Election model predicts Clinton over Trump

is the title of this piece in The Hill which looks at the model from Moody Analytics.   I came across it as a result of a tweet:

Moody’s Analytics latest fearless forecast predicts Hillary will crush Trump in the Electoral College, 332-206. https://t.co/hNgvcaeCjo

Given that track record I thought it worth letting people know.

The write-up notes that the Democrats are running ahead in Ohio, Virginia, Pennsylvania, Florida and Nevada.

It acknowledges recently polling data that might seem to indicate differently, but Moody goes beyond polling data.  As I note in another writeup about his model, from a pro-Republican writer in this WXYZ piece

t works like this: Moody's Analytics breaks states down according to how the electoral college will vote and also uses the economy as a predictor for how an election will sway.

The good news for Republicans is we're still six months away from the election, and predictions deserve every grain of salt we can provide. The bad news? Moody's is remarkably reliable.

Moody's Analytics has correctly predicted who'll be president in every election since 1980 and has a 90 percent success rate in guessing state outcomes.

Returning to the write-up in The Hill,

“President Obama’s approval rating has crossed over the important 50 percent threshold for the first time in almost four years,” said Dan White, a Moody’s economist who oversees the model.  

Aside from rising gasoline prices, a drop in Obama’s approval rating is the only model variable that could possibly move far and fast enough to push the model in Republicans’ favor by November, White said.

“This sudden surge could be a result of the messy primary season or a relative lull in geopolitical news from overseas,” White said.

We are going to see a welter of models, some of which including polling data, some of which don’t, just as we already have during the primary season (and so far, our own Benchmark Politics has done a remarkable job of predicting the results based on county level demographics with some polling data.

And incidentally, here are a couple of recent tweets relevant to the forthcoming California primary:

Looking at CA, I see no way Sanders competes. I can't think of a county in a primary where he has won any place with >30% Hispanics.

We have Clinton +12 in California vs Sanders. So many counties 40-50% Hispanic... These are counties Sanders has extreme trouble with.

Returning to the write-up:

Moody’s model awards Electoral College votes to each party based on state-by-state outcomes.

The most important economic variable is income growth by state, including job and wage growth, hours worked and the quality of the jobs being created in the two years leading up to an election. The model also factors in home and gasoline prices on a state level, as well as the presidential approval numbers.

The Moody’s equation also includes an additional dummy variable that penalizes Democrat incumbents, stemming from the theory that Democrats and Democrat-leaning independent voters are more likely to switch sides and vote for a Republican candidate than vice versa. 

Polls are indicators at a moment of time.  Models can predict to some degree based on the data currently available. 

The California primary is still more than two weeks out.

The general election is more than 5 months in the future.

Thus I take nothing for granted, on either event, although I have a fairly high degree of confidence in the outcomes of both.

Offered for your cogitation.

Make of it what you will.