Polling Society accurately predicts election
By Dan O'Brien
October 21, 2013
October 21, 2013
The student-run Emerson College Polling Society says it accurately predicted the outcome of the U.S. House of Representatives 5th District of Massachusetts primary election before voters went to the polls last Tuesday, October 15.
The Democratic winner in the heavily Democratic Congressional district is Katherine Clark, a state Senator from Melrose, with 32 percent of the vote, according to the Boston Globe. Four days before the election, the Emerson College Polling Society announced that Clark was the front-runner with 29 percent support in an automated poll of 471 likely Democratic voters administered October 2-8.
The poll also said Clark had a 10-point lead, which turned out to be exactly true when 5th District residents voted. Clark beat Peter Koutoujian, the Middlesex County Sheriff, 32 percent to 22 percent.
According to MassLive.com, the Emerson polling club shed light on an election in which there was “almost no public polling.”
“I’m very proud that ECPS predicted the election results with such accuracy,” said Siobhan Robinson ’14, president of the polling club. “It gives us credibility, which is so important as we are a student-run group.”
Spencer Kimball, faculty advisor to the group who is also scholar-in-residence in the Communication Studies Department, said he could not find another independent poll that examined the 5th District race before Election Day. However, there were three internal polls conducted by the campaigns of three Democratic candidates (Clark; State Senator Karen Spilka; and State Representative Carl Sciortino).
“[This] demonstrated to the students that there are jobs available for those with this skill set,” Kimball said.
“This race really taught me about the power of endorsements,” Robinson said. “Koutoujian was able to win in Revere and Winthrop thanks to Speaker [of the House of Representatives Robert] DeLeo endorsing him.”
Students were also able to see firsthand how a low turnout election necessitates the use of a “likely voter” model when polling as opposed to a “registered voter model,” which is often used when turnout will be higher.
The Emerson College Polling Society collects its data through Stratic Networks using an interactive voice response system, which eliminates a live human from injecting bias into questions. The students write the polls then collect the responses from Stratic Networks before running an analysis.