The 2020 Democratic primary process is heating up. The field of Democrats is already winnowing, and there are a lot of data-driven tea leaves that pundits are using to prognosticate the possible outcome. We took a look at one particular set of tea leaves and find Warren and Biden to be in good positions as the wide field of candidates narrows.
We came to this conclusion by looking at Democratic voters’ second-choice candidates. It may seem weird to look at voters’ second choice preferences for candidates, but if we assume voters can rank order their preferences, and the field will narrow, then we can use information about voters’ second choices to anticipate how support for remaining candidates will change, when candidates drop out.
We can also use the data to help us understand the consensus point of the Democratic Party. During a primary process a party is focused on finding a candidate who (1) can win in the general election, and (2) represents the consensus of Democratic party policy views.
“Electability” has been a high priority for Democratic voters, although it’s not clear what that means. But the other consideration is, which candidate best represents a cross-section of the party’s views? The answer is not necessarily the most popular candidate—Donald Trump led the GOP 2016 polls but was clearly not a standard-bearer for the classic Reagan coalition. Instead, parties tend to seek a candidate who is acceptable to all party factions.
A recent poll by Morning Consult provides one way to measure “acceptability.” Morning Consult’s pollster asked Democratic primary voters who their second choice candidate would be, and provides data on the second choice preferences among supporters of the top five candidates: Biden, Warren, Sanders, Harris, and Buttigieg.*
Using the data provided by Morning Consult, we took a look at the second choice candidate preferences for the polling sample. We represent a visualization of those choices in the following graph, known as a chord diagram, because it can help us see movement from one candidate to the next. Looking at these rankings can help us to see how the field might change as candidates drop out of the race.
The graph is a bit complicated, but once you know what to look for, it’s quite revealing. To see how to read it, take a look at Harris as an example. Voters who support Harris as their first choice are shown in maroon. The maroon stripes that go to Warren, Biden, and Sanders, in decreasing order of thickness, show that Harris’s supporters prefer those candidates as their second choice. This means, if Harris were to drop out, most of her supporters would go to Warren, some to Biden, and a few to Sanders.
Sanders benefits a lot if Biden drops out, some if Warren drops out, and a little if Harris drops out, but he picks up none of Mayor Pete’s supporters if he decides to call it quits.
Looking around the diagram, it appears that Buttigieg is no one’s second choice and is unlikely to benefit if any of the top 4 candidates above him were to drop out. Also, Harris does not appear to gain that much from other candidates dropping out. Warren and Biden appear to be the second choice for many who were polled, and both will benefit if others in the field call it quits. As it stands, Warren gains more overall support from voters who don’t already rank her first, than Biden does. She gains more than any other candidate from any winnowing.
The very visible primary process that Democrats are going through provides wonks with a lot of data to help us understand the process. Using this particular snapshot of a sample of Democratic voters’ rankings, Warren has an upper hand at the moment. Recent data from online betting markets are consistent with this view, where Warren has overtaken Biden across the board. We show the summary of those markets here.
Warren’s trajectory has been positive over the last few months, where other candidates have been static, declined, or had only momentary surges of support. We will have to wait and see how long the trend will last.
* Specifically, the top three choices of supporters for the top five candidates. Thus, we have no data on:
a) the second choices of voters supporting lower-ranked candidates (such as Tulsi Gabbard)
b) the second choices of voters who supported a top candidate, but whose second choice was a lower ranked candidate (e.g. a voter who preferred Harris, then Castro)
c) the second choices of voters who supported a top candidate, but preferred Buttigieg. For each of the other four candidates, Morning Consult does not report how many voters would switch to Buttigieg, perhaps because he was lower ranked, but we can’t be sure.