Estimating the Average Treatment Effects of Nutritional Label Use Using Subclassification with Regression Adjustment

• Activity Number: 236
• Type: Topic Contributed
• Date/Time: Monday, August 5, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Health Policy Statistics Section
• Author(s): Michael Lopez*+ and Roee Gutman
• Companies: Brown University and Brown University
• Keywords: Causal Inference ; Propensity Score ; Potential Outcomes ; Ordinal Exposures ; NHANES

Abstract

Propensity score methods are common for estimating a binary treatment effect when treatment assignment is not randomized. When exposure is measured in a dose (i.e., low - medium - high), however, propensity score inference requires advanced techniques which have received limited exposure (Imbens, 1999; Imai & van Dyk, 2004). One interest with an ordinal exposure is the average treatment effect between each exposure pair, from which it is possible to determine an optimal dose among all subjects. Traditional methods, including dichotomization of the exposure or a series of binary comparisons across exposure pairs, are generally inadequate for identification these effects. We combine subclassification with regression adjustment to estimate transitive, unbiased causal effects of an ordered exposure, and apply our method on the 2005-06 National Health and Nutrition Examination Survey to estimate the effects of nutritional label use, measured on a 5-point scale, on body mass index (BMI). Evidence suggests label use may cause lower BMI only when used with often or always frequency.