Skip Navigation

Produced by the Office of Marketing and Communications

Subscribe Now

‘Social Sensing’ Could Better Predict Elections, Other Trends, Researchers Find

By Maryland Today Staff

In the run-up to the 2016 U.S. presidential election, many polls predicted a win for Hillary Clinton, leaving wonks and pollsters alike scratching their heads at Donald Trump’s victory.

A perspective piece published today in Nature by a team including a University of Maryland researcher presents a new approach to forecasting elections and trends: “social sensing,” which relies on human interaction and social awareness to fill in gaps where surveys and other scientific analyses fall short.

In short, the researchers propose that asking participants who their friends are likely to vote for could provide more reliable results than asking who they themselves will vote for. That’s because among respondents, some may change their minds after being polled, while others may be embarrassed or afraid to report their true intentions, according to the paper for which Professor Frauke Kreuter of UMD’s Joint Program in Survey Methodology and of the Ludwig-Maximilians-University of Munich is a co-author.

While it’s widely assumed that cognitive bias clouds our assessment of the people around us, their research and that of others suggests that in fact, our estimations of what our friends and family believe are often accurate, said corresponding author Mirta Galesic M.S. ’05 of the Santa Fe Institute. Thus, researchers can gather highly accurate information about social trends and groups by asking about a person’s social circle rather than interrogating their own individual beliefs.

The paper’s co-authors hail from the Santa Fe Institute, the University of Southern California, Purdue University, the Sloan School of Management at MIT and New York University.

Read the full news release by UMD’s College of Behavioral and Social Sciences.

Maryland Today is produced by the Office of Marketing and Communications for the University of Maryland community on weekdays during the academic year, except for university holidays.