Household scanner data are a rich resource for understanding food purchasing habits in the United States. The IRI Consumer Network provides a detailed account of the retail food purchases for a large, nationally representative sample. These data further include self-reported height and weight for a subset of households that complete the MedProfiler survey. Together, the Consumer Network and MedProfiler surveys provide a unique opportunity to study the relationship between diet and obesity. This report includes an assessment of the MedProfiler height and weight data in determining body mass index (BMI) for children and adults, using MedProfiler data from 2012 to 2018 and National Health and Nutrition Examination Survey data from 2011–2012 to 2017–2018. In addition, because self-reported height and weight may often be misreported in survey data, the report explores adjustment methods to account for any self-reporting measurement bias. Finally, since food-purchase data are collected at the household level, the report includes a comparison of methods for defining the obesity status of a household.
Details
Title
Assessment and Adjustment of Body Weight Measures in Scanner Data
Record Identifier
https://ageconsearch.umn.edu/record/338949
Language
English
Total Pages
40
Note
This study examined patterns in body weight status across individuals and households using the IRI Consumer Network household panel survey and the IRI MedProfiler survey from 2012 to 2018. The ERS researchers used height and weight data from the National Health and Nutrition Examination Survey (NHANES) of the National Center for Health Statistics to compare with self-reported height and weight data from the MedProfiler. To correct for possible measurement bias in BMI calculations based on MedProfiler data, the researchers considered three adjustment methods: (1) removing outliers based on the minimum and maximum measured height and weight values reported in NHANES; (2) removing outliers based on the MedProfiler interquartile range; and (3) predicting BMI in the IRI MedProfiler using measured BMI and percentile rankings of self-reported BMI in NHANES. Using the percentile-ranking adjustment method for adults, with no adjustments for children and youths, the researchers classified households by body weight status as normal weight, overweight, and obese. They compared household obesity levels for four possible ways of defining obesity at the household level, based on obesity of: (1) the primary shopper, (2) any member of the household, (3) at least half the household members, or (4) all household members. In a subsample of households with children, four additional definitions of household obesity were considered based on obesity of: (1) the primary shopper, (2) any child, (3) at least one adult and one child, or (4) all household members. Finally, the researchers compared results for household obesity status for all households, by race and ethnicity of the primary shopper, and—for all households only—by the number of household members.