Forget ‘Lanes.’ The Democratic Primary Is A Whole Freaking Transit System.

If you’ve read our Slack chats or heard our podcast, you’ll know that I get annoyed whenever the discussion of “lanes” — such as the left/liberal lane versus the establishment/moderate lane — comes up in the context of the Democratic primary. Without getting too far into the weeds,¹ I basically have two beefs with most of the analyses I see.

Beef Numero Uno: These analyses tend to treat voter preferences as being frozen in time — they’re about who the voter supports now, and perhaps who the voter currently lists as their second choice. But to the extent that lanes are interesting, it’s because they tell us something about the route a voter took in coming to a decision. A voter who backed Bernie Sanders in 2016 but has since switched to Elizabeth Warren won’t show up as a Sanders supporter if you ask about her current preferences, for instance. But the evolution of this voter’s choice of candidates would actually be evidence of the existence of a left/liberal policy lane.

Second, these analyses are usually oriented around an effort to either prove or debunk the existence of a particular type of lane, usually the liberal or moderate lanes. But voters come around to their decisions for lots of different reasons. Some voters do vote on the basis of ideology or policy. Some vote on demographics or identity. Some factor in more abstract considerations like “leadership” or “electability.” And some just haven’t thought deeply about their choices yet. When candidates are aligned along several different dimensions — for instance, Pete Buttigieg and Beto O’Rourke have much in common in terms of their demographics and their policy platforms — you can see especially strong correlations in which voters support them, with one tending to rise or fall at the other’s expense. But most of the time, it’s messier, with different threads pulling in different directions.

To extend the lanes metaphor, you wouldn’t expect to see a single, parallel set of lanes connecting all voters in the Democratic primary. Instead, you’d expect something more like a complex, multi-modal transportation network, where some parts of the network are more robust than others, but there are a lot of somewhat redundant options. This doesn’t mean that paths between different candidates are irrational or random, any more than the choices of, say, Manhattan commuters are. But they might look chaotic in the aggregate if you don’t recognize that different voters have different priorities and take different routes to get there.

Let’s think about voters who supported Sanders in 2016, for instance, and which candidates they might be supporting now if they aren’t sticking with Bernie. Voters four years ago might have had several reasons to prefer Sanders to Hillary Clinton:

• Some voters — such as those who do choose candidates on the basis of ideology — liked Sanders because he was proud to wave the flag for democratic socialism. These voters would probably be inclined to back Warren, if they aren’t staying with Sanders.
• Some voters liked Sanders because he was an anti-establishment outsider. I’m not sure where these voters would go today. They certainly wouldn’t be inclined to go to Joe Biden or Kamala Harris, who have the most support from the party establishment. But they might also see Warren as too establishment-friendly. Maybe more eccentric candidates like Tulsi Gabbard and Andrew Yang would work for this group. But I’d expect a lot of anti-establishment types to stick with Sanders. In fact, that seems to describe a lot of the voters he’s currently left with. If you look at donations, for example, a lot of Sanders donors don’t seem to be interested in donating to any of the other candidates, suggesting that he’s somewhat of an island unto himself.
• Some voters liked Sanders because he is a white man. I’m not saying this is an especially large group of voters. But there are certainly some, and maybe quite a few. They have lots of other white guys to choose from if they want to move on from Sanders: Biden, O’Rourke, Buttigieg and so forth.
• Relatedly, voters who associated Sanders with the white working class and supported him on that basis, or who believe that a candidate who does well with white working-class voters is more “electable,” might be attracted to Biden this time around, in addition to Sanders.
• Some younger voters liked Sanders because of his emphasis on policies like free college tuition, and because he came to unexpectedly represent a break from the hegemony of Baby Boomer-dominated politics. It’s not totally clear who these young voters are supporting yet — they can be hard to pin down in polls, and they can sometimes be late to engage in the process. But Buttigieg, Yang and O’Rourke have the potential to overperform with young voters relative to the electorate overall.
• Finally, some voters just aren’t tuned in yet. The notion that Sanders and Biden’s support is just name recognition is mostly misguided, in my view. But it’s certainly true for some voters. These voters’ preferences may simply reflect who they’ve been hearing about in the news lately, which recently has meant Biden, Warren and Harris in addition to Sanders.

So if you run through this list, you certainly do see some major highways that connect Sanders and Warren. But there are also a lot of ways to get from Sanders to Biden. And there are some byways that link Sanders to several of the second-tier or minor candidates, most notably Gabbard, Buttigieg, Yang and O’Rourke. The one major candidate who doesn’t have a lot of obvious connections with Sanders is probably Harris. It’s sort of like trying to get from the West Village to the Upper East Side. You can certainly do it, and they aren’t that far apart as the crow flies, but it requires an extra transfer or two.

All of that might sound good in theory, but is there any data on who Sanders’s 2016 voters are supporting now? Actually, yes! Emerson College, in its national polls, has broken out results based on who Democrats said they supported four years ago. Since the sample sizes are a bit small, I combined the last two Emerson national polls, which were conducted in early and late July, respectively.

This data actually does a good job reflecting our “predictions” from above about how Sanders’s support from 2016 might map to the various candidates this year. (Granted, it’s easy to make “predictions” when you get to see the data before making them, as I did in this case.) Start with some of the lesser-known candidates. Gabbard and Yang might not have that many supporters, but the ones they do have are drawn disportionately from former Sanders voters. O’Rourke and Buttigieg also do better with former Sanders voters than with ex-Clinton ones.

Among the major candidates, there are quite a few Sanders 2016 –> Biden 2020 voters, although not nearly as many as there are Clinton 2016 –> Biden 2020 voters. Harris gets her support mostly from Clinton voters; relatively little comes from 2016 Sanders voters, consistent with our hypothesis.

Warren is, by contrast, drawing about equal shares of Clinton 2016 and Sanders 2016 voters. Maybe you’re surprised that Warren’s numbers aren’t more slanted to former Sanders supporters, but keep in mind that (i) there are plenty of connections between Clinton and Warren too, e.g. in their appeal to college-educated women; (ii) whereas Clinton’s voters need to look around for a new candidate, Sanders 2016 voters have the option of picking Sanders again. One way to look at it is that 44 percent of Sanders 2016 voters are voting for either Sanders or Warren this time around, while just 24 percent of Clinton 2016 voters are.

So there almost certainly is a robust left policy/ideology lane in the Democratic primary. It’s probably even one of the more well-traveled routes. It’s just far from the only road in town.