New York Times Polling Expert Says It ‘Fudged’ Trump’s Election Night Numbers

“We did not allow it to go higher than 95 [percent] for fear of contradicting projections,” Nate Cohn says

New York Times
Getty Images

New York Times polling guru Nate Cohn said the paper “fudged” a number on election night — the 95 percent chance it gave Donald Trump of winning.

TheUpshot election forecasting model started off giving Hillary Clinton an 85 percent chance to win the election, a forecast based on polling data. Near the end of election night, the same model gave Donald Trump a greater than 95 percent chance of winning.

“That was fudged, by the way, a little inside knowledge. We did not allow it to go higher than 95 for fear of contradicting projections,” Cohn said on the Times’ “The Run-Up” podcast.

Podcast host Michael Barbaro simply said, “sure, that’s fascinating,” without asking a follow-up question about what that meant, exactly. Trump, who will take over the White House on Jan. 20, likes to call the media “dishonest” and “unfair,” and this revelation probably won’t do much to change his opinion.

Politico once called Cohn “a precocious numbers-cruncher and polling whiz who has emerged as the new face of election analysis for the paper of record.” But the Times’ initial prediction of a Clinton victory was very off, like most news outlets’ predictions.

Cohn was speaking casually on the podcast with a co-worker, after working long hours. But we asked for more transparency about what he meant by “fudged,” given the flaws in the paper’s earlier predictions.

The Times’ response: “Our live election night chances were displayed as >95%. Note the greater than sign. We know there are errors in live election results and that models are imperfect: they have to make guesses about things like the number of votes left to be counted. The cap reflected this uncertainty. We did this for both candidates, in advance.”

We asked for a little more detail than that, and received this response from a New York Times spokeswoman:

“Fudged was a poor word choice. To be clear nothing was fudged. As I explained earlier, The Upshot model was following protocol to account for uncertainty with the models.”

She then pointed out a Cohn tweet that said, “In case it’s not clear: our >95% tickers don’t go any higher than that.”

We have requested to speak with Cohn for further explanation. Stay tuned.

Comments