All errors should be reported to

Thursday, November 03, 2016

Applying the polls to the Electoral College

Byron York of the Washington Examiner is a rare pundit who actually gets out of the office and talks to people. But he shares in common with the rest of punditry a misunderstanding of numbers, and so he is allowing poll numbers to negate what he sees and hears.
In a column today, York said Trump has a big challenge to get to 270 Electoral College votes. he used Real Clear Politics averages as his source on polling.

York wrote:
On election eve in 2012, the RCP average of national polls had Barack Obama over Mitt Romney by 0.7 points. Obama won by 3.9 points. Yes, the poll leader won the race, but that kind of gap is enough to make any poll watcher cautious.
Pollsters talk about a margin of error and dismiss the huge miss accordingly.

So the national polls were off by 3.2 points in 2012.

State polls were off as well. From York:
Right now, Trump is leading Clinton by 0.6 points in the RCP average of Florida polls. (I'm going to use the 2-way numbers from now on.) On this date in 2012, Romney led Obama by 1.2 points. As the election neared, Romney's lead grew a little, to 1.5 points on election eve. Then Obama won by 0.9 points.
So a 3.2-point error nationally translated into a 2.4-point error in Florida in 2012.

York wrote:
Next comes another must-win state for Trump, Ohio, with 18 electoral votes. Trump leads Clinton by 3.3 points in the RCP average, a lead Trump has held for more than two weeks. 
On this date in 2012, Obama led Romney by 2.3 points in Ohio. He went on to win the state by 3.0 points.
So a 3.2-point error nationally translated into a 0.7-point error in Ohio in 2012.

York wrote:
The RCP average in North Carolina is tied — exactly tied, at 46.4 points for Clinton and Trump. In 2012, Romney led the state at this point in the race by 3.8 points in the RCP average. The race tightened a little in the last few days, and Romney won by 2.0 points.
So a 3.2-point error nationally translated into a 1.8-point error in North Carolina in 2012.

And so on.

The correlation between national and state polls is more likely to be happenstance because all these polls are wrong. But they are all we have.

Also, the political makeup of a state may change over time. From 1932 to 1996, West Virginia was stoically blue in presidential elections. In 2000, it flipped in the blink of an eye.

But applying 2012 to 2016 can illustrate that when national numbers go up, numbers in individual states tend to rise.

With a 0.7-point lead in 2012, Obama took Ohio. When that lead grew to 3.9, Obama took Florida as well.

The importance of the Electoral College is exaggerated because we had a close election in 2000 in which the plurality winner in the popular vote failed to gain election. Since then the margins in the popular vote were outside 1 percent, which meant the one with the most votes won. fleshed out the idea:
In November 3 polling averages:
5 battleground states are virtually tied.
Trump leads in 3 battleground states.
Clinton leads in 6 battleground states.
That’s a shift for Trump in recent days, although, since Clinton was ahead in so many states, the new Trumpmentum has turned some states into dead heats.
Trump's big number is that national one. If he beats Hillary by one point or more, he is the 45th president.

The rallies show Trump winning. So does the bad news for Clinton who once again is under a criminal investigation. The Electoral College map is a fun parlor game, but I have a feeling Trump and his supporters will render it meaningless on Tuesday.


Hate the media? Love The Donald? "Trump the Press" is the book for you. It skewers media experts with their own words. "Trump the Press" is available as a paperback, and on Kindle


  1. Don, you amaze take all this information, interpret it, and present it to us in a way that we can understand it. This is why your blog is a must-read every day.

  2. Why is it that those who call for diversity enforcement never see the need for diversity of opinion?

    Why does Ms "Stronger Together" Clinton constantly seek division rather than points of unity?

    1. Dave
      Tacitus on The Ancient Brits, who were very Tribal and had a woman as a war chief vs the Romans who were in a far away land and only had themselves to rely on:
      "They ( the Brits) fought individually,so were conquered collectively"

  3. Caddell thinks it could be a 40 state romp.

    I've been thinking that for a while.

    And do you notice how the pollsters are hedging their bets?

  4. Remember, Trump needs to win by a wide margin to overcome the cheating by the Dimocrats. Dead people vote Dimocrat and in every recount, Dimocrats always find more Dimocrat votes. Always! Sometimes it takes three recounts like it did when Christine Gregoire won in Washington. And illegals can supply the winning margin like in Minnesota when Al Franken defeated Norm Coleman.

    1. Illegals in Minnesota as in illegal felon voters.

    2. Illegals in Minnesota as in illegal felon voters.

  5. What I most fear is massive voter fraud. As Denny points out, it's already happened and will continue.

    This country- actually all of its individual states- really needs to carefully inspect its voting apparatus. Electronic voting should be disallowed, it's far too easy to rig; each ballot should be a paper one, and they should all be kept in case of irregularities; no early voting should be allowed (exception: Military and overseas citizens,) and mail ballots should be heavily restricted and only available to those who cannot vote in any other way. Voter rolls should be closely inspected and easily challenged by any citizen who suspects foul play! Valid ID should be required everywhere!

    Often I wonder if we'd have elected Obama if vote fraud hadn't happened. Remember, he took *over* 100% of the vote in several precincts!