2016 Election Model Forecast

Unlike corporate mainstream polls, the 2016 Election Model provides two forecasts:  the Recorded Vote and the True Vote. Pollsters are usually quite accurate in their projections of the Recorded Vote. But they avoid the fraud factor. The fraudulent Recorded Vote is never the same as the True Vote.

The Election Model projects that Clinton will win the fraudulent Recorded Vote by 45.6-43.6% (294-244 EV) and that Trump will win the True Vote by 49.2-44.5% (367-171 expected EV).

The  Election Model  is based on the effects of changes in party affiliation (Dem, Rep, Ind) from 2012 to 2016. The polls are adjusted to Gallup party-affiliation survey weights. Gallup is the only poll dedicated to tracking national  party affiliation.

Each published poll has a different Party-ID split .Theoretically, they should all have the identical split. After all, they are NATIONAL polls – and there is only ONE theoretical NATIONAL Party-ID split at any given point in time.

After adjusting the polls for the Gallup voter affiliation split (40I-32D-28R),  undecided voters are allocated (UVA) to derive the final adjusted TRUE poll share. Typically the challenger (in this case Trump) gets approximately 75% of the undecided vote.

Undecided Voter Allocation Sensitivity Analysis

UVA  Trump Clinton  ExpEV  Pop.vote Win Prob.
50%….47.1….45.6…….310….. 73%
60%….47.6….45.1…….332….. 86%
75%….49.2….44.5…….367….. 98%

The estimated popular vote win probability and corresponding Electoral Vote are calculated for each poll. The 2016 party-ID for each state is calculated by applying the  proportional  change  from the 2012 party-ID to  the Gallup 2016 survey. The state votes  are calculated by applying the candidate national poll shares to the state party-ID.

The electoral vote is  calculated two ways: 1)  the total EV  (snapshot) in which the winner of the state wins all  of the state electoral votes and 2) the expected EV (state win probability times the state electoral vote).

The Sensitivity Analysis tables show the effect of incremental vote shares on the total vote.

The 2008 and 2012 election models exactly forecast the electoral votes (365 and 332 for Obama). But the True Votes were quite different. The 2008 model forecast that Obama would win 420 votes with a 58% share, exactly matching the state unadjusted exit poll aggregate. He led the unadjusted National Exit Poll by 61-37%.

The 2012 model forecast Obama: 51.5% recorded and 55% True vote (380 EV}. But the pollsters failed to exit poll in 19 states, so I could not confirm the results. https://richardcharnin.wordpress.com/2014/09/14/summary-2004-2012-election-forecast-1968-2012-true-vote-model/

Gallup Adjusted