Skip to main content
ARS Home » Northeast Area » Beltsville, Maryland (BHNRC) » Beltsville Human Nutrition Research Center » Nutrient Data Laboratory » Research » Publications at this Location » Publication #275734
Title: ARS, USDA updates food sampling strategies to keep pace with demographic shifts

Author
item Pehrsson, Pamela
item PERRY, CHARLES - Consultant
item DANIEL, MARLON - Johns Hopkins School Of Public Health

Submitted to: Procedia Food Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 1/27/2013
Publication Date: 5/29/2013
Citation: Pehrsson, P.R., Perry, C., Daniel, M. 2013. ARS, USDA updates food sampling strategies to keep pace with demographic shifts. Procedia Food Science. 2:52-59.

Interpretive Summary: The National Food and Nutrient Analysis Program (NFNAP) was implemented in 1997 as a collaborative food composition research effort between USDA and NIH. The goal of this program is to obtain nationally representative estimates of the nutritional components of important foods consumed in the U.S. for inclusion in the USDA National Nutrient Databank System; to date, analytical food composition data generated for over 1800 foods have vastly improved overall data quality in the database. The NFNAP sampling approach was updated in 2001 using 2000 U.S. Census data and recently updated to use 2010 Census population estimates. This design, like the 2001 design, employs a three-stage, stratified, probability-proportional-to-size (PPS) sample selection process; 1) county selection (based on population density); 2) supermarket outlets within selected counties (based on annual sales); and 3) specific brands of foods (based on market share data). In the first stage, Census regions (4), divisions and states were used to obtain a self weighting sample of population centers, ensuring geographic dispersion across the 48 conterminous states; 48 locations were selected, with nested subsets of 24,12 and 6 locations. Due to demographic changes in the population and congressional redistricting it was necessary to revise the sampling scheme to reflect these changes. With the increased penetration of warehouse-type retail outlets into the grocery industry, the sampling frame was adjusted to include these purchase locations. Food samples which are collected nationally according to a statistically rigorous sampling approach are consistent with national representativeness and allow better estimates of the mean and variability than convenience sampling or less rigorous options.

Technical Abstract: The Nutrient Data Laboratory, USDA implemented the National Food and Nutrient Analysis Program (NFNAP) in 1997. The goal of this program is to obtain nationally representative estimates of the nutritional components of important foods consumed in the U.S. for inclusion in the USDA National Nutrient Databank System; to date, analytical food composition data generated for over 1800 foods have vastly improved overall data quality in the database. The NFNAP sampling approach was updated in 2001 using 2000 U.S. Census data and recently updated to use 2010 Census population estimates. This design, like previous iterations, employs a three-stage, stratified, probability-proportional-to-size (PPS) sample selection process; 1) county selection (based on population density); 2) supermarket outlets within selected counties (based on annual sales); and 3) specific brands of foods (based on market share data). In the first stage, Census regions (4), divisions and states were used to obtain a self weighting sample of population centers, ensuring geographic dispersion across the 48 conterminous states; 72 locations were selected, with nested subsets of 24, 12 and 6 locations. Due to demographic changes in the population and congressional redistricting it was necessary to revise the sampling scheme to reflect these changes. With the increased penetration of warehouse-type retail outlets into the grocery industry, the sampling frame was adjusted to include these purchase locations. Food samples which are collected nationally according to a statistically rigorous sampling approach are consistent with national representativeness and allow better estimates of the mean and variability than convenience sampling or less rigorous options.