Posted on 09/25/2016 10:28:46 AM PDT by Voluntaryist
For those of you who read and follow Peoples Pundit Daily, particularly our election projection models, you know I am a true believer in the average of polls and aggregate data. I do not believe in unskewing the polls nor the practice of weighting polling results for party identification. Party identification, which is fluid, is not the same as party registration and respondents should dictate to pollsters what the party breakdown of the electorate will be, not the other way around.
That being said, the 2016 presidential election is testing my faith. First, there are too many polls and not enough pollsters. Simply put, the same firms that completely blew the 2014 midterm elections and, subsequently, the 2016 primary elections, are doubling down on being statistically stupid. Most clearly refuse to improve on their models assumptions and methodologies and, as a result, the aggregate polling is over the place.
National polls, for those who arent polling junkies, are typically thought to be a leading indicator of state polls. In 2012, it was certainly the other way around, though they inevitably coalesced. Yet, despite clear movement in most battleground states toward Trump, the national averages remain skewed by suspicious outliers. Take the polling released in the past month, for instance.
Clintons dominance during the month of August, which was the combined result of her convention bounce and a media barrage covering Trumps self-inflicted wounds, was shattered with the release of a CNN Opinion Research Poll. Trump led by 2 points in the four-way race and 1 point in the fictional two-way race. The TIPP Poll, conducted for Investors Business Daily, which is one of the most accurate, highly-rated pollsters analyzed by the PPD Pollster Scorecard, had the 4-way race tied.
(Excerpt) Read more at peoplespunditdaily.com ...
I think this is what Ed Rollins was trying to say tactfully on the Lou Dobbs show. He said to ignore national polls and focus on states. Lou was adamant that we should pay attention to national polls but Ed stuck by his statement.
Wow!
I think national polls are useful for gauging the trend. But, I agree with Ed: the individual state polls are far more important at this time. However, it's expensive to conduct a statistically significant poll in 57 states.
At some point, a national poll will reflect the states -- i.e. it's currently impossible to have a 10 point gap in a properly-conducted national poll without causing a large swing in electoral votes. If the country continues to become more polarized, that might no longer be true... but we aren't there yet.
Good AND well reasoned article.
Polls don’t factor in election fraud, either.
Very interesting piece. And it makes total sense. Do not be discouraged by make believe national poll numbers.
They know the voter's state and even an approximation of congressional district, based on zip code. It would be interesting to see the cross-tabs for the individual states. But, I don't know if they publish that data, even to subscribers.
Some do, especially those that show Hillary 6 point up nationally. They are factoring fraud into their equation.
What makes you think the polls, national or otherwise, are creditable?
These “polls” are meant to massage the “narrative”. If the polls are done at all, they are skewed to influence the outcome.
I will posit another notion. A news outlet commissions a poll. The pollster knows what the buyer wants and the trend to be shown, he waits a few days and draws up a data base from his files, re-jiggers the numbers and turns the completed poll in. Collects his fee and moves on to the next news outlet. This operation can be run out of a shoe-box. Who verifies that the pollster did the work.
I call BS on the whole polling scam, except for the internal polling and that is reflected in where campaigns are making appearances and the money injected into state campaigns, and maybe ad buys. Although, the time around, that may not be too good of an indicator.
Nate Silver wrote yesterday what we’ve all been posting. It’s statistically impossible to be leading nationally by 6 and losing Ohio by 5 and Iowa by 7. Not to mention a dead heat in Maine which Republicans haven’t won since 1988.
2:00 am 11/9/16.
So why does 538 have Trump <43%?
Very funny cartoon by someone named Stilton Jarlsberg:)
I saw one where the beast was up by 2. I looked at the internals. They listed the states. North Carolina, New York, Maryland, Minnesota and two others that we both deep blue. Only NC was reasonably within the red realm. They oversampled women 64-36. They oversampled Democrats 36-24-30-10 The 10 are Libertarians and Greens.
Another, they did 0 Independent, 0 Lib/Green, 57-43 Dem/Rep. She was up 7.
They’re not even trying to hid their propaganda this cycle. It’s not news, but propaganda. Don’t believe the lying polls.
Why not? with such huge leads in CA, MD, WA, and MA????
Actually, many, if not most, polls are designed to facilitate fraud. The pollsters MUST report the election as being close so that fraud can be conducted without appearing obvious.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.