16-Feb-24 Elections

Tales of the unexpected

Donald Trump’s electoral performance in both 2016 and 2020 suggests that returning to the White House is likely to prove an uphill struggle.

Modern U.S. electoral history is full of surprising comebacks and stumbling frontrunners. At this time four years ago, Joe Biden’s presidential bid was in the doldrums, following devastating losses in Iowa and New Hampshire.[1] Biden’s middling showing in Nevada only reinforced the conventional wisdom that the party’s activist base would deliver the nomination to Bernie Sanders; after all, the leftwing Senator had won the most votes in each of the first three states.[2] Events such as Biden’s subsequent comeback in South Carolina are worth recalling, whenever pundits are all too certain about how they expect a U.S. election year to unfold by extrapolating from early contests.[3] All simple rules of thumb tend to be broken – eventually.

The underlying reason is that U.S. presidential election forecasts typically rely on small samples of not necessarily representative data, given that elections only happen once every four years. Making reliable inferences typically requires judgement calls in how to analyze the available data and being clear eyed on what the data actually shows. Our Chart of the Week starts off election coverage by looking at popular vote shares of major U.S. presidential candidates in 2016 and 2020, compared to all runner ups since 2000, with 1988 also added for comparison.[4]Superficially, it might seem like a rematch between both major parties’ elderly front runners should be a toss-up. After all, both Biden and Donald Trump have won before.

Popular vote shares of Biden and Trump in 2020 vs. previous losing U.S. presidential candidates

Sources: Federal Election Commission, DWS Investment GmbH as of 2/13/24

However, our chart should serve as a reminder of just how different those two victories were. Biden won in 2020 with 51.3% of the popular vote, a vote share tending to entail victory in most electoral systems.[5] By contrast, Trump’s 2016 share at 46.1% of the popular vote nationally was roughly in line with such losing presidential candidates as Mitt Romney in 2012, John McCain in 2008 and, indeed, Trump himself in 2020.

Of course, what counts in the end for the U.S. presidency is not the popular vote but the Electoral College (EC), a topic we will no doubt return to regularly. Suffice it to say here that since the 1980s, campaigns have grown increasingly sophisticated in targeting voters in battleground states to gain EC advantage.[6] The problem, for campaigns and forecasters alike: Which states will be decisive and what it will take to win often critically depends on which third party or independent candidates gain ballot access.

As we argued four years ago, the most reasonable way to think about the EC early on in any given election year is that it massively increases uncertainty in both directions.[7] Still, Trump’s electoral performance in both 2016 and 2020 suggests that returning to the White House is likely to prove an uphill struggle. It is likely to require Republicans either growing their electoral coalition, third party or independent candidates depressing Democrats vote share in critical battleground states or a combination of both. And while head-to-head match-up polling tends to have limited predictive value this early in a presidential race, according to FiveThirtyEight polling does suggest that both Biden and Trump are very unpopular.[8]Meanwhile, the generic Congressional ballot suggests general election voters are roughly equally unhappy with both major parties.[9]Throughout Biden’s time in office, special elections, including last week’s in New York’s third Congressional District, have underlined that Democrats’ vote share has been unusually resilient for the party holding the White House.[10]

If Republican primary voters were to prioritize electability, it would thus not be all that surprising for them to play it safely and choose someone else. And at least by historical standards, it is still early days in the Republican nomination fight, despite Trump’s wins in Iowa, New Hampshire and Nevada. So might Nikki Haley stage a similar upset at the South Carolina Republican primary on February 24, to the one that propelled Biden to victory in 2020? For now, the polls in her home state do not appear all that promising for Haley.[11] To be fair, polling primaries is hard, there have only been very few high-quality ones in South Carolina recently and, in addition, voters in the state’s open primary tend to decide late.[12] Incidentally, South Carolina used to be famous, in modern Republican primary folklore for always picking the eventual nominee. Until it did not, by voting for Newt Gingrich, in 2012, who ended up eventually losing to Mitt Romney.[13]

Read more

Discover more

1. For Joe Biden, Two Bad Losses and a Precarious Path Forward - The New York Times (nytimes.com), Feb 12, 2020

2. 5 Takeaways From the Nevada Caucuses (The Big One: Sanders Takes Control) - The New York Times (nytimes.com), Feb 23, 2020

3. Biden touts landslide victory in South Carolina, swipes at Sanders 'revolution' | Fox News, Feb 29, 2020

4. In other words, only on the instances, where only the two major party candidates were seen as having a realistic chance as has been the case since 2000. That excludes 1992 and 1996, when Ross Perot was running as a third-party candidate.

5. These results, moreover, turn out to be quite robust to election forensics - statistical detection mechanisms of systematic election irregularities social scientists have developed for countries such as Russia at both the national level and for swing states in the U.S. context. See, for example, Kobak, D., Shpilkin, S. and Pshenichnikov, M. (2012) “Statistical anomalies in 2011–2012 Russian elections revealed by 2D correlation analysis.” Physics and Society, https://doi.org/10.48550/arXiv.1205.0741 and Johnson, C. (2021) “Constraining Localized Vote Tampering in the 2020 US Presidential Election.” https://doi.org/10.48550/arXiv.2110.03661 For an accessible discussion of such tools, see Russian elections once again had a suspiciously neat result (economist.com), Oct. 11, 2021. A recent, and methodologically deeply flawed example of how not to do election forensics can be found here: Feb-24-2020-Election-Analysis-vWeb_Final.pdf (heartland.org)

6. See, for example, Issenberg, S. (2013) The victory lab: The secret science of winning campaigns. Broadway Books, pp. 238.

7. An early roadmap to U.S. politics in 2020 and beyond (dws.com)

8. How Popular Is Joe Biden? | FiveThirtyEight and Donald Trump : Favorability Polls | FiveThirtyEight

9. Generic ballot : 2024 Polls | FiveThirtyEight

10. Dems flip seat as Suozzi wins crucial special congressional election in New York | Fox News and Turnout Data Reveals the Core of Democrats’ Success in Special Elections - The New York Times (nytimes.com)

11. South Carolina : President: Republican primary : 2024 Polls | FiveThirtyEight

12. Debates, late deciders propel Gingrich to win in South Carolina - The Washington Post, Jan 21, 2012

13. Newt Gingrich wins South Carolina Primary by uniting Reagan Republicans | Fox News, Jan 21, 2012

CIO View