Survey Response Bias and the Determinants of Substance Use
2008/10/28
研討會日期 : 2008-10-28
時間 : 14:30
主講人 : 余士迪教授
地點 : C棟103室
演講者簡介 : 余士迪教授為美國羅徹斯特大學(University of Rochester)經濟學博士。目前為清華大學計量財務金融系教授。其主
要研究領域為個體經濟計量學、財務計量及計量經濟學。
演講摘要 : We investigate the impact of survey response bias on the analysis of the determinants of substance use. We demonstrate that the prevailing approaches in the literature produce biased and misleading estimates for the determinants of substance use because survey response bias is ignored. We present an alternative approach which can correct for survey response bias and produce consistent estimates for the determinants of substance use. Our approach can simultaneously identify: (i) the determinants as well as the magnitude of survey response bias, and (ii) the determinants of substance use in a way that is free of survey response bias. To illustrate our approach, we provide an empirical application with four different types of substance: alcohol, cocaine, marijuana, and cigarette. We find that the presence of survey response bias in substance use data is not negligible. The bias is substantial especially for illicit substances. Our approach is applicable to other types of self-reported survey data where response bias is a concern.
