Validation Study of Commercial Food Store and Restaurant Availability in Commercial Business Lists

Purpose: This study validated the information on available food stores and restaurants found in two commercial business lists. Methods: Ground-truthed data on food stores and restaurants were compared to business list data from Dun & Bradstreet (D&B) and InfoUSA for 425 randomly drawn census tracts from the Chicago MSA and its 50 mile buffer. Validity was assessed for the full sample and across tracts by racial, ethnic, income and urbanicity characteristics. We estimated sensitivity and positive predictive value, each of which measured the extent to which the commercial lists contained retailers on ground and the extent of those listed retailers existed on ground, respectively. Results: Among all food stores observed on ground, 39.53% and 44.02% existed in Dun & Bradstreet and InfoUSA, respectively. The positive predictive values for D&B and InfoUSA, respectively, were 47.64% and 58.19%, respectively. Significant differences across tracts by income, race and ethnicity were found only for D&B, whereas both lists had significant differences by urbanicity. 53.22% and 63.17% of restaurants observed on ground were found, and 62.44% and 76.59% of listed restaurants were observed on ground in D&B and InfoUSA, respectively. No statistically significant differences were found for restaurants by income, although there were some differences by race, ethnicity and urbanicity. Validity also varied by store and restaurant type. Conclusions: The results imply that the validity of those commercial business lists varies by census tracts, in particular by urbanicity characteristics, and researchers need to address this in generalizing their results based on those lists.