2016 Presidential Election Model Prediction

Based on my statistical analysis I published a primary prediction model which stated:

“Donald Trump will be just under the 1,237 delegates needed to win after the Washington caucus on 5/24/16”

You can read the entire analysis of how I came to my conclusion here:

http://www.freerepublic.com/focus/f-chat/3410261/posts

Indeed, months later, a CNN story states:

“Donald Trump is less than 10 delegates shy of clinching the Republican presidential nomination following a win Tuesday in the Washington state primary.” That article is here:

http://www.cnn.com/2016/05/24/politics/washington-republican-primary-results-donald-trump/

Trump then, as I predicted, coasted to victory by picking up many more delegates in the remaining primary elections. There was no "brokered convention".

My General Election model will use a different method than my Primary Model as we don’t have any states that have already completed their votes in the General Election in order to project results of future states. A model that takes into account past General Election results and has accurately predicted future results is called for. There are several out there that use different methods.

The best is from a professor of Political Science at Stony Brook University, Helmut Norpoth. He has published “The Primary Model” since 1996. “It is a statistical model that relies on presidential primaries and an election cycle as predictors of the vote in the general election.”

http://primarymodel.com/

It has correctly picked the winner in each election since it was first published in 1996. Furthermore, he has looked at how his model would have predicted Presidential elections going back to 1912. With only one exception, it would have correctly picked the outcome of every election since 1912. The single exception was when the model predicted Richard Nixon to beat John Kennedy in 1960. Many feel that the election was "rigged" and stolen from Nixon, so even that "incorrect" result of the model is somewhat in question.

His model for 2016 states:

“In a match-up between the Republican primary winner and each of the Democratic contenders, Donald Trump is predicted to defeat Hillary Clinton by 52.5% to 47.5% of the two-party vote.”

Some may find it hard to believe that Trump could get 52.5% of the vote. A republican hasn't got that high of a percentage of the vote since George H.W. Bush in 1988. Here's how Trump beats Clinton 52.5% - 47.5%.

First, it's not prediction that Trump will get 52.5% of all Presidential votes cast. It's a prediction that of votes cast for Trump or Clinton, "The Donald" will get 52.5% of those votes. Third party candidates could garner 10% of the popular vote in this election. If Trump were to get 52.5% of the other 90% of the votes cast, that would be 47.25% of the overall vote, which is quite possible.

Secondly, less than 60% of eligible Americans vote in an average Presidential election. That leaves a HUGE untapped reservoir of potential voters ripe for the picking. Is Hillary Clinton going to inspire millions of people to vote who have given up on politics and wouldn't otherwise vote? That seems unlikely. Democratic turnout was down dramatically during the hotly contested primary race between herself and Bernie Sanders. If she can't even get her normal democratic partisans to turn out, how will she inspire new voters to come out and vote? On the other hand, republican turnout was up by record numbers. Trump got the most primary votes ever by a GOP candidate, and it wasn't even close. Being an outsider, not closely aligned previously with the republican party, Trump stands a much better chance of turning out new voters who believe they finally have a non-politican to vote for. He is "their voice".

Trump also has a better chance of taking more blue-collar voters away from the democrats than any GOP candidate since Ronald Reagan due to issues involving trade and immigration.

He will get some of Bernie Sanders supporters who feel "berned" by the democrats.

He also stands to do better among blacks and Hispanics than the media has blindly reported.

Trump is a well-known celebrity, and Americans love their celebrities. Arnold Schwarzenegger ran as a conservative republican in California and won the governor's chair. Another celebrity, Jesse Ventura, was elected governor in Minnesota. It's not much of a stretch to see a celebrity having nation-wide appeal as well. Democrats have been trying to "dumb down" our education system for decades. They accuse Trump supporters of being "low information" voters. How ironical it would be if "dumb-downed, low-information" voters put a republican into the White House because of his celebrity status.

Professor Norpoth’s model is predictive of the national popular vote only, not state-by-state popular, or more importantly, Electoral Votes. My statistical analysis takes it from there.

I looked at how a 52.5% national popular vote would break down state by state, and how that would equate in Electoral Votes. Tweaking results from recent Presidential elections I removed and added certain statistical biases to varying degrees in all the states. For example, adjusting for how Presidential and Vice Presidential candidates do in their home states had to be accounted for. Additionally, Mitt Romney’s (and others) effect on LDS voters has also been factored in.

I did all of this and completed my model in July of this year, once the candidates were locked in. The polls have gone up and down, and there have been many controversies, surprises, twists and turns since then. There may be more still in the next few days before the election but none of that effects the model, so it's the same today as it was in July. Some days since then it has looked like it was selling Trump short, while other days it looked like it was selling Clinton short. So, perhaps when the dust settles it will show to be accurate after all, as my Primary Model prediction turned out to be.

With that said, here is my state-by-state statistical breakdown. Remember, it takes 270 Electoral Votes to win the election.

First, let’s look at deeply red and blue states, respectively, that should be easy wins for the candidates.

For Trump:

Alabama - Trump wins 63.84% - 36.16% (9 EVs)

Alaska - Trump wins 58.26% - 41.74% (3 EVs)

Arizona - Trump wins 53.90% - 46.10% (11 EVs)

Arkansas - Trump wins 65.51% - 34.49% (6 EVs)

Georgia - Trump wins 57.50% - 42.50% (16 EVs)

Idaho - Trump wins 57.81% - 42.19% (4 EVs)

Indiana - Trump wins 66.17% - 33.83% (11 EVs)

Kansas - Trump wins 65.69% - 34.31% (6 EVs)

Kentucky - Trump wins 66.73% - 33.27% (8 EVs)

Louisiana - Trump wins 60.19% - 39.81% (8 EVs)

Mississippi - Trump wins 57.55% - 42.45% (6 EVs)

Missouri - Trump wins 61.15% - 38.85% (10 EVs)

Montana - Trump wins 62.42% - 37.58% (3 EVs)

Nebraska - Trump wins 65.97% - 34.03% (5 EVs)

North Dakota - Trump wins 66.08% - 33.92% (3 EVs)

Oklahoma - Trump wins 70.93% - 29.07% (7 EVs)

South Carolina - Trump wins 58.41% - 41.59% (9 EVs)

South Dakota - Trump wins 65.52% - 34.48% (3 EVs)

Tennessee - Trump wins 65.25% - 34.75% (11 EVs)

Texas - Trump wins 61.80% - 38.20% (38 EVs)

West Virginia - Trump wins 71.97% - 28.03% (5 EVs)

Wyoming - Trump wins 70.55% - 29.45% (3 EVs)

That gives Trump 185 Electoral Votes to start with.

For Clinton:

California - Clinton wins 60.30% - 39.70% (55 EVs)

Connecticut - Clinton wins 53.52% - 46.48% (7 EVs)

Delaware - Clinton wins 53.68% - 46.32% (3 EVs)

D. C. - Clinton wins 88.65% - 11.35% (3 EVs)

Hawaii - Clinton wins 70.10% - 29.90% (4 EVs)

Maryland - Clinton wins 61.58% - 38.42% (10 EVs)

Massachusetts - Clinton wins 59.11% - 40.89% (11 EVs)

Michigan - Clinton wins 54.12% - 45.88% (16 EVs)

New Jersey - Clinton wins 57.19% - 42.81% (14 EVs)

New Mexico - Clinton wins 54.38% - 45.62% (5 EVs)

New York - Clinton wins 62.42% - 37.58% (29 EVs)

Oregon - Clinton wins 54.61% - 45.39% (7 EVs)

Rhode Island - Clinton wins 61.51% - 38.49% (4 EVs)

Vermont - Clinton wins 65.29% - 34.71% (3 EVs)

Virginia - Clinton wins 52.44% - 47.56% (13 EVs)

Washington - Clinton wins 56.39% - 43.61% (12 EVs)

Wisconsin - Clinton wins 52.82% - 47.18% (10 EVs)

That gives Clinton 206 Electoral Votes to start with. I know Trump has been competitive in New Mexico and perhaps thought Michigan might be in play, but the model shows that not to be the case. It also shows that while he may have areas of strong support in New York and New Jersey it is not likely to be nearly enough to take those heavy-liberal states away from the democrats.

The model shows Clinton will win the popular vote overall in Maine 55.32% - 44.68%; however, Trump to win one of the Congressional districts. Due to their unorthodox system (only Maine and Nebrasksa are not "winner take all), it would give Trump 1 Electoral Vote, while Clinton would be awarded the other 3 EVs, making the score: Clinton 209 EVs, Trump 186 EVs.

Perhaps the biggest surprise is Utah, with the model predicting Clinton wins 57.76% - 42.24% for 6 EVs. It's hard to imagine a hard-core leftist winning perhaps the most conservative state in the country; however, the conservative vote is going to be splintered between Trump and several 3rd-party candidates due to Mitt Romney, Glenn Beck and others trashing Trump, and some conservative voters may not even make a Presidential selection. Meanwhile, all the democrats will likely be united behind their candidate, perhaps explaining this unlikely win, making it Clinton 215 EVs, Trump 186 EVs.

Now, the good news for Trump. The model shows a Trump sweep of the following battleground states:

Florida - Trump wins 55.81% - 44.19% (29 EVs)

Iowa - Trump wins 51.26% - 48.74% (6 EVs)

New Hampshire - Trump wins 51.49% - 48.51% (4 EVs)

North Carolina - Trump wins 54.49% - 45.51% (15 EVs)

Ohio - Trump wins 51.74% - 48.26% (18 EVs)

Pennsylvania - Trump wins 52.70% - 47.30% (20 EVs)

That gives Trump a total of 278 EVs, 8 more than needed to clinch the victory. It shows much more comfortable wins for Trump in Florida and North Carolina than many anticipate. The closest of these battleground states are Iowa and New Hampshire, but still show him winning by a fairly comfortable 2 1/2 - 3% points over Clinton. Even if the model is off a little in Trump's favor and he loses 1 of those 2 states, he would still win the election, although it would make it a razor-thin margin. He can't, however, afford to lose any of those other battleground states.

Finally, for the 4 remaining states, the model predicts the following:

Colorado - Clinton wins 50.39% - 49.61% (9 EVs)

Illinois - Clinton wins 50.22% - 49.78% (20 EVs)

Minnesota - Clinton wins 50.60% - 49.40% (10 EVs)

Nevada - Clinton wins 50.95% - 49.05% (6 EVs)

Even if Clinton does indeed sweep those last 4 states she would finish with 260 EVs, 10 short of victory. Notice, however, how close these states are expected to be. If the model is off even a little in Clinton's favor, Trump could easily pick up 1 or more of those states. Colorado and Nevada would not be a huge surprise. Minnesota would be, while Illinois would be the shocker of the night. Trump could see his Electoral Vote tally go over 300 if he pulls off winning Illinois and 1 other of those states, or if he loses Illinois but gets the other three.

2016 Presidential Election Model Prediction: Trump 278 EVs Clinton 260 EVs.

I wonder if Al L the Republicans who stated they would vote for Hillary Clinton are still going to stand up and vote for her against Donald Trump

Excellent analysis - I think you’ll see a Colorado, Michigan, Wisconsin, and a few others go Trump as Hillary voters flip to Stein or Trump or Johnson. An informal poll in NJ said that 20% of D votes for Hillary are considering NOT voting for her, but it’s probably less than half of that. Even then, that’s quite substantial.

This week is crunch time - who knows what will break Monday through Friday.

Notice, however, how close these states are expected to be. If the model is off even a little in Clinton's favor, Trump could easily pick up 1 or more of those states. Colorado and Nevada would not be a huge surprise.

In the short time remaining, will your model pick up the changes that are occurring as the result of the re-opened investigation?

Since I don't know what the inputs to your model are, I'm not sure how it would change.

In the short time remaining, will your model pick up the changes that are occurring as the result of the re-opened investigation?

No, the model was done in July. It's not supposed to reflective of what the state of the race was then or today, but how it will turn out on November 8th.

Interesting.

But I just don't buy that Hillary gets Utah.

If Trump carries the national popular vote by greater than 3.5% he will win CO, IA, NH, PA, OH, VA, FL, and NC. If he wins the popular vote by 5% I would guess that WI and MI would be in play as well.

I wonder if it is all talk in public but when they vote, will vote for Trump.

Bookmarking this one; many thanks!

But I just don't buy that Hillary gets Utah.

I have many friends and relatives in Utah. If they cost Trump the election, I will never forgive them for the destruction that will befall all of us.

Bkmk

I doubt Utah will go for Hillary and I bet a few of Clinton states are in play.