Jon Krosnick
Jon Alexander Krosnick is a professor of Political Science, Communication, , and , and director of the at Stanford University. Additionally, he is the Frederic O. Glover Professor in Humanities and Social Sciences. Krosnick has served as a consultant for government agencies, universities, and businesses, has testified as an expert in court proceedings, and has been an on-air television commentator on election night.
Some of Krosnick's work focuses on the design and methodology of questionnaires and surveys, and he has taught extensively around the world on that topic. He has also studied the psychology of attitudes and researched how voters make up their minds and how campaigns influence them. He has conducted research on American attitudes toward global warming, how negativity in campaigns affects turnout, ballot order effects and other topics.
Krosnick was a principal investigator leading the American National Election Studies from 2005 to 2009, along with Arthur Lupia of the University of Michigan.
Personal life
Krosnick's mother was an educator and opera-singer, and his father was a physician who was a diabetes specialist, professor, expert witness, and researcher.. Jon Krosnick has a sister, Jody Arlyn, who is a surgeon. He became interested in music at an early age, starting to learn how to play the piano at age 6 and going to the National Music Camp at Interlochen at age 9, where he first encountered jazz drummer Peter Erskine. Erskine would later be a major musical influence for him and a personal friend. Krosnick continued playing percussion instruments from elementary school on, performing as a soloist with various orchestras, including the Philadelphia Orchestra, as well as playing jazz with many ensembles, including the electric jazz band, and the Latin jazz group, The .Krosnick went to the Lawrenceville School in Lawrenceville, New Jersey and graduated in 1976. He later graduated magna cum laude from Harvard University in 1980 with a B.A. in Psychology. He then received both an M.A. in 1983 and a PhD in Social Psychology in 1986 from the University of Michigan, Ann Arbor. On June 1, 1986, Krosnick married Catherine Ann Heaney. He joined the departments of psychology and political science at Ohio State University, Columbus, as a lecturer in 1985, became an assistant professor in 1986, and was promoted to associate professor in 1991. He became a full professor and was a member of the Ohio State University political psychology program and co-directed the OSU Summer Institute in Political Psychology. In 2004, Krosnick became a professor at Stanford, where his wife is also a faculty member. The couple have a daughter, , who graduated from Stanford as an undergrad, got a PhD from Columbia University's Mailman School of Public Health, was a post-doc at the University of California Berkeley School of Public Health, and in 2023 joined the faculty of the Herbert Wertheim School of Public Health and Human Longevity Science at the University of California, San Diego. Jon and Catherine live in Portola Valley, next to Stanford. Like his parents’ home, Krosnick's under-construction home was seriously damaged by a fire in 2016.
Work in survey methodology
Questionnaire Design
The largest area of Krosnick's work in survey methodology lies in questionnaire design. In his article, Optimizing Survey Questionnaire Design in Political Science: Insights from Psychology, which he co-authored with Josh Pasek, he "shows the general principles of good questionnaire design, desirable choices to make when designing new questions, biases in some question formats and ways to avoid them, and strategies for reporting survey results."When deciding whether to use ratings or rankings as response option for a question, Krosnick and Alwin found, in a study with thirteen rating scales, that 42 percent of individuals evaluated nine or more of the objects identically. Krosnick argues that such non‐differentiation is most likely to occur under the conditions that foster satisficing. Krosnick has also found that although ranking questions take more time, rankings acquire responses that are less distorted by satisficing and are more reliable and valid than ratings.
In a series of separate studies that were designed to test the effects of having a "don't know" response option, Krosnick and his colleagues found that candidate preferences predict actual votes better when researchers discourage “don't know” responses. This is one of the reasons that Krosnick argues that discouraging “don't know” responses collects more valid data than does encouraging such responses. In his article, Optimizing Survey Questionnaire Design in Political Science: Insights from Psychology, Krosnick also says that respondents who truly are completely unfamiliar with the topic of a question will say so when probed, and that answer can be accepted at that time, thus avoiding collecting measurements of non‐existent "opinions." Thus, because many people who initially say "don't know" do indeed have a substantive opinion, researchers are best served by discouraging these responses in surveys.
Krosnick also argues for a phenomenon known as response order effects, which is another form of satisficing in which a respondent chooses the first plausible response option he/she considers. More specifically, there are two types of effects that can be seen. One is called a primacy effect, which is the tendency to choose the items at the beginning of a list of options, and the other is called a recency effect, which refers to the tendency to choose items at the end of a list of options. To decrease response order effects, Krosnick suggests that researchers use what are called seemingly open‐ended questions.
Survey Response Rates
Krosnick and colleagues compared Internet surveys, telephone surveys, and face-to-face surveys of probability samples and found that people provide socially desirable answers more often in telephone surveys than in the other two cases. They found that face-to-face-survey respondents respond more accurately than telephone respondents. Since face-to-face interviews are costly, Krosnick conducted a study providing computers and an Internet connection to a set of randomly sampled people, and inviting them to answer survey questions online over a year. This method is known to produce samples, after subtracting those who refused to participate, reflecting population counts of various groups proportionately.Krosnick's research has also focused on the effects of response rates on the accuracy of survey results, and his work indicates that women were overrepresented in RDD surveys relative to the population. Additionally, in the same analysis, Krosnick found that an RDD survey sample included more high-income respondents and fewer low-income respondents than the population.
Opt-in surveys
Krosnick has published studies questioning the use of Internet opt-in surveys. Such surveys do not result in a random sample because participants are a self-selected group. Along with David Yeager, Krosnick concluded such surveys produced results varying from traditional surveys even after they were statistically adjusted to cancel effects from their non-random nature. Another study found such studies could not be used to compare how a group's behavior or attitude changed over time, or how their responses to different issues related to one another. Krosnick and Yeager used the same procedure to weight the raw data demographically in order for their surveys to be equally representative in terms of gender, age, race, etc. They then calculated the average error for the surveys on 13 additional measures of "secondary demographics" and other non-demographic factors. The responses of opt-in Internet surveys differed from those in traditional surveys. Krosnick reached similar conclusions using two surveys collected for the U.S. Census Bureau, with one being a traditional poll and the other an Internet opt-in one.In another study, Krosnick and a collaborator, LinChiat Chang, compared probability samples interviewed by telephone and via the Internet to an opt-in Internet sample. This study found the latter to be less representative of the population in terms of demographics and to over-represent people with high interest in the topic of the survey. To further explore the generalizability of these findings, Krosnick, along with David Yeager and other colleagues, collected data on a variety of topics via an RDD telephone survey, an Internet survey of a probability sample, and Internet surveys of seven non-probability opt-in samples of American adults. The estimates from each survey were then compared to benchmarks from official government records or high-quality federal surveys with very high response rates. Using a sample of 1,000 participants, the results showed that all of the non-probability sample Internet surveys were significantly less accurate than the probability sample Internet survey in terms of primary demographics, and all but one of the non-probability sample Internet surveys were significantly less accurate than the telephone survey.
Work in political psychology
Issue Publics studies
Krosnick has done extensive research on attitude strength, with a particular focus on attitude importance. He has shown that the U.S. adult population is divided into small groups, each of which is passionate about a particular policy issue. That passion translates into effortful information gathering and information processing on the topic. And people passionate about a policy issue form stable preferences on that issue and use those preferences when making vote choices.Ballot order studies
Krosnick and a colleague, analyzing data from an Ohio election, concluded the candidate whose name is listed first on a ballot received roughly 2% votes more in half of the races they studied. The effect was stronger in races where the voters had no clear a priori choice. While this effect has been known for more than a century, the study produced evidence. His testimony to this effect led a court to invalidate an election in Compton, California. The effect has also appeared in other areas.Krosnick and others conducted a study of the 2000 U.S. Presidential elections in Ohio, California, and North Dakota and found that candidates gained votes when listed first on the ballot as opposed to when listed later. For elections, Krosnick hypothesized the effect may be from voters, feeling compelled to cast a vote, choosing the first choice on the list. He believes George W. Bush benefited from this effect in the 2000 presidential election in Florida, and that the exit poll of the 2004 U.S. Presidential election was skewed toward the Democratic candidate, John Kerry, because he was listed first on the questionnaire.