. Author manuscript; available in PMC: 2014 Oct 1.

Published in final edited form as: Am J Prev Med. 2013 Oct;45(4):462–473. doi: 10.1016/j.amepre.2013.06.009

Table 5.

By primary data–gathering approach, evidence for validity of secondary retail food data reported (n=19)^a

	Commercial Sources^b	Government Sources^c	Local Directories^d	Omnidirectional Sources^e
Ground-truthed, GPS-Assisted (n=4)^12,14,17,18
Percentage Agreement^f	Almost Perfect 0.85¹⁸^g 0.90¹²^g	Substantial to Almost Perfect 0.64¹⁴^h 0.76¹⁷ 0.82¹⁸^g	Substantial 0.64¹⁴^h 0.77¹⁸
Sensitivity	Moderate to Substantial 0.59¹²^g 0.65¹⁸^g	Moderate to Almost Perfect 0.42¹⁸^g 0.82¹⁷	Moderate 0.55¹⁸
Positive Predictive Value	Moderate to Substantial 0.49¹⁸^g 0.62¹²^g	Fair to Almost Perfect 0.31¹⁸^g 0.92¹⁷	Moderate 0.41¹⁸
Cohen’s Kappa Coefficient	Moderate 0.43¹⁸^g	Fair 0.24¹⁸^g	Fair 0.24¹⁸
Concordance	Moderate 0.42¹⁸^g 0.44¹²^g	Fair 0.26¹⁸^g	Fair 0.35¹⁸
On-Site Verification (n=12)^{13,19–27,29,30}
Percentage Agreement^f	Substantial to Almost Perfect 0.65^{20 i} 0.72³⁰ 0.73¹³^g 0.77^25,26 0.86²⁹^g	Fair to Almost Perfect 0.34²² 0.50³⁰ 0.64²⁹ 0.70²¹ 0.77¹³ 0.80²⁰ 0.85²⁴ 0.86²³^g 0.88²⁷	Fair to Almost Perfect 0.37¹⁹^h 0.54²²^g 0.65²⁶^g 0.71²⁹ 0.88²⁵	Fair to Substantial 0.37¹⁹^h 0.69²⁴ 0.78²⁹
Sensitivity	Moderate to Almost Perfect 0.60¹³^g 0.84²⁶ 0.90²⁹^g 0.96³⁰	Moderate to Almost Perfect 0.46²³^g 0.50³⁰ 0.68¹³ 0.75²⁹ 0.84^21,22 0.85²⁴	Moderate to Substantial 0.52²²^g 0.66²⁶^g 0.74²⁹	Almost Perfect 0.81²⁹
Positive Predictive Value	Substantial to Almost Perfect 0.70³⁰ 0.82¹³^g 0.90²⁶ 0.94²⁹^g	Almost Perfect 0.81²⁹ 0.82²¹ 0.89¹³ 0.89²³^g 0.92²² 1.00³⁰	Almost Perfect 0.81²²^g 0.95²⁹ 0.98²⁶^g	Almost Perfect 0.95²⁹
Concordance	Almost Perfect 0.94²⁹^g	Fair 0.23²⁹	Fair 0.27²⁹	Almost Perfect 0.87²⁹
Targeted Observational Field Data (n=3)^28,31,32
Percentage Agreement^f	Almost Perfect 0.88²⁸			Fair to Almost Perfect 0.36³² 0.92³¹
Cohen’s Kappa Coefficient	Moderate 0.48²⁸			Fair 0.21³¹

Levels of agreement for all evidence for validity findings reported were interpreted using the Landis and Koch criteria (<0.00 poor, 0.00–0.20 slight, 0.21–0.40 fair, 0.41–0.60 moderate, 0.61–0.80 substantial, and 0.81–1.00 almost perfect).

Averages findings reported on Dun & Bradstreet (U.S.); InfoUSA or ReferenceUSA; InfoCanada; Krak Denmark (Web-based search engine); Stockman Company (chain food addresses); and Tamec Inc.

Averages findings reported on City Health Department (United Kingdom and U.S.); County Health Department (U.S.); State Department of Agriculture (U.S.); State Department of Health–authorized WIC retailers (U.S.); State Department of Taxation and Finance (U.S.); State Department of Health (U.S.); State Liquor Authority (U.S.); state-authorized lottery ticket retailers (U.S.); USDA-authorized Supplemental Nutrition Assistance Program (SNAP) retailers; Country Food Administration (Denmark); and National Tax Registry.

Averages findings reported on the variety of online and local telephone directories examined.

Averages findings reported on Google Earth (U.S.); Google Street View (U.S.); and Google Maps Denmark.

Frequencies or dispositions percentages, when necessary, were used to calculate a percentage agreement.

Average findings reported across a combination of data sources (e.g., ReferenceUSA and Dun & Bradstreet or multiple government sources)

Not all studies reported evidence for validity by specific data source (e.g., Sharkey¹⁴ grouped local/area telephone directories, Internet telephone directories, and a list of Current Food Establishment Group Firms from the Texas Department of Agriculture) so the total evidence reported was used for each data source examined.

ⁱ

Comparisons were made between results generated using primary versus secondary data for fast-food density and proximity, convenience store proximity, and food deserts.

USDA, U.S. Department of Agriculture; WIC, U.S. Department of Agriculture’s Special Supplemental Nutrition Program for Women, Infants, and Children