AAAS 2013: The science of politics

Photo via Thinkstock.

Photo via Thinkstock.

Until yes­terday, I hadn’t thought too much about the term “polit­ical sci­ence.” I prob­ably first heard it in high school or col­lege, when I accepted it as an item of poten­tial aca­d­emic study that I would not pursue and went on with my life (I con­sider pol­i­tics to be the single most abstract and frus­trating ele­ment of being a human). To me, polit­ical sci­ence wasn’t quite pol­i­tics (although I assumed it to be a poten­tial route toward a career in that field), and it wasn’t quite sci­ence. In fact, I never really stopped to think about why that word was even tagged onto the end of the term in the first place — until yesterday.

In a AAAS annual meeting ses­sion orga­nized by David Lazer, pro­fessor of polit­ical sci­ence and com­puter and infor­ma­tion sci­ence at North­eastern, five researchers blew my notion of polit­ical sci­ence out of the water. While the ses­sion was titled “The Sci­ence of Pol­i­tics,” which puts a dif­ferent kind of spin on things, I sus­pect the five speakers would all agree that polit­ical sci­ence needs to own the sci­ence space. And indeed, they argued that with new methods and tech­nolo­gies at its dis­posal, polit­ical sci­ence is finally becoming the sci­ence of pol­i­tics. Below are a couple exam­ples from three of the speakers.

Donald Green of Yale Uni­ver­sity kicked things off by demon­strating the dif­fer­ence between obser­va­tional studies — the tra­di­tional method used in polit­ical sci­ence — and exper­i­mental studies. The latter more closely resemble ran­dom­ized clin­ical trials of new drugs or ther­a­pies, but instead of testing out­comes of health inter­ven­tions, they look at polit­ical inter­ven­tions. They allow us, Green said, to get “an objec­tive sci­en­tific under­standing of polit­ical processes.”

Ques­tions like “what impact do direct mail­ings have on voter turnout?” (answer: very little) and “do our social con­nec­tions influ­ence our voting ten­den­cies?” (answer: yes) can be tested through care­fully designed field exper­i­ments cou­pled with pub­licly avail­able voting records. The obser­va­tional approach would gain infor­ma­tion about “exper­i­mental” con­di­tions by asking people spe­cific ques­tions after the fact (ie., did you receive any direct mail encour­aging you to vote?) while the field exper­i­ment would ran­domly sep­a­rate people into con­trolled groups. One group would receive direct mail, the other would not. So, instead of having to rely on inher­ently unre­li­able mem­o­ries, polit­ical sci­en­tists can do con­trolled exper­i­ments that yield much higher quality data.

Brown Uni­ver­sity pro­fessor Rose McDer­mott’s talk on the genetics of pol­i­tics had me a little ner­vous at the outset but bouncing with excite­ment by the end. It may not be all that pro­found to say some­thing like “con­ser­v­a­tives are more risk averse than lib­erals,” but to back that kind of state­ment up with phys­i­o­log­ical and genetic data is pretty amazing. McDer­mott star­tled us all by showing a ten-​​foot tall pic­ture of a taran­tula crawling toward someone’s eye­ball. Turns out con­ser­v­a­tives and lib­erals respond very dif­fer­ently to this kind of image (by the way, she was quick to point out that her studies do not look at dif­fer­ences between Amer­ican Democ­rats and Repub­li­cans, but rather indi­vid­uals’ con­cep­tual ori­en­ta­tions across a spec­trum of ideologies).

Con­ser­v­a­tives tend to focus their eyes on the taran­tula while lib­erals spend more time looking at the eye­ball, she said. Con­ser­v­a­tives’ heart rates increase faster and they sweat more when they’re faced with images like the one she showed us. Other researchers have recently shown that they can pre­dict with 85% accu­racy a person’s polit­ical lean­ings based solely on how their brains respond during a risk-​​taking task having nothing to do with pol­i­tics at all.

The var­ious ways we respond to fright­ening infor­ma­tion are her­i­table traits, McDer­mott said. And if con­ser­v­a­tives tend to become more fright­ened, it’s not very hard to under­stand why they might be more likely to sup­port things like mil­i­tary spending and the death penalty: those actions will help pro­tect us from “evil.” It’s not that con­ser­v­a­tives are more risk averse, as I said above, but rather that more risk averse people tend to be conservatives–an impor­tant dis­tinc­tion. Like­wise, there is no “con­ser­v­a­tive” gene or “lib­eral” gene as the media might like to pre­tend. “Its not that simple. It never will be that simple,” said McDer­mott. “These things are mul­ti­fac­to­rial.” Instead of a single gene, it’s a “con­stel­la­tion of path­ways” inter­acting with our envi­ron­mental expe­ri­ences, she explained.

Lazer’s talk focused on the ways net­works and big data can inform our ability to answer large scale ques­tions about the spread of polit­ical ide­ology through pop­u­la­tions. He specif­i­cally spoke about three studies exploring the impact of per­sua­sion, peer influ­ence, and col­lec­tive action. The per­sua­sion study was an example of the field exper­i­ments Green spoke of, with a ran­domly assigned group being privy to an online town hall event in which par­tic­i­pants could directly interact with their polit­ical rep­re­sen­ta­tive. The par­tic­i­pants in the meet­ings were more likely to vote for the member of con­gress than those who didn’t par­tic­i­pate. “We found sub­stan­tial changes in the direc­tions of the member’s opin­ions,” Lazer added, showing that the town hall meeting not only influ­enced the par­tic­i­pants’ feel­ings toward the can­di­date, but also the candidate’s feel­ings towards his con­stituents’ desires.

His peer influ­ence study looked at the changes in polit­ical view­points of freshman entering col­lege dor­mi­to­ries, in which they would reside for the next four years of col­lege. They asked the stu­dents to list how they were con­nected to all of the other people in the dorm and then looked at cor­re­la­tions between their var­ious polit­ical view­points. It turned out that esteem and close friend­ship influ­enced stu­dents’ voting prac­tices, but who they spent the most time with, spoke about pol­i­tics with, and even dis­liked, did not influ­ence their polit­ical leanings.

Finally, his project Money Bombs, a visu­al­iza­tion of how Bosto­nians gave to polit­i­cans during the 2012 elec­tion revealed the under­lying net­works that mobi­lize money, he said, showing how col­lec­tive action influ­ences polit­ical behaviors.

Susan Hyde of Yale, and Daniel Dier­meier of North­western also spoke about ways their work is bringing pol­i­tics into a more sci­en­tif­i­cally rig­orous frame. Dier­meier demon­strated ways in which polit­ical phe­nomena can be math­e­mat­i­cally mod­eled and Hyde spoke about her work using field exper­i­ments to explain the anomaly that fraud­u­lent democ­ra­cies invari­ably invite for­eign observers to mon­itor their elec­tions, despite the obvious cost they incur by doing so.