Stephen Ansolabehere, the Frank G. Thompson Professor of Government, and Elliott Morris, Editorial Director of Data Analytics at ABC News and 538, dissected the 2024 election data in a Center for American Political Studies (CAPS) and Institute of Politics (IOP) event – moderated by Ryan Enos, Professor of Government and director of CAPS.
The well-attended event, held in the Tsai Auditorium in CGIS South, featured an in-depth conversation on the numbers as well as the outcome – with a host of interesting questions from the audience.
Kicking off with an outline from Ryan, in which he introduced Stephen and Elliott, he said “There’s been a ton of commentary on the election in the last couple of weeks. And unfortunately, I think from a lot of our perspectives, almost all that’s basically based on vibes”.
We then launched into a series of discussions and graph assessments, which were produced by all three speakers as well as their counterparts across the US. The trio covered a wide range of data analysis from swing state voting shifts, exit poll data, and forecasting models.
“For me, the non-findings are interesting,” said Stephen. “A lot of them don’t fit into the narrative [such as] no change from 2020 to 2024 in white males voting democrat, the gender gap was smaller, and Hispanic men went 20 points towards Trump”.
Elliott added: “There is also the big mystery of why people voted for Democrat Senators but not Harris. People tend to think of election outcomes by the candidate’s ability to persuade people to vote and turnout – this could corroborate the theory that they do correlate.”
Another big topic of discussion was polling on election night. Elliott heads up 538 – which focuses on opinion poll analysis, politics, and economics – and Stephen is a consultant, supported by a small team of Gov graduate students, for CBS on election night.
Talking about exit polls, Stephen explained that they never call the election from exit poll data but they give a good indication of the national voting direction. He said “In 2016 we knew by 5:15pm, 2020 we knew it would be rocky but looked good for Biden, and again this year we knew from 5pm onwards. We don’t call at that point but it gives a pretty good picture of how the night will go.”
The next marker is then 9:30pm – at which point he had a complete model of Republicans winning by using a House Model. “CBS wanted to keep sticking with groups rather than national tides – but this was a way better predictive model on the night. 12am is the next shot” he added.
We also heard from Gov Ph.D. Candidate Noah Dasanaike – who delivered a brief presentation having won a CAPS competition for his election modeling in which he compared national vs local data and its impact on voting choice. He found a decrease in nationalization over time, comparing data from 1984 to 2024, with his graph showing voting dissimilarity from matched counties.
The event wrapped up with questions from the audience – which ranged from discussing the Democrats’ core base flip to Republicans, global issues of dissatisfaction with incumbent governments, understanding bias in forecasting, and the impact of early voting.
Overall it was a fascinating discussion so thank you to Ryan, Stephen, and Elliott for their insights. The Harvard Crimson also attended the event and published a full write-up, which you can read here.