Data department targets voters, helped communicate Obama's 2012 campaign message| Nov 7, 2013
Data processing played a pivotal role in the Obama campaign’s ability to target and persuade potential voters in the 2012 election, former campaign data director Ethan Roeder argued in a lecture on Thursday evening.
Roeder, who served as data director for both the 2008 and 2012 Obama presidential campaigns, explained that his department pioneered the use of vast data collection operations to aggregate open-source information about voters’ preferences so that they could be approached and persuaded by volunteers.
Using polling data, consumer data and open voter files the data department created models to predict how likely voters were to vote for Obama or be persuaded to support or volunteer for the campaign.
“It’s stunning to me how relatively simple data can give us such an edge in predicting where people are at in terms of voting,” he said.
Open voter files in the United States contain simple information like voters’ names, addresses, ages, genders, voting histories, party affiliations, and racial backgrounds. Roeder said that his data team, comprised of 107 staff members across 17 states, worked to compile that data with information gathered by volunteers contacting voters.
Such models, while useful for helping to “take the temperature of where the electorate is,” are just tools for conveying the campaign’s political message, Roeder explained.
“A model is not something that dictates strategy because there are still some decisions you have to make about how you use that tool,” he said.
The Obama team’s campaign strategy was all about going from wholesale to retail politics for the sake of voter persuasion, he explained. Data helped show the Obama campaign which voters to target, but Roeder noted that the actual process of persuading those voters relied heavily on human interaction and engaging each voter in a political conversation to understand and hear stories of why they vote the way they do.
“Because we treated stories with the same importance as the numbers and other bits of info we received, we were able to better communicate with voters,” Roeder said.
He denied the validity of analysis offered by many media sources in the aftermath of the 2012 election, which praised the Obama campaign’s data and analytics departments as responsible for the victory.
“Tech didn’t win the campaign; tech and tools helped to communicate the message that we had a candidate worthy of winning,” Roeder said.
Contrary to media reports in the election’s aftermath, the Obama campaign never collected data on who was watching pornography online and how people were likely to vote based on their internet browsing histories, Roeder added, laughing.
“Even a billion dollar campaign has limited resources,” he said. “Sometimes the simplest approach works best.”
The lecture, titled “Obama, Data, and Porn: What really went down on the 2012 campaign,” was hosted by Princeton’s Center for Information Technology Policy in 222 Bowen Hall.