August 25, 2020

Here's why I'm wrong.

Here's my latest  filtered projections, that I've built entirely in Excel (including the map and bubble visual).  I've filtered the RealClearPolitics polls for each battleground state, to exclude polls of registered voters or all adults, polls with a Margin of Error of >4.5%, and a sample size of under 500. 

According to the filters I've applied Biden should win. He's already captured back Michigan, Pennsylvania and North Carolina.  If you concede his likely states of New Mexico, Colorado and Virginia, he's already at 273.  That's not to mention Nevada and  Minnesota which are also fairly firm in Democrat historical support.  That would put Biden at 289 and in that scenario he probably also wins New Hampshire and ends up at 293.  It won't matter if Trump wins Florida, Arizona, Ohio and Iowa.  He still tops out at 245.

Here's why I'm wrong.

North Carolina.  My weighted average results show Biden ahead 50.7% to 49.3% on the strength of 2 polls.  One of the two polls is the CBS/YouGov polls which I expressed concerns with back in 2016.  The Emerson poll actually had Trump ahead very slightly.  MY own opinion is that there is only one viable poll and that is not enough to call the state.

Pennsylvania.  The same two pollsters as in North Carolina have Biden ahead 53.6% to 46.6%.  Emerson, the more reputable pollster of the two in my estimation, actually have Biden +9%.  But it's only one poll and that's not enough to call the state.  I believe the state is actually too close to call, but that remains to be seen.  My suspicion is that the Trump vote is underrepresented. 

Michigan.  Two polls have Biden ahead +8%.  Once again one of the polls is the YouGov poll and the other has Biden +11%.  This state more than any other so far indicates that even if the polling is done reasonably (and without seeing the crosstabs even that may be questionable), there's something amiss.  Trump down 8% on average or 11% in one poll, just doesn't seem reasonable.  In 2012 president Obama won re-election by 9.5%.  Hillary Clinton, projected to win by 5%, lost by 0.23%   To go from Democrat +9.5 to Republican +0.2 back to Democrat +8 in the course of three elections does not seem feasible.  I need to see more polls.  

Wisconsin.  We're looking at just two polls, YouGov and Marquette, in Wisconsin.  YouGov has Biden +5% and Marquette +3%.  The weighted average is Biden +4%.  The problem here is the same as Pennsylvania and Michigan. Too little data, suspect polls and a potential silent Trump vote.

These are just the states that have moved back to the Democrat from Trump.  One of the states that is undecided there is only the Emerson poll that has Biden +4%.  The state has been trending Democrat demographically, but to go from a historically R +5% to +10% to even just +3% for Trump to suddenly Biden +4% indicates a polling problem, not a true trend.

I don't believe the polls.  Clinton was +9% in August 2016 if the polls then were to be believed. In October 2016, CNN announced she was ahead by 12 points.  She was never ahead by those numbers.  But even if she was, right now RCP has Biden +7.6% nationally, slightly over-performing Hillary Clinton at the same point in the race, where the RCP average had her +6% nationally. On election day she was ahead by 3.2% in the RCP average.  She won the popular vote by 2.1% despite losing the electoral college.

If we extrapolate Biden's out-performance of Hillary and assume the race 'mysteriously' tightens closer to election day by the same historical amount, Biden should end up ahead of Trump by 2.66%.  Remember libertarian Gary Johnson got 3.3% of the electoral college and a lot of that vote, some begrudgingly, some happily will gravitate towards president Trump because there is no Gary Johnson level libertarian in this race.  Do you know who Joanne Jorgensen is? Probably not.  To be fair, Green Party voters might gravitate to Biden the same way.  But they only got 33%  of the support of Gary Johnson last time around.  So it's too close to call with all of those variables out there.  And on top of all of that, these poll numbers now are just too sketchy to believe.

No comments:

Post a Comment

Disagreement is always welcome. Please remain civil. Vulgar or disrespectful comments towards anyone will be removed.

Related Posts Plugin for WordPress, Blogger...

Share This