Ask an expert: The numbers game of the 2024 presidential election
Penn State statistician Cory McCartan brought his expertise to the CBS team during the 2024 election week to help analyze the precinct-level data live and talk viewers through the trends surrounding those results. As the inaugural Hoben and Patricia Thomas and Thomas and Ann Hettmansperger Early Career Professor in Statistics at Penn State, he applies statistics to questions in the social sciences, specifically elections, legislative redistricting, racial disparities and missing data.
In the following Q&A, McCartan discussed his research and the role statistics play on election night.
Q: What was it like working at CBS during election week?
McCartan: It was incredible to be able to work at the CBS News Data Desk and play a small role in understanding the election results as they came in. It's quite a big operation, with producers, technicians, camera crews, news reporters and data analysts all crammed into one studio in Times Square — not to mention the sets, lights, servers and all the other equipment needed to produce a live broadcast. It was great to see all of the data and results coming in and to be able to hear from the other analysts in real time. The news team has to balance telling a compelling story, based on the partial data they see early in the night, with the need to wait and see what the full results bring. It's a tricky balance, and it was great to watch them pull it together so well.
Q: What role do statistics play in presidential elections?
McCartan: Statistics is used all over the election world. News organizations run pre-election polls to give the public an idea of how the race is going, and that’s a heavily statistical process. The campaigns themselves also have polling and data operations to figure out which voters to target, where to run ads, where to send the candidates for visits, where to send mailers, which messages work and which don’t.
On election night itself, news organizations bring together academics, including statisticians and political experts, to look at all the data as it’s coming in and figure out what’s going on with the race. All that involves looking at data as it’s collected in real time and trying to weave it together into a story and do good statistics.
Q: What kinds of data do statisticians use to predict and calculate election results?
McCartan: Statisticians get to bring in different kinds of data sources to predict and work with election results. We bring in data from the census, demographics, where people live, past election results and public voter files. When you register to vote, data like where you’re registered and what party you’re registered with is taken and studied by statisticians. There’s also data on early ballot returns that we can use. And of course, there’s public polling where people are sampled randomly and asked their views on political issues. Statisticians integrate all that data to try and get a better picture about what’s going on with the election.
Q: How do major news outlets project winners of elections even though only a portion of precincts have reported their vote tallies?
McCartan: The results you are getting aren’t representative of the country as a whole. We get states like Indiana before California and make up the difference by comparing previous election results. Also, these days, a lot of people are voting early, and we know going into Election Day how many registered Republicans and Democrats have voted in many states in the country.
For very close races, you can’t really know until all the results are in, and so what news organizations will do is wait until someone is mathematically eliminated. This is what we saw in 2020 in Pennsylvania. News organizations waited until Philadelphia had enough votes to make a fully statistical race call.
At the end of the day, there’s always a human looking at all the data and making a call. Humans aren’t always perfect, but the track record of these news organizations making race calls is really good. It’s a great reminder that data can do amazing things and humans can do amazing things with data, but at the end of the day, you have to take everything and have someone who is going to make the call and be responsible for it.
Q: What’s so exciting about being a statistician during election season?
McCartan: I started studying statistics because I thought it was a great tool to understand the real world. For me, there is nothing more exciting about election season than how this data can affect what is going to happen in this country for the next four years. To be able to take the things I’ve learned and teach students about them every day and put them into practice is super exciting.
Q: What is your area of research?
McCartan: Broadly, I study computational social science, or the intersection of statistics and social science. Social science data is messy: people don’t respond, you have missing data, and all the relationships you can think of are all hidden and complex. There are variables you don’t observe and things that are hard to measure. So, that creates a lot of challenges for figuring out what’s really going on in the real world, but it also creates a lot of opportunities for research.
As a statistician, we’re asking how we can work with this messy data. How can we deal with all these problems? I like to investigate specific problems that we can try to solve, and how can we come up with new tools, methods, and statistics to solve them.
Q: What are some early lessons from your first semester at Penn State?
McCartan: Suddenly jumping from being a Ph.D. student to a professor is a big change. It’s been a fun challenge to learn how to integrate time spent doing research, teaching and designing classes, and interacting with students. It’s been super rewarding. The Penn State students have been great; I’ve really enjoyed our statistics majors and other students I’ve worked with in and out of class.