In early May of 2019, Elizabeth Warren found herself trailing Joe Biden in the RealClearPolitics polling average by over 33 percentage points (41.4% to 8%). Over the next three months she began a rapid rise. By Sept. 9, she had overtaken Biden in the weekly YouGov poll to become the front-runner for the Democratic presidential nomination.

In the following months, YouGov routinely placed Warren at over 30% support among likely Democratic primary voters. However, the last month has cost the Massachusetts senator significant momentum. She’s fallen below 20% in the YouGov survey and to about 16% in the RCP average.

Journalists and commentators have attributed this slide to her support for a “Medicare for All” health policy. But while her health care positions may have hurt her to some degree, it seems unlikely to be the sole -- or main -- cause of her decline. In early October, when she was at her high point, about one quarter of Democrats said health care was the most important issue to them and, of those voters, 28% preferred her (23% supported Biden and 20% preferred Bernie Sanders). This suggests that even a substantial shift among these voters would not account for Warren’s rise and her subsequent fall.

Rather, the barrier for Warren is a broader concern about her electability in November. In the early September YouGov poll -- when Warren first overtook Biden -- she was doing quite well among liberal Democrats (who make up about 55% percent of likely primary voters) and comparatively worse among the 40% who identify themselves as moderate or conservative. Table 1, below, shows the shifts in support for the top four candidates over the past four months.

In September, Warren had twice the support of Biden among liberals and a 17-percentage-point lead over Sanders. By the December poll Warren had fallen among liberals to 24% while Biden and Pete Buttigieg had both gained support. Among more moderate Democrats, Warren remained at about 10% while Biden increased his lead a little and Buttigieg rose to double digits. Sanders stayed about the same among liberals, but lost seven points among more moderate Democrats. What this table demonstrates is that Warren’s drop in the polls can be attributed to her standing among liberals, who supposedly favor her Medicare for All ideas.      

Why, then, has Warren slipped?  One answer is electability. Even if primary voters are closer on issues to one candidate, they will often vote for another if they think the second candidate is more likely to win the general election.  In the 2004 contest for the Democratic presidential nomination, for instance, former Vermont Gov. Howard Dean had a 2-to-1 lead over John Kerry late in 2003, with the Iowa caucuses only a month away. Yet, when the ballots were counted, Kerry won twice as many delegates as Dean and went on to win 12 of 14 contested primaries and the nomination.

Why the switch when polls generally showed voters closer to Dean on the issues? The answer is that Democratic primary voters wanted badly to defeat incumbent George W. Bush.  Writing for Slate, William Saletan observed at the time, “Kerry won by racking up a 4 to 1 advantage over Dean among voters who chose their candidate because he could beat George W. Bush in November.”

In the 2020 primary season, a similar force might be at work. When asked whether it was more important that the potential nominee agree with them on the issues or that the nominee could unseat Donald Trump, 60% of likely Democratic voters chose the latter. More importantly, when this number is broken down by ideology, it is liberals -- rather than moderates -- who place greater emphasis on winning than on agreement (67% to 57 %).  Given the level of distaste for Trump among progressives, it is not surprising to that they want to pick a winner.

Table 2, below, compares perceptions of electability for the four top candidates over the September-to-December time period. Looking at all likely voters, Biden has slipped somewhat during that interval but still has 63% believing he can unseat the incumbent president. Warren has fallen from 63% to 49%, with Sanders is at 50%. Buttigieg moved up slightly, but two-third of Democrats don’t see him beating Trump. Liberal Democrats have less faith in Biden in December than they did in September, but he now leads Warren 63%-59% among that demographic. 

In September, 73% of liberal Democrats thought Warren would beat Trump; that number is now down to 59%. Sanders’ numbers remained unchanged at 60% over these three months, while Buttigieg fell slightly. Among moderate/conservative Democrats, Biden held steady while both Warren and Sanders fell to 37% from 49% and 46%, respectively. In sum, Warren has fallen because Democrats, especially liberal Democrats, perceive that she has an electability problem -- in all likelihood largely driven by her positions on health care and the economy. Sanders, the other major candidate on the left, appears to have a similar electability problem, but his slide has been among more moderate Democratic voters.

Fortunately for Warren, her recent struggles have not dropped her out of contention for the nomination.  Despite having fallen into a near-tie for second place with Sanders, she remains the candidate with the most “second-choice” votes (13%) and few Democratic voters say they would be disappointed by her nomination (17%).  These metrics suggest that she has the potential to recover her previous position. 

David Brady is a professor of political science at Stanford University and the Davies Family Senior Fellow at the Hoover Institution.