An Investigation of the Effect of Ignoring Extreme Survey Response Style

• In Event: Marriott Roundtable Session 1
  In Roundtable Session: Assessment Methodologies in Higher Education

Abstract

Response style effect is a well-known limitation of surveys. By applying a generalized item response theory (IRT) model on extreme response style (ERS) to the Global Perspective Inventory data in 2014 National Survey of Student Engagement survey, this study intends to provide estimates of respondents’ ERS tendency and purified estimated construct of interest. Then, we compare respondent ERS tendency across groups for three demographic factors: gender, institution Carnegie classification, and major category. Further, in order to demonstrate the issue of ignoring ERS in survey data interpretation, we compare the unpurified construct of interest and the purified construct of interest across demographic groups. Preliminary results suggested that there was significant difference in ERS tendency across major categories.

Authors

• Xiaolin Wang, The University of Kansas

• Amy Ribera, Indiana University

• Justin Paulsen, Indiana University - Bloomington