Store-level weights for the IRI InfoScan data allow researchers to develop projections from the retail stores currently included in the data purchased by the Economic Research Service (ERS) to the population of stores in the United States.1 IRI prepares the InfoScan datasets from data provided by retail establishments across the United States that have agreed to provide weekly retail sales data (revenue and quantity) for products with Universal Product Codes (UPCs) and random-weight (or perishable) products. The types of stores covered include grocery, drug, convenience, mass merchandiser, club, dollar, and defense commissary stores. IRI provides some of the InfoScan data to ERS at the store level, but in cases where the retailers did not approve release of their data at the store level, IRI provides the data at the retailer marketing area (RMA) level. Unlike the Consumer Network household scanner data purchased by ERS, IRI does not provide weights for stores in the InfoScan data.
The importance of weights was recently highlighted in the National Academies of Sciences, Engineering, and Medicine’s (NASEM’s) report A Consumer Food Data System for 2030 and Beyond (see Recommendation 4.9 in NASEM [2020]). Store-level weights can be used in a wide variety of research projects to calculate sales quantity, sales value, or other estimates that are representative of the population of stores. The store-level weights are useful because they allow analysts to create population estimates of sales quantity and sales value for foods and beverages at the national level or for major metropolitan areas. Without weights, estimates would underrepresent the total quantities or values.
The purpose of this report is to describe the approach and results of developing weights for stores in InfoScan for 2012 through 2018 and provide a user’s guide. This report will be updated in the future to include weights for 2019 and 2020 and replicate weights that can be used in variance estimation.
Contract No. GS-00F-354CA, Task 3
Final Report
