Many Democratic politicians and voters rallied strongly behind Joe Biden before and after Super Tuesday, aiming to halt what appeared to be Bernie Sanders’s momentum. Many did so because they feared Sanders is too ideologically extreme to beat President Trump in November. Were they right?

In short, probably so. We’ve researched a century of U.S. presidential elections and found that a moderate Democrat is more likely to beat Trump in November.

How we did our research

We examined the Republican candidate’s share of the vote in the 25 presidential elections between 1920 and 2016. Past research has shown prosperity and peace generally make it more likely that the incumbent party will win, so for each election, we accounted for both the economy and U.S. involvement in international conflict. We also adjusted for public appetite for change, which can happen when one party holds the presidency for more than two consecutive terms. Finally, we examined where each party’s candidates stood on a “left-right” (liberal-conservative) scale in successive presidential elections.

We measured parties’ ideological positions by analyzing the policies promised in their election platforms, using a “Manifestos” database that codes these policies on quantitative scales for each party for each of the 25 presidential elections between 1920 and 2016. This database includes summary “left-right” (liberal-conservative) scales for each party for each election, based on policies such as government involvement in managing the economy and promised spending in areas such as health care, social welfare and defense. Our model includes these summary left-right variables for the Democrats and the Republicans in each election.

Republican presidential candidates gained votes when the Democrats moved leftward

Our statistical model then analyzed the Republican candidates’ percentage of the popular vote in each election, accounting for all these factors. The chart below shows the Republican candidate’s percentage of the vote in each of the 25 presidential elections from 1920 to 2016, against the Democratic Party’s scores on the “Manifestos” left-right ideological scale. A lower score indicates a more left-wing position.

As you can see, the Republican popular vote tends to be higher in elections when Democrats are further to the left ideologically.

Let’s look at some examples. We found the most left-wing Democratic manifesto in the 20th century was offered by Adlai Stevenson in 1952 — who was easily defeated by Republican Dwight D. Eisenhower. Twenty years later, Democrat George McGovern offered a very left-wing manifesto in 1972 — and lost overwhelmingly to President Richard M. Nixon. Running on an equally left-wing platform in 1980, President Jimmy Carter was soundly beaten by Ronald Reagan.

Meanwhile, in 1992, Bill Clinton offered the most centrist Democratic manifesto of the entire 20th century — and defeated President George H.W. Bush. Barack Obama won in 2008 with an equally centrist platform; that was revised leftward in 2012, and Obama lost ground — but kept the presidency. Four years later, former secretary of state Hillary Clinton doubled down on Obama’s 2012 move and ran on a platform as left-wing as McGovern’s and Carter’s. Although she won more individual votes than Trump, Trump won the electoral college.

What does this tell us for 2020?

These results clearly suggest Biden is more electable than Sanders. We treated Sanders as a Scandinavian-style social democrat, giving him the Swedish Social Democratic Party’s average position on a left-right scale since 2000. Because we did our work before the emergence of covid-19, we further assumed also that economic growth would stay at its current rate (2.1 percent), that the country is not at war and that the Republican platform is at the same left-right position as it was in 2016. Our model predicts that the Republicans would get 54.7 percent of the popular vote and a large majority (368) in the electoral college. If Sanders is the nominee and the coronavirus pandemic or some other factor kicks the economy into a recession like that in 2008, the result is still a Republican victory, with the GOP getting 51.6 percent of the popular vote and 334 electoral college votes.

We find something different with Biden. We assumed that Biden would run on a moderate platform, much like Obama’s in 2008. In this scenario, our model predicts the Republicans would get 48.5 percent of the popular vote and a knife-edge majority (273) in the electoral college. But if Biden is the nominee and the economy is in recession, our model translates the result into a 45.7 percent Republican popular vote and 219 GOP electoral college votes. If Biden runs on a moderate platform in the context of a struggling economy, he decisively defeats Trump.

To be sure, these results come with caveats. Studies that use survey-based measures of candidate ideology have concluded that candidate ideology is only weakly tied to election outcomes. Also, Democratic voters have moved to the left over the past three decades. This could make it more difficult for an ideologically extreme candidate like Sanders to split the party, as they have in the past.

Nevertheless, our findings suggest moving sharply to the left could harm the Democrats’ prospects of evicting Trump from the White House — and that Biden appears to be a better bet for the nomination.

Harold Clarke is the Ashbel Smith professor in the school of economic, political and policy sciences at the University of Texas at Dallas and co-author with Matthew Goodwin and Paul Whiteley of “Brexit — Why Britain Voted to Leave the European Union” (Cambridge University Press, 2017).

Paul Whiteley is a professor in the department of government at the University of Essex and the co-author most recently of “Brexit — Why Britain Voted to Leave the European Union” (Cambridge University Press, 2017).