Friday, August 28, 2020


In 2016, Hillary Clinton led Donald Trump in nearly every poll throughout the campaign. In one poll, however, Trump regularly led: The USC Dornsife Election Daybreak Poll, which was done in conjunction with the Los Angeles Times. In that survey, Trump routinely led by 1 to 4 points; the final USC Dornsife survey gave Trump a 3-point lead. (Since the poll was intended to estimate the national popular vote, this was actually off the mark -- Clinton won the popular vote by 2 points.)

The USC Dornsife poll is back this year, and so far the results are a bit different: Joe Biden is currently at 52.73%, while Trump is at 40.32%. That's more than a 12-point Biden lead.

And USC Dornsife is reporting daily numbers. The numbers above were posted yesterday. The numbers from Sunday, August 23 -- before the Republican convention started -- were Biden 52.23%, Trump 40.45%. So Biden has gained a tiny amount of support during the convention, while Trump has slipped by a similar amount.

I don't know whether to believe these numbers. This poll wasn't accurate four years ago. But the pollsters certainly can't be accused of having dismissed the possibility of a Trump win and building that bias into their approach.

Four years ago, an explanation of USC Dornsife's polling method was posted on its website:
The poll’s probability approach sought to measure participants’ level of certainty in their plans to vote for the candidates, rated on a scale of 0 to 100, and the intensity of their commitment to a candidate, rather than simply their preference for one candidate or asking them outright whom they would vote for in an election. It is based on a methodology used by the RAND Continuous Presidential Election Poll that was successful in 2012....

The numbers showed Trump ahead for most of the poll’s run.
If the methodology hasn't changed, and Biden is well ahead, that suggests that pro-Biden (or at least anti-Trump) voters are more intense in their feelings than Trump supporters -- which is contrary to what many observers believe.

I think that's a good sign.

No comments: