Think you can correctly predict the results of November's Senate elections? You could be the winner of $2,500.
The political science department is holding a contest to see which participant can predict the most results of the 2014 U.S. Senate elections—with the winner receiving the monetary award. In order to enter, students must list the name, party and the vote percentage received by the winner in each of the Senate races. Scott de Marchi, associate chair and professor of political science and one of the contest's organizers, noted that the aim of the contest is to get undergraduates students interested in the connections between data science and social science.
"A lot of undergraduates are not aware that social sciences are a pretty natural hub for studying data science and human decision-making," de Marchi said. “In this case, we have a pretty great election coming up, where the Senate is up for grabs between the two main parties, and recently there’s been a lot of data work about who is going to win. Every main journalistic outlet is trying to do data science, so it was a good opportunity for undergraduates to try it out and see what the process looks like and how well they can do."
Of the 100 seats in the U.S. Senate, 36 are open in the November elections. With 21 of those seats currently held by Democrats and 15 currently held by Republicans, many political analysts predict that the GOP will take control of the Senate this year by gaining at least five seats.
“A good chunk of the Democratic seats are in states that President [Barack] Obama lost when he ran for re-election, and only one of the Republican seats is in a state that Obama won, so it’s a very disadvantageous contest for the Democrats, in addition to the problems of having a president in his sixth year who is not very popular," said David Rohde, Ernestine Friedl professor of political science. "All of these things work to the disadvantage of the Democrats and puts their Senate majority in jeopardy."
Senior Eric Wu noted that the contest was a great demonstration of the interdisciplinary applications of statistics, politics, and also computer science.
“I have a passion for politics, and I’m also deeply interested in how to use and analyze data and combine those with computer science, so it seemed like the perfect intersection of all the interests that I have," Wu said.
Wu explained that the statistical component of project allowed his team to write programs and apply computer science in a real world application.
"I think that’s one thing that we really don’t learn enough about in our curriculum and academia, so this is really a great opportunity for us to get our hands dirty and see if we can really apply the knowledge we’ve learned in class into a real world problem," he said. "That’s one of the very exciting parts, that something I just learned in class I can apply in a very real world way and see the immediate results of it.”
De Marchi also added that he hopes students will deal with data and run into problems as they work on the contest.
“The joy of modeling is that you can be wrong. Being wrong is just one of the more powerful tools for advancing science," he said. "Frame the problem in the way that you can get the wrong answer, and good things happen.”
To students considering participating in the contest, Rohde suggests looking at micro-data in the states that will have very close races.
De Marchi, however, recommends creating a good prediction model and avoiding over-fitting that model to small little differences.
“Take a shot at it, and the way to get better is to do it again, keep doing it again and keep in mind general lessons you’ve learned,” he explained.
The $2,500 prize, funded by departmental money, is the incentive to get students interested in the contest, but De Marchi hopes that undergraduates will be able to create solid, if wrong, modeling approaches and keep track of what they’ve learned.
"We might donate the money if we win," Wu said. "This is really just a fun project for us to do, so we’re really interested in just using that money to help people.”
Get The Chronicle straight to your inbox
Signup for our weekly newsletter. Cancel at any time.