PA- Bloomberg: Clinton +9

[BoAtlantis]

As David Plouffe said, PA is a fantasy for Republicans.

[BoAtlantis]

StatesPoll doesn't even bother opening the link.


"October 13 (Bloomberg) -- The Bloomberg Politics Pennsylvania Poll, conducted Oct. 7-11 by Selzer & Co. of Des Moines, Iowa, is based on interviews with 806 Pennsylvania residents who say they will definitely vote or have already voted in the 2016 general election. In order to look more closely at voters in the four counties that surround Philadelphia (Bucks, Montgomery, Chester, and Delaware counties), an oversample of 215 likely voters in those counties was conducted, leading to a total of 373 likely voters in these suburban Philadelphia counties.

Interviewers with Quantel Research contacted 1,233 Pennsylvania adults with listed landline and cell phone numbers supplied by TargetSmart to reflect the age 18 and over population of Pennsylvania. Responses from the full probability sample were weighted by age and race to reflect the general population based on recent census data. Interviews were administered in English.

Percentages based on the sample of 806 statewide Pennsylvania likely voters may have a maximum margin of error of plus or minus 3.5 percentage points, and those based on the sample of 373 likely voters in the four suburban Philadelphia counties may have a maximum margin of error of plus or minus 5.1 percentage points. This means that if this survey were repeated using the same questions and the same methodology, 19 times out of 20, the findings would not vary from the percentages shown here by more than plus or minus 3.5 percentage points for the statewide sample and plus or minus 5.1 percentage points for the suburban counties. Results based on smaller samples of respondents—such as by gender or age—have a larger margin of error."

[BoAtlantis]

This is also from Selzer who has a heavy Trump tilt in her likely voter screens all cycle. It's very likely Clinton is ahead by double digits

You have to stop unskewing polls, Selzer is probably the best pollster in the country. They know what are they doing.

Having said that, Trump is finished.

No, no, that's not "unskewing."  It's adjusting for so-called "house effects."

Pollsters often have house effects -- sampling practices and other factors that mean that they generate results that are more Democratic or Republican than the average poll.  For instance, if we only have one Indiana poll, and it's from a pollster that routinely finds unusually Democratic results in other states, it's very reasonable to assume that the poll in Indiana is also too Democratic.  This is perfectly reasonable, because pollsters with partisan house effects do exist, and it would be a bad idea to ignore that if the polls in a given state come from pollsters with the similar house effects.

"Unskewing" is re-weighting a poll based on either non-static political factors (like party ID), which is a terrible idea.  Sometimes "unskewing" is also applied to the practice of re-weighting based on shoddy prior data like race composition from exit polls.  In the former case, you're re-weighting a poll based on something you should reasonably expect to change; in the latter case, you're re-weighting a poll based on information that probably was inaccurate in the first place.

Neither of those things are reasonable.  Done right, re-weighting for house effects is the right thing to do.

Absolutely correct. Party unskewing is what Dick Morris did and his results ended up being bigly terrible. Nate Silver adjusts for House-effect all the time and he already re-weighted this poll from +9 to +10.