Whom the Democrats Nominate in 2020 Matters -- a Lot
Conventional wisdom about presidential campaign strategy changed around the turn of the current century. Traditionally, candidates were advised to move to the center in the general election campaign after catering to the party bases in the primaries. Not anymore. George W. Bush’s two presidential campaigns exemplify the shift.
In 2000, Bush ran as a “compassionate conservative” appealing to a broad cross-section of the country. He won the presidency despite losing the popular vote. Four years later, his campaign focused on turning out the Republican base. The most common explanation for discarding decades-old practice was, as The Economist put it in 2002: “The 50:50 nation appears to be made up of two big, separate voting blocks, with only a small number of swing voters in between.” Academic research supports the view that fewer voters split their tickets now than in earlier decades. Also, fewer Americans change their party preference from election to election. If only a small number of unattached voters are up for grabs, the emphasis of campaigns naturally has shifted to the “base.” Feed them red meat and maximize their turnout.
We have been skeptical of this newfound conventional wisdom. One of the reasons many voters seem more set in their partisan voting inclinations is that today’s parties are much more homogeneous than in the past, nominating the same types of candidates from office to office. For example, liberal antiwar candidate George McGovern in 1972 differed far more from moderate Jimmy Carter in 1976 (“a respectable alternative to George Wallace”) than Barack Obama in 2012 and Hillary Clinton did in 2016. For that reason, we would expect more voters to change their decisions between 1972 and 1976 than between 2012 and 2016. The question takes on heightened importance in light of the crowded field of candidates for the 2020 Democratic presidential nomination. Many Democrats consider Donald Trump such a flawed president that they believe any Democrat can defeat him. If so, the party has an opportunity to elect a president closer to the views of the base than it would with a more positively regarded Republican incumbent. Some data support this view. One recent poll included trial heats pitting eight potential Democratic candidates against Trump. All eight won.
Is it really the case that any Democrat can beat Trump? On closer inspection, that outcome looks less certain. For starters, history shows that one shouldn’t put much faith in trial-heat polls 18 months ahead of a presidential election. Moreover one recent survey experiment by YouGov indicates that a surprisingly large proportion of the electorate – about 40 percent -- reports that the choice between President Trump and a Democratic challenger depends on the identity of the Democrat. These voters are “in play” or “up for grabs.” There are more of them than there are completely committed Democratic or Trump voters.
YouGov polled 3,000 Americans between Feb. 28 and March 3. We asked our respondents how they would vote in an election between Trump and nine declared or potential Democratic candidates (Joe Biden, Cory Booker, Julian Castro, Kirsten Gillibrand, Kamala Harris, Amy Klobuchar, Beto O’Rourke, Bernie Sanders, and Elizabeth Warren). To avoid voter fatigue we did not ask each voter to make eight decisions. Rather, each voter chose between Trump and three randomly selected Democrats, so roughly 1,000 respondents participated in each trial heat. If respondents said that they would vote for the Democrat in all three trial heats or Trump in all three, we then asked: Would you always (never) vote for any Democrat?
Democrats led in seven of the nine trial heats, but the “always Trump” and “never Trump” blocks were nearly identical in size: 36 of respondents said they would vote for the Democratic candidate in all three matchups and 35 percent said they would vote for Trump in all three. When asked the follow-up question, about one in seven said that there was a Democratic nominee who would cause them to switch parties. The good news for Democrats is that there are more Trump supporters who are willing to consider voting for a Democrat than vice versa. The bad news for Democrats is that 40 percent of the electorate say that their vote in 2020 depends upon the identity of the Democratic nominee. This is the largest segment of the electorate and consists of three groups:
- People who preferred Trump over a Democrat in one or two of the three trial heats and a Democrat over Trump in the others (4%).
- People who favored all of the three Democratic candidates in the matchups we asked them about, but in the follow-up question said that there was a Democrat who, if nominated, would cause them to vote for Trump (11%).
- People who were unsure of how they would vote in at least one of the three trial heats (26%).
Unregistered voters are disproportionately represented in the last group. Among registered voters, the Democratic advantage is larger (38% never Trump, 31% always Trump, and 31% depends), but nearly a third of registered voters are “available” for Trump depending on whom the Democrats nominate.
Who are these potential swing voters? They fit the expected description. Over half (56%) identify themselves as independents; 38% classify themselves as “moderate” or are non-ideological; 40% did not vote in the 2018 midterms; 8% said they voted for Trump in 2016 and a Democratic congressional candidate in 2018. They appear to be in between the parties, neither MAGA true believers nor members of the resistance. They are not part of either party’s base.
In liberal states such as California, New York, Illinois, and Massachusetts, to name the more prominent ones, Trump is not competitive. It is almost certain that he will ignore them and aim for an Electoral College majority rather than a popular vote majority. For that reason, we ran additional analyses on: (a) the 17 states where no candidate won by greater than 10% in 2016; (b) the 13 states where a swing of 5 points or less would have changed the outcome; and (c) on six competitive “Midwestern” states (Iowa, Minnesota, Michigan, Ohio, Pennsylvania and Wisconsin) where Trump will try to eke out an Electoral College victory in 2020. The following table contains these alternative figures.
These numbers differ very little from the national figures, although Trump does a bit better in the competitive states.
What about the results of the national trial heats between the nine Democrats and Trump? Mock heats provide little information this far in advance of the election, and largely reflect name recognition. But for what it is worth, over the entire national sample Biden and Sanders do best against Trump, each garnering 46 percent. Six of the nine Democrats get a plurality against Trump, although none of them gets a majority. (Trump does not lose to O’Rourke, Harris or Castro, but the undecideds are especially high in their trial heats.)
The polls tighten up in the competitive states, with seven of the nine Democrats either tied or trailing Trump — only Sanders and Gillibrand maintain small leads. In these states the voters are more likely to report they are not sure whom they would vote for in the matchups that involve lesser-known candidates. In the competitive Midwestern states, Sanders turns in his best performance, leading Trump 50%-33%. Biden and, perhaps surprisingly, Booker, have six-point leads in those states. Castro and O’Rourke trail badly.
In sum, these findings suggest that at this time the pool of potential swing voters in 2020 is larger than commonly believed. We are a year and a half away from the election, of course, but barring an economic slowdown or serious foreign policy crisis, we can probably expect little change on the Republican side. Donald Trump’s persona is already “baked in” to voter judgments. Whether he is reelected or rejected is largely in the hands of Democratic primary and caucus voters.