Title: Semiparametric estimation of the distribution of episodically consumed foods measured with error
Authors: Felix Camirand Lemyre - Université de Sherbrooke (Canada) [presenting]
Raymond Carroll - Texas A and M University (United States)
Aurore Delaigle - University of Melbourne (Australia)
Abstract: One purpose of collecting dietary data in national surveys has been to estimate the distribution of usual intake of various nutrients and food groups in populations and subpopulations. When dietary intake data are obtained by 24-hour dietary recall, it is now widely recognized that the usual intake is assessed with considerable measurement error. There is a vast literature on measurement errors for estimating usual intake distributions when the food/nutrient is consumed daily. However, the classical methods cannot be used when considering the usual intake of episodically consumed foods (e.g., fish or whole fruits) because in that case those methods are typically biased. We address this problem by using a mixture of a parametric and a non-parametric approach.