to 7 points. These polls probably came from that sampling which is why we are seeing these lean heavily republican when each is considered a toss-up.
WI-Sen: Marquette: Russ Feingold (D): 49, Ron Johnson (R-inc): 44 (Clinton 45-41)
Marquette did release their party affiliation breakdown.
The partisan makeup of the full registered-voter sample, including those who lean to a party, is 41 percent Republican, 49 percent Democratic and 9 percent independent. The long-term estimate over the previous 35 statewide Marquette polls, with 30,540 respondents, is 42 percent Republican and 48 percent Democratic, with 9 percent independent. The partisan makeup of this sample, excluding those who lean to a party, is 25 percent Republican, 30 percent Democratic and 41 percent independent, compared to the long-term estimate of 27 percent Republican, 31 percent Democratic and 39 percent independent.
Quote:
Originally Posted by emm74
I was going to make the same point as bellhead about the demographics - lots of discussion in the twitterverse about that. Quinnipiac underpolls people of color - sure, maybe their assessment that white voters will become a larger percentage of the electorate could be right, but most people don't think so given the changing demographics of the country at a whole.
as of now, we don't know much about the demographics of the NYT polls - and it's kind of odd they haven't released the crosstabs, so it's hard to say whether it looks demographically representative of what the electorate is likely to be. We do know that Trump is underperforming with people of color, sometimes in an astonishingly bad way, so these numbers are very relevant to how much credence to give any polls. Well, to the extent that polls now have all that much credence right now anyway, as I argued upthread. I think that many people are just now starting to dial into this election, they know who the nominees are and will be curious about the VP picks, but they haven't been thinking about it all that much and to the extent they are undecided, they haven't begun a whole lot of information gathering to change that yet.
The actual sample breakdown is largely irrelevant if the data are weighted to reflect current party affiliation percentages prior to tabulating/analyzing the results.
Further, while 538/Nate Silver does specialize in predictive modeling, most pollsters do not. Their universes are pretty much always based on past behavior, not possible future behavior.
As such, whites may seem to be over-sampled in many of these polls, however, their percentages are based on previous turnout combined with actual census numbers as those are the only consistently reliable numbers available.
Similarly, Republicans may seem to be over-sampled in some states, but if turn-out among Republicans is consistently higher than it is among Democrats or Independents, it would make sense to reflect that in the sample.
That would be the rationale anyway.
Of course, it is all just so much hocus-pocus as was clearly evident in 2012.
Somewhat related study on the number of white voters here:
The actual sample breakdown is largely irrelevant if the data are weighted to reflect current party affiliation percentages prior to tabulating/analyzing the results.
Further, while 538/Nate Silver does specialize in predictive modeling, most pollsters do not. Their universes are pretty much always based on past behavior, not possible future behavior.
As such, whites may seem to be over-sampled in many of these polls, however, their percentages are based on previous turnout combined with actual census numbers as those are the only consistently reliable numbers available.
Similarly, Republicans may seem to be over-sampled in some states, but if turn-out among Republicans is consistently higher than it is among Democrats or Independents, it would make sense to reflect that in the sample.
That would be the rationale anyway.
Of course, it is all just so much hocus-pocus as was clearly evident in 2012.
Somewhat related study on the number of white voters here:
Please register to post and access all features of our very popular forum. It is free and quick. Over $68,000 in prizes has already been given out to active posters on our forum. Additional giveaways are planned.
Detailed information about all U.S. cities, counties, and zip codes on our site: City-data.com.
Please register to participate in our discussions with 2 million other members - it's free and quick! Some forums can only be seen by registered members. After you create your account, you'll be able to customize options and access all our 15,000 new posts/day with fewer ads.
CBS/NYT - dead heat heading into convention...
