Latent Class Analysis of Consumer Expenditure Reports

Previous research by Tucker et al. (2007), working with the Consumer Expenditure Interview Survey (CE), explores the efficacy of measurement error indicators such as: interview length, extent and type of records used, the monthly patterns of reporting, and income question missing in a latent construct. Later research by Tucker, Meekins, and Biemer (2008) extend this latent class model to include indicators of response behavior across multiple interviews in a panel. This research develops a number of plausible models which possess the qualities of reliability and validity, in that they appear to accurately capture measurement error, but prove unable to explain a large amount of the variance associated with expenditure reports. This work extends past research by including the relatively recently recorded indicators from the Contact History Instrument added to the CE since 2005. In addition, this work examines measurement error by mode of collection (also a recently collected item).