The FCID Commodity Consumption Calculator is an application that uses the NHANES/WWEIA FCID food commodity intake information to provide estimated mean and various percentiles of consumption of a user-selected food commodity or series of food commodities. Specifically, the number of consumers in the database, the mean consumption, and consumption estimates associated with various percentiles from the 5th to 100th (i.e., maximum reported) are provided. The database user should keep in mind, however, that statistical estimates based on a small number of survey respondents may be less statistically reliable than estimates based on larger numbers of respondents. To the extent that the data user desires better approximations or additional details such as standard errors, confidence limits, or design effects/variance inflation factors, the user should instead use the raw data files accessible through the orange "Database Contents" button along with appropriate survey-capable statistical software(e.g., SAS, Stata, R, or SUDAAN) that properly incorporates information on sample weights, PSU, and strata.
The "Third Report on Nutrition Monitoring in the United States" makes recommendations regarding minimal reporting requirements (see LSRO 1995 or p.2-3 of "Healthy People 2010 Criteria for Data Suppression (2002) at http://www.cdc.gov/nchs/data/statnt/statnt24.pdf ) and suggests that estimates may be unreliable if the value of the calculation ([Number of Respondents])*(1-[Proportion]) is less than 8*([Variance Inflation Factor]). The Percentile calculator here provides reasonable estimates of consumption based on relatively simple and basic algorithms that properly take into account the NHANES survey sampling weights, but the Calculator provides no indication as to the statistical reliability or confidence limits around those estimates; the estimates are necessarily less reliable (and confidence limits correspondingly wider) with smaller numbers of respondents or for commodities that are less commonly consumed, and it is left to the user of the database to make appropriate judgements as to whether the data are "fit for purpose". For additional details regarding the appropriate use of the data with respect to statistical inferences for commodities with low numbers of reported consumption events or for the tails of the distribution, the user should consider consulting with a qualified statistician for advice and counsel.