Caroline Soler’s Deep Dive Into Voter Data

The senior has been crunching the numbers as an analyst on a major election survey

Even as a little kid, Caroline Soler, A25, was fascinated by election night. “I loved watching those guys look at the numbers and tell you, ‘If this county comes in at 52%, then this state’s going to go blue,’” she said. 

While she grew to care about political issues as she got older, she came to Tufts expecting to major in math, which she enjoyed because of its formulas and its certainty. Only later did she see how she could use the precision of math to help people make sense of the seeming chaos of politics. 

Now a political science major with a math minor, she was hired in June as a research associate for the Cooperative Election Study (CES), which at 60,000 participants is the largest academic survey focused on American elections. She’s been gathering and analyzing data under the guidance of the study’s co-principal investigator, Brian Schaffner, the Newhouse Professor of Civic Studies in the political science department and Tisch College. She’s also co-authored essays about hot topics like undecided voters and poll weighting, and put together snappy graphs that reveal some unexpected things about public opinion. 

You came to Tufts as a math major. What changed?

I’ve been taking courses with Professor Schaffner since first semester freshman year, when I took Intro to Civic Studies. When I took his class on Poli Sci Research Methods, that’s when I realized this sector—this combination of math and humanities and politics—exists. It’s the perfect mix of these things of I’ve always wanted to do. It makes my heart feel warm, and it makes my brain fire up. 

What is the Cooperative Election Study and why it is valuable?

There aren’t a lot of 60,000-people surveys that analyze this many kinds of opinions and demographics. Because the survey has been ongoing since 2006, we have data on demographic groups going back nearly 20 years. In election years, it asks the same people some of the same questions before and after the election, so we can see how the opinions of specific individuals are changing. It has validated voter info, which means it matches respondents to voter files to show us if they voted or not. That’s one reason we were able to say, based on data from 2020, that most voters who are still undecided in October end up staying home and not voting at all

Can the average person see the data from the study?

I’ve worked on the Shiny apps for our website, interactive tools that display the CES data in ways that you can access without being a coder. We added some filters, so that if you want to look at gender, you could look at women just between the ages of 18 and 27, for example. Or you could look at Black people who are Republicans. You can see how their opinions have changed or their vote has changed. 

It adds transparency. There’s a lot of bias in media right now, and many people point out trends that maybe aren’t happening. For example, people were talking about Black voters abandoning the Democrats, which we didn’t see when we looked at our data. So if you see something in the news, you can run to the app and see what is actually happening using a huge survey you trust.

You put a lot of energy into the graphs you make. Why are graphs important?

People who aren’t super into politics might not spend time reading an article about numbers, but if they’re on Twitter and they see a pretty graph, they might stop and look. I’ve been working on having really good titles and axis labels so that it makes total sense for someone who has no other context of what’s going on.

A graph compares state abortion laws to constituent preferences

What is a favorite graph that you’ve made recently?

I think the graph that shows the differences between state abortion policies and constituent preferences is really powerful, because you can see almost everyone wants abortion to be legal to about 20 weeks, which is slightly before viability. Maryland, for example, is one of the most Democratic states in comparison to Kentucky, yet people in both states want the same number of weeks. I wanted to show how this compares to the actual limits in different states, some of which have no cut-off dates and some of which ban abortion entirely. Specifically on abortion, politicians don’t really listen to their constituents’ preferences, because almost all people support abortion to some extent. I think this graph does a good job of explaining that. 

Did you ever picture yourself doing research like this in college?

Absolutely not. I don’t think I understood what humanities research was. But I’m so grateful I’m doing it. It’s changing how I think, making me a harder worker, and I’m learning so many new skills.

Back to Top