Data Quality in Online Surveys: The Case of Survey Trolls

Thu, September 26, 2019 11:00 AM at 104 S. Kedzie Hall

Professor Hillygus has published widely on the topics of American political behavior, campaigns and elections, survey methods, public opinion, and information technology and politics. She is co-author of The Persuadable Voter: Wedge Issues in Political Campaigns (Princeton University Press, 2008) and The Hard Count: The Social and Political Challenges of the 2000 Census (Russell Sage Foundation, 2006). From 2003-2009, she taught at Harvard University, where she was the Frederick S. Danziger Associate Professor of Government and founding director of the Program on Survey Research.

Here is an abstract for her talk:

Online surveys now represent a large segment of the academic research on public attitudes and behaviors.  There is widespread recognition that online surveys can be plagued by lower levels of data quality than traditional survey modes, especially low levels of respondent attention that can introduce noise into survey estimates. We evaluate a variety of metrics of data quality and introduce another source of measurement error: survey trolling, or mischievous responding, whereby respondents give insincere answers in an effort to be provocative or humorous.  We offer a technique for identifying mischievous responding and evaluate the implications for survey research. Our analysis examines the incidence and implications of survey trolling across multiple surveys, with a particular focus on evaluating the extent to which it biases estimates of political misinformation in the American public.