Search Results

Source: Johns Hopkins University
Resulting in 1 citation.
1. Bush, Ryan
Essays on Recovering Latent Characteristics Using Nonparametric and Semiparametric Methods
Ph.D. Dissertation, Department of Economics, Johns Hopkins University, 2016
Cohort(s): NLSY97
Publisher: Johns Hopkins University
Keyword(s): Drug Use; Modeling; Self-Reporting

Permission to reprint the abstract has not been received from the publisher.

This dissertation explores applying nonparametric and semiparametric methods to recover latent characteristics in various settings. The second chapter employs panel data to recover latent behavior related to the health and drug use of survey respondents. The misreporting problem of drug use in self-reported surveys can severely affect the validity of estimation results in empirical work. In this paper we use an eigen-decomposition method to nonparametrically estimate the misclassification errors under various assumptions and settings. We use the longitudinal data of NLSY97 and focus on the years from 2005 to 2009, when the cohort is aged in their mid-20s. We find that the overall proportion of participants who actually use marijuana is higher than the reported proportion. Moreover, participants' inclination to misreport their drug use status is related to their current and previous actual drug use as well as their "habit" for misreporting in surveys. In general, males are more likely to underreport their drug use than females.
Bibliography Citation
Bush, Ryan. Essays on Recovering Latent Characteristics Using Nonparametric and Semiparametric Methods. Ph.D. Dissertation, Department of Economics, Johns Hopkins University, 2016.