Good Item or Bad: Can Latent Class Analysis Tell? Examining the Effectiveness of the Latent Class Analysis Approach to Item Evaluation

• Activity Number: 278
• Type: Contributed
• Date/Time: Tuesday, August 8, 2006 : 10:30 AM to 12:20 PM
• Sponsor: Section on Survey Research Methods
• Author(s): Frauke Kreuter*+ and Roger Tourangeau and Ting Yan
• Companies: University of Maryland and University of Maryland and Institute for Social Research
• Keywords: latent class analysis ; measurement error ; social desirability bias ; mode effects ; questionnaire pretesting

Abstract

Latent class analysis (LCA) has been used to model measurement error, to identify flawed survey questions, and to compare mode effects in the survey context. Compared to other approaches, LCA has the advantages that it doesn't require true values ("a gold standard") or the stringent assumptions usually assumed in classical measurement error analysis. However, there has been no evaluation of this technique yet in terms of its accuracy and effectiveness for detecting bad questions. This paper seeks to determine whether LCA yields plausible results that agree with accepted procedures for assessing questionnaire items. Specifically, we will examine, in the presence of a gold standard, how well the results from LCA converge with those from a comparison of survey responses with a gold standard.