Performance of the NCI method of dietary intakes in small sample sizes

Abstract

Modeling the population distribution of usual intake of episodically-consumed foods, primarily using 24-hour recall, is a challenging problem. The NCI method is the most recent major step in the evolution of specialized techniques that address it. This thesis is a contribution to the effort of spreading the usage of the method to a wider research community. We implement the basic method in the R environment, increasing the number of researchers and practitioners in the position to use the method. We observe the performance of the method in small samples using two different datasets, and in three potentially-useful scenarios. From these observations we draw recommendations for the minimal sample composition and size.