Learn More about Deriving Group-level Estimates from Individual-level Intakes

The focus of the Diet Assessment Primer is on group-level assessments of dietary intakes for population research, such as surveillance, epidemiology, and evaluations of interventions. In population research, the objective is to estimate parameters, such as [glossary term:] mean intake, [glossary term:] usual dietary intake distributions, or the [glossary term:] correlation between dietary intake and health [glossary term:] outcomes for a group of individuals. This is distinctly different from the clinical setting in which the objective is to assess the nutritional status or dietary habits of a single individual for the purpose of personalized treatment or counseling.

The distinction between group and individual level dietary intakes is potentially confusing because dietary intake data are collected at the individual level, even for research in which the parameters of interest relate to the population. Also, there is a common misconception that population-level research requires a true value for every individual in the sample. However, that is not the case.

Statistical modeling allows for the use of sample data from each person to estimate population parameters (Learn More about Statistical Modeling). For example, a study that captures two 24-hour dietary recalls (24HRs) on each individual in the sample can estimate the distribution of intake for the sample, even though the average of the two 24HRs for any one person would not provide a very close approximation of that person's true habitual intake.

In summary, researchers interested in making group-level estimates need to collect representative data from each individual to use in estimating population parameters, even though the data would not be sufficient to provide satisfactory individual-level estimates.