Click on photo(s) to enlarge.

Jon Krosnick, The Frederic O. Glover Professor in Humanities and Associate Director of IRiSS, with his research team

Mark Tuschman

Why were the polls so wrong in the January 2008 New Hampshire Democratic primary? Jon Krosnick, professor of communication, helped solve the mystery.

In the past, New Hampshire primary ballots had a different candidate order in each precinct. But in 2008, the candidates were listed alphabetically on every ballot. According to Krosnick's research, that gave Hillary Clinton at least a 2 percent advantage. Pollsters negated this advantage by randomly changing the order of the candidates for each person they interviewed, and said Obama would win easily.

States should rotate the order of names on ballots so that an equal number of voters see each candidate's name listed first, concluded Krosnick, the Frederic O. Glover Professor in Humanities and Social Sciences. He hopes that publicizing his research findings will encorage states to reform their ballot process.

In an era where America's political races have become closely contested, Krosnick's work has important implications for the validity of election results. His study of name order effects in elections in Ohio, California, North Dakota, and elsewhere was done with a team of Stanford undergraduate and graduate students who conducted the most precise research to-date. Various statistical methods had been used to test name-order effects in past elections, but no studies had directly compared methods to one another to see whether they yielded different results, which Krosnick's team did.

"My research with Jon has been exciting because it made me realize how statistical tools can be used to spot policy problems like ballot design," said undergraduate Daniel Blocksom'08. The team analyzed election returns from the 2004 presidential race in all precincts in Ohio, where name order is rotated across precincts. Having data on more than ten thousand precincts provided them with great statistical power, which they used to determine that all presidential candidates received significantly more votes when listed first. They also discovered name-order effects were most pronounced in precincts with the least educated voters and in precincts that voted on touch screen machines.

Krosnick's innovative methods were supported by The Institute for Research in the Social Sciences (IRiSS), which facilitates the formation of multidisciplinary research teams and provides them with the resources to tackle major social issues. Working closely with faculty mentors, students at the Institute work on problems of democratic stability, poverty, population and disease transmission, and the impact of philanthropy and entrepreneurship. Through its efforts, IRiSS strives to increase the impact of social science research on public policy, business innovation, and social entrepreneurship.