Like on Facebook   Follow on Twitter
2020 AVERAGES: NATIONAL VOTE | ELECTORAL COL. | HOUSE | SENATE | Battlegrounds: FL-OH-AZ-GA-IA-NC-MI-PA-WI-NH-NV-TX-MN | BEST IN 2016  
METHODOLOGY
FREEDOM WINDOW TURNOUT ASSUMPTIONS
 

Writing as PoliticalRef, using polling analysis this page predicted 48 of 50 states correctly in 2016. Compare our results to others. See the real time predictions. Final Prediction - Facebook | Twitter. Earlier predictions - Podcast | YouTube. A big reason for the success were the turnout assumptions.

 

OUR OBJECIVE: DETERMINE WHO WOULD WIN IF THE ELECTION WERE TODAY AND BY HOW MUCH

To best determine who would win an election if it were held today and by how much, we need to determine who will likely vote. Exit polls represent the best political poll available because pollsters interview people who are actually turning up to vote, not just telling you they will.

Those who vote early or by absentee are included in exit polls. When phone or online polls identify those who have voted, they are added to the certain to vote category and to the exit polls. Because exit polls are the best poll out there, I use the 2016 and 2018 exit polls as a standard to look to when estimating turnout for 2020.

2020 turnout almost certainly won't match 2016 exactly, but it will likely be very close. For this reason, in doing my projections, I adjust the partisan distribution to come close to 2016 if it substantially varies from that standard.

Also, I do not use polls in my projection that do not report partisan distribution or the ratio of Biden/Trump vote within each party. Without this data, one cannot reliably report a poll based on the current voting moods of the public. For this reason one ought to consider polls failing to report this information incomplete presentations not worthy of full credibility.

     
     
2016 Media Exit Poll - FNC   D+3 without leaners. Democrats 36%, Independents 31%, Republicans 33%. Ideology - 26% liberal, 39% moderate, 35% conservative.
     
     
     
Pew Research study of validated voters   D+3 with leaners. Voters were 51% D or D Lean/48% R or R Lean. Nonvoters who were eligible to vote favored Clinton, 37 Clinton to 30 for Trump, 9 for Johnson. Eligible nonvoters were 55% D or D lean, 41% R or R lean. This data demonstrates why polls of adults or registered voters, that do not filter out unlikely voters, artificially boost Dem support. Pollsters who fail to screen for likely voters necessarily benefit Democrats in most elections. It's really that simple.
     
     
     
2018 Media Exit Poll - FNC   D+3 with leaners. Democrats 46%, Independents 11%, Republicans 43%.
     
     
     
2018 Media Exit Poll - CNN   D+4 without leaners. Democrats 37%, Independents 30%, Republicans 33%
     
     
     
KEY FILTER FOR REGISTERED VOTER POLLS  
How Registered Voter polls can be adjusted without party distribution or Biden/Trump ratio within each party
 
This filter is far from perfect but is essential when pollsters insist in relying so heavily on a polling method that consistently favors one party over another to the detriment of the poll's accuracy. Using this filter will lead to a more accurate result more often than not.
 
There were 200 million registered voters in 2016 of which 138 million voted. 31% of registered voters, therefore, did not vote. We know that about 55% of eligible nonvoters would vote Democratic and 41% of eligible nonvoters would vote Republican. To account for this, a registered voter poll can be reduced by 31% in size where 17 percentage points (55% of 31) are deducted from the Democrat and 12.7 points (41% of 31) are deducted from the Republican. The new result should then be extrapolated out to the percentage of the poll composed by intending to vote Biden or Trump. This screen is only applied when party distribution and Biden/Trump ratios within each party group are not provided by the pollster. Following is an excerpt from Pew Research Center's, "An examination of the 2016 electorate, based on validated voters." I am using an analysis of 2016 because one candidate is the same and the Democratic candidate has many of the same problems motiviating voters as Clinton did.
   
  Among members of the panel who were categorized as nonvoters, 37% expressed a preference for Hillary Clinton, 30% for Donald Trump and 9% for Gary Johnson or Jill Stein; 14% preferred another candidate or declined to express a preference. Party affiliation among nonvoters skewed even more Democratic than did candidate preferences. Democrats and Democratic-leaning independents made up a 55% majority of nonvoters; about four-in-ten (41%) nonvoters were Republicans and Republican leaners. Voters were split almost evenly between Democrats and Democratic leaners (51%) and Republicans and Republican leaners (48%).
     
     
   
ONLINE ONLY FILTER   Evidence suggests that online only polls lean Democratic by as much as five points. See New York Times, Are State Polls Any Better than they Were in 2016. This article suggests online surveys will consistently lean Democratic within a range of 2.9 to 5.0%. I will compare online polls to phone only polls, or polls that have weighted its online portion with data collected through phone calls. If this is not possible, I will adjust by 2.9 to 5.0% based on the extent to which the poll seems to vary from the mean or otherwise contains unusual results.
     
     
     
RECALLED VOTE FILTER   Increasingly pollsters are weighting their polls to match the 2016 vote by asking respondents to recall who they voted for. This method benefits the losing party because more people remember voting for the winner and some will not admit to voting for the loser. If the pollster applies this weight and it matches 2016 very closely, the poll is likely oversamplng Democrats and an adjustment should be made.
     
     
     
NY Times 2016 Review: Turnout Wasn't the Driver of Clinton's Defeat   Black turnout matched the predictions, but was substantially lower than when Obama was on the ticket. The real driver of Clinton's defeat was the 1 in 4 whites who voted for Obama that defected to Trump. My take: Without an Obama or other black candidate at the top, it seems unlikely that black turnout surges to 2008 or 2012 levels. It remains to be seen if Biden's association with Obama translates to increased black turnout. It's unclear also how the protests/riots impact all of this.