Ruy Teixeira and John Halpin, two liberal politics watchers and activists associated with the Center for American Progress, believe they’ve figured out how Donald Trump loses. They’ve identified what they call Trump-Biden voters (voted for Trump in 2016 but now seem ready to vote for Joe Biden). Teixeira and Halpin surmise that such voters are more numerous than others who voted for Hillary Clinton in 2016 but now say they will vote to re-elect Trump), enough to change the result. That’s an oversimplification. The headline, with proper caution, says “Trump-Biden voters could decide the 2020 election.”
Of course they could. Or maybe not. The Teixeira-Halpin piece is properly framed and hedged. Naturally, I was glad to see their numbers. But at some level, I know better. Yes, I spend a chunk of time every day looking at polls. I’ll tell you what I’m seeing, much of which you already know, but first I’ll emphasize why it’s not such a great use of time, especially six months before Election Day (assuming Trump doesn’t succeed in canceling the election).
Merely a snapshot
As I and every poll addict like me should constantly remind ourselves, fresh poll results are, as the cliché has it, merely a snapshot in time. Another, harsher saying has it that fresh poll results are like “crack cocaine for politics junkies.” Not only are they addictive, but their effect doesn’t last.
By the time you read a poll result, it’s out of date. And the out-of-date snapshot is also somewhat out of focus. As you know, every poll result comes with a margin, plus-or minus three-or-four percentage points for “sampling error,” which mostly means it’s an estimate, based on the fact that only a sample of the electorate was interviewed and no sample perfectly captures the larger population it “represents.”
Then there are problems like what portion of those identified as likely voters will actually vote, and which the undecideds break, and so on.
You know all this. But your (and my) great desire to know what’s going to happen on Election Day may overpower our ability to take all that into account enough to remember that that shaky portrait of the recent past is not a reliable picture of what will happen on Election Day.
If you (yes, like me) are extremely anxious, even more so than usual, to know how the presidential election of 2020 will turn out, you should mostly disregard national polling samples — as two of the last five presidential elections have reminded us, thanks to the Electoral College, a terrible system that doesn’t even reflect the intentions of the framers who gave it to us.
(If you have the time, I explored the enormous differences between how the Framers pictured this system working and what has evolved as part of a longer series on the Constitution here. Of course few of you will do this, but this is the miracle of the internet. And I also listed 10 problems with the Electoral College system in the same series, accessible via this link. And I wrote all of this when it had nothing to do with the current incumbent, elected with neither a majority nor even a plurality of the popular vote. And please note, I stated that long-held position on the Electoral College before the 2016 election.)
Currently, the picture augurs well for Biden
Setting aside all those irrelevant historical and constitutional diatribes, and ignoring my own advice above, I obsess on the latest polling and, as you probably know, the picture augurs well for Joe Biden winning the November election (assuming the current incumbent doesn’t find a way to call off the election off, steal it or limit the franchise to rich, angry white males).
According to most recent polling, Trump does poorly among females, poorly among non-whites, poorly among the young and, as I mentioned recently, he does poorly among the fairly large group of voters who say they dislike both him and Joe Biden. These groups make up some pretty dang big pieces of the overall electorate, more than a supermajority. If you look at current polling of a nationwide sample, most such polls show Biden ahead by more than five but less than 10 percentage points.
I’m forcing myself (and so should you) to be cautious about such polls. It’s way too early. The leads are not insurmountable. If you take the full “margin for error” seriously, they reflect a result that could actually be a small Trump lead, and personally I expect to see Trump pull some mighty questionable maneuvers to disenfranchise those who might vote in a way that displeases him, or to postpone the election, or to refuse to accept the result. So now you have some sense of my worries.
If we are going to obsess on polls at all (I recommend against it but can’t help myself), it’s the polls in the swing states that offer the best clues to how things are going.
There, too, things look mighty good for Biden. Polls generally show Biden leading (although the leads are mostly within the margin for error) in enough swing states to win a pretty large electoral vote majority (if they come out the way they currently look, if, if, if).
For any of you who are (like me) so obsessed that you will look at a bunch of state-level polls six months before Election Day, I’ll tell you how to find them (or, at least, how I find them).
This link will get you to a page on FiveThirtyEight.com, the site maintained by obsessive political number crunchers under Nate Silver, showing the most recent polls in the presidential race. Most of them are national numbers, and look at them if you like. But I use that link to get the latest polling in the swing states.
When I looked Thursday at noon, it had Michigan polls by four different pollsters, all of which showed Biden leading in that key swing state by between three and eight percentage points. (Again, take into account margins for error.) By the way, LV next to a poll means likely voter. RV means registered.
The most recent Pennsylvania poll, and the only one on the list, had Trump up by four.
The only Wisconsin poll had Biden up by nine.
Everyone knows that those were the three key swing states that enabled Trump to unexpectedly win the Electoral College. But, again relying on polling, there are a lot more swing states than those three, including several that might surprise you if you don’t follow these things obsessively.
For example, there are three Florida polls on the current list, all of which show Biden leading, by margins ranging from one to six percentage points. Florida and Ohio used to always be considered two big swing states. And Trump carried Florida by just one percentage point in 2016, but for some reason, the three I just mentioned get most of the buzz. Anyway, the three Florida polls all show Biden leading, by six, one and four percentage points.
Speaking of Ohio, which Trump carried by an impressive eight points in 2016, FiveThirtyEight.com doesn’t list any, and the most recent two I could find elsewhere show Trump up by three points in one and Biden leading by three in the other, taken two weeks apart in late April and early May.
Many states aren’t listed on 538, and, of course, the pollsters are mostly interested in the ones considered close. But there are some surprises on that list for those who don’t obsess as much as I do on these things. For example, the most recent poll out of Georgia showed Biden leading Trump by an insignificant 48-47.Georgia has gone red in the last six elections and eight of the last nine.
Most states don’t make the 538 list, presumably because they haven’t been polled recently, and that’s presumably because they are not considered to be seriously “in play.” Interestingly, Minnesota, which used to be listed as a potential Trump pickup because he came so close in 2016, didn’t make the list.
If you want to join me in obsessing on this stuff (I don’t really recommend it, but I can’t help myself), save that link, check it once a week or so, and, my advice is, don’t waste a lot of time on the national numbers — focus on the swing states. Or, even better, take a walk in the woods and come back in October (maybe September).