November 10, 2010
Political Science Class Conducts Election Day Exit Poll and Analysis
Students Andrew Hunt and Anthony Cuoco
at their polling stationExit polls. Ever wonder how they work? Professor Brian Schaffner (political science) is an expert on survey research, a regular contributor to Pollster.com, and has made it his business to lead 17 students in his class, Political Polling and Survey Research, through the process of developing, conducting and analyzing an exit poll of voters during the 2010 election. As it turned out, none of the major polling organizations chose to do exit polls in Massachusetts this year, so Schaffner’s class conducted the only one in the state. When the media figured that out, they were all ears to hear the results, which turned out to be quite accurate. But before getting into analysis, a look at process and methodology helps set the scene for the classroom.
In the weeks leading up to the election, students developed an understanding of how surveys allow for making inferences about public opinion. While learning how to constructively critique existing research, they determined what they wanted to find out through the exit poll. (Each student is developing their question into a lengthy research paper, incorporating data analysis with existing literature.) Then they developed skills for writing good survey questions and how to properly sample and interview respondents.
“I’m particularly interested in female voters,” said Sarah M. Schultz ’11 (political science and history). “I’d like to know the relationship between occupation and political choices.”
“My interest lies in how people view healthcare reform and then tying it to a preferred news source,” added Dan Schulke ’11 (political science).
Anthony Cuoco ’13 (political science) said, “Knowing more about people’s beliefs on global warming and how that might impact climate change theory and testing is important to me.”
“The students are so invested,” says Schaffner. “They chose to conduct a statewide poll, over a local one. That requires a lot more work, but they’re energetic! Our goal was to determine how different demographic groups voted, which issues were most important to voters, and what we could learn about voting behavior and public opinion in general.”
The methodology approximated the approach networks take for their exit polling. First, to ensure representation of different geographic areas and racial/ethnic groups, students selected a probability sample of towns and cities within the state using a method called stratification. They randomly selected two towns in each of four regions across the state, plus two more communities to ensure representation of Hispanics, African Americans and rural voters. Precincts were randomly selected and voters were randomly sampled, using a formula determined by previous turnout.
Everyone in the class participated in the exit poll for the entire day. They’d informed all the precincts that they were coming, learned the rules about set ups and how to factor in rejection, and were armed with helpful hints on best practices. Instructed to arrive early and to work on their tally sheets throughout the day, the teams of two (including 3 volunteers) dispersed with their clipboards, boxed exit surveys, t-shirts (for identification purposes), pens, and tally sheets to their polling stations.
Schaffner spent the day driving around the state, making sure that things ran smoothly. When the polls closed, he gathered each team’s tally results and did some quick number crunching to determine initial results for the media.
The exit poll correctly estimated the vote for governor, attorney general, secretary of state and Proposition 1 within a margin of error. While men narrowly favored Republican Charlie Baker, Democratic incumbent Deval Patrick had a 25% edge among women, which allowed him to win re-election. Baker won 97% of the vote among Republicans, but just 53% among independents. Patrick won 86% of the Democratic vote, and Democrats comprised 42% of the electorate.
While 41% of the Massachusetts electorate either strongly or somewhat supported the Tea Party, 45% strongly opposed the movement. 82% of voters who supported Scott Brown in January voted for Baker; however, 12% of voters reported that they didn’t vote in January and 60% of that group supported Patrick.
The economy was the most important issue to voters, followed by healthcare, education, and taxes. 37% of Massachusetts voters thought the economy had improved during the last year, while 14% said it had worsened and 50% thought it was about the same. 36% thought the economy would improve in the coming year while 33% thought it would get worse. Of the former group, 93% voted for Patrick; of the latter, 75% voted for Baker.
Opinion on the new healthcare reform law was divided: 42% approved while 41% disapproved and 17% were not sure.
In a debriefing, the class shared experiences. “It was a really long day,” said one student. “Rejection is not fun,” said another. “It’s amazing how many ways people can find to avoid taking a survey!” “Some people wouldn’t believe that it was confidential.” “Some wouldn’t take it because it wasn’t in Spanish.” “People complained about the length, but no one stopped taking it because of the length.” Several students mentioned being greeted by UMass alumni who were excited to see them conducting the exit poll.
One student noted, “It was interesting that many voters said they support the Tea Party but then voted for Patrick and support Obama. There’s definitely a difference between supporting a movement as opposed to a candidate.” Another added, “The experience made me think a lot more about biases—demography seems to determine the issues. I think politics would be a lot different if younger people actually turned out to vote.” Almost everyone mentioned the paucity of young voters at their polling locations.
Now the students are learning how to conduct statistical analysis. Looking for more in-depth answers to the research questions they proposed for their papers, they are using statistics to make inferences about relationships between variables. Before submitting their research papers, each student will make a presentation of their topic, followed by a question/answer session with classmates.
“I’ve never seen students so excited about doing research,” notes Schaffner. “I think that’s because they’ve been intimately involved in every detail of the process while studying something inherently interesting. I definitely hope to offer the class again in 2012.”