Are Self-Reported Smoking Information As Good As Clinical Measures? Evidence From a Data Driven Approach
34 Pages Posted: 16 May 2018
Date Written: May 3, 2018
Abstract
Self-reported smoking variables are commonly utilized by researchers, however, they are prone to be misreported. Clinical researchers tend to rely on some form of objective or clinician generated smoking measure, especially in randomized controlled trials. This paper uses the Lung Health Study (LHS), a randomized smoking cessation study, to analyze the discrepancies in self-reported and objective smoking measures, and provides some guidance regarding the level of misreporting. Utilizing large changes in Body-Mass-Index (BMI), carbon monoxide (CO) and cotinine (COT) levels after the smoking intervention, we create our won estimate on the level of misreporting. We find that misreporting levels are relatively low with 8% of smokers classified as smokers by clinicians stating that they are non-smokers. Our own measure of misreporting concludes that the clinicians objective smoking measures are most likely overstating misreporting levels by up to 50%, which also implies that objective smoking measures may not be preferable over self-reported measures.
Keywords: Misreporting, Smoking, Mixture Mode, BMI, Carbon Monoxide
JEL Classification: C1, C11, D03, I1
Suggested Citation: Suggested Citation