Saturday, November 12, 2016
Norpoth vindicated - or not? / Trump 2016, Truman 1948
Did Professor Helmut Norpoth predict Trump's victory on November the 8th?
When I first encountered Helmut Norpoth's primary model, I was impressed by it and its consistent track record of predicting the two-party vote in presidential elections (within a margin of error) which in turn determined, more often than not, the winner. He predicted that the Democrats would win 50.2% of the two-party vote in 2000: the party went on to win 50.3% - so this was a highly accurate call - but went on to lose the electoral college vote; in his own defence, Norpoth states that his model only makes the claim to predict the two-party vote, not the electoral college result, and so, in this instance, didn't really predict the winner.
After November the 8th, his model has garnered a great deal of attention because it predicted a Trump victory through the two-party vote. Norpoth swam against the tide this year, because his was one of only three models - the others being Allan Lichtman's and Alan Abramowitz's - to state that Trump would win. Norpoth deserves the acclaim for sticking his neck out and getting right. But there's a problem: Norpoth's model predicted that Trump would win 52.5% of the two-party vote, and while Trump was ahead in the popular vote on election night, he's been behind for several days now. Below we see the results so far from CNN:
So Trump has only won 49.7% of the two-party vote to date, but, like Bush 43 in 2000, won the electoral college (albeit to a greater degree than Bush). Norpoth's model did not predict this.
As of now, millions of votes haven't been counted, and it could be that Trump will pull ahead, thereby proving Norpoth right. This will take time, however, and so we shouldn't be showering Norpoth with plaudits just yet.
Trump's victory bears an uncanny resemblance to Truman's in 1948. Truman won 303 electoral college votes, Trump is projected to win 306; Truman won 28 states, Trump 30. Norpoth predicted Trump would win 52.5% of the two-party vote (and perhaps Trump will win something approaching that figure); Truman won 52.4%. In 1948 and 2016, a consensus of media and pollsters concluded that both Trump and Truman would lose. And, just before November the 8th 2016, Newsweek magazine even gave us a Chicago Daily Tribune moment, with a President-elect Clinton commemorative issue printed up before November the 8th:
These issues were recalled, and now fetch a high price on EBay.