The polls, the Electoral College and Democratic hand-wringing
There are daily calls or emails from friends — Democrats — unnerved by reports that battleground states now are toss-ups, over a headline last week on a poll showing that Biden’s lead is narrowing in Wisconsin.
In the final month and a half of the most intense, consequential presidential contest of our lives, the chatter among much of the press and politicos is about numbers: the polls and the 270 electoral votes needed to win.
It’s easy to caution against obsessing over small shifts; I do each day, right before turning to RealClearPolitics or FiveThirtyEight to get the latest data.
Here’s a guide on how to view those numbers in the stressful days and weeks ahead:
The Electoral College is a factor only if the vote is decided by about two points or less. That’s been the case in two of the last five contests, but the odds are for a more decisive outcome.
If it is close, there are two must-win states: Florida for Trump, and Pennsylvania for Biden, which means he’s carrying Michigan too.
If it’s a really tight outcome, the listed battlegrounds are Arizona, Wisconsin and North Carolina, all carried by Trump last time. He would have to win all of them — or if he loses one, take a couple 2016 Clinton states, such as Minnesota or Nevada and New Hampshire.
The nervous Democrats — and some press reports — make a big deal over the much closer contest in the battleground states than overall. That current national four-point spread is the same as four years ago — except Biden is running around five points ahead of Clinton.
On polls, that Wisconsin tally last week that had Biden leading, 47 to 43 — compared to 49-44 a month before — suggested, as the Milwaukee Journal reported, the “stability” of the race, not a narrowing lead.
It’s tempting to focus only on the battleground states, not the national surveys. But the scope of any national outcome matters. In 2016 most of the major national polls got it right, showing Clinton narrowly winning the popular vote.
The best, I believe, are Iowa-based Ann Selzer and the Wall Street Journal/NBC News poll, perhaps reflecting that I worked years with both. Others with a good track record include the Washington Post/ABC poll and surveys from Quinnipiac and Marist.
State polls last time had a poor record, as a number of them stopped polling too early; Trump outperformed in most of these surveys. There are good ones: Selzer’s Iowa polls and the Marquette Law School’s Wisconsin survey cited earlier.
For pathological poll watchers, here are some rules of the road for the next 48 days: knowing all polls aren’t the same, look at their track record.
Make sure they reflect the demographic breakdowns. Gender has become a partisan divide, with Republicans much stronger with men. Women in most states comprise about 52 percent of the vote.
The methodology of each poll is important. Surveys that rely on phone interviews, a mixture of landline and mobile, generally are more reliable than online polls or those conducted with automated digital calls.
With an electorate deeply polarized between two well-known men, be wary of polls with a double-digit undecided. If one, for example, shows a 17 percent undecided black vote, it’s pretty safe to assume 80 percent to 90 percent of those votes are going to Biden.
Every good poll is weighted at least for gender, race, geography and educational levels to project the most likely composition of the electorate. No good poll weights for party preference; it’s not a fixed variable: “That’s a red flag, as the electorate shifts enough that nobody knows for certain,” Selzer told me.
Education level is crucial, as more-educated voters are now decisively Democratic. Sean Trende, the perceptive senior elections analyst for RealClearPolitics, says the 2016 polls, especially in the midwestern states — where the supposed “blue wall” crumbled — “were badly underestimating” turnouts for white non-college educated voters.
Trump got more than two-thirds of this group and among white males without a college degree got almost 70 percent in Michigan and Wisconsin.
Trende is not sure polls have made appropriate adjustments this time. One who gets this is Jim Gerstein, a leading Democratic poll taker, who acknowledges the issue and says his surveys often are weighted to reflect more non-college educated voters: “Those with college education are more willing than others to take the surveys.”
There are any number of key sub-groups to monitor: independents, suburban voters, Latinos. Keep an eye on seniors whom Trump carried by eight points last time but who are tilting Biden today.
Finally, always look first at Trump’s number: Anything that’s not above 45 percent nationally or in any important state spells trouble for an incumbent.
Al Hunt is the former executive editor of Bloomberg News. He previously served as reporter, bureau chief and Washington editor for the Wall Street Journal. For almost a quarter century he wrote a column on politics for The Wall Street Journal, then the International New York Times and Bloomberg View. He hosts 2020 Politics War Room with James Carville. Follow him on Twitter @AlHuntDC.