Accuracy of commercial geocoding: assessment and implications

Abstract

BackgroundPublished studies of geocoding accuracy often focus on a single geographic area, address source or vendor, do not adjust accuracy measures for address characteristics, and do not examine effects of inaccuracy on exposure measures. We addressed these issues in a Women's Health Initiative ancillary study, the Environmental Epidemiology of Arrhythmogenesis in WHI.ResultsAddresses in 49 U.S. states (n = 3,615) with established coordinates were geocoded by four vendors (A-D). There were important differences among vendors in address match rate (98%; 82%; 81%; 30%), concordance between established and vendor-assigned census tracts (85%; 88%; 87%; 98%) and distance between established and vendor-assigned coordinates (mean ρ [meters]: 1809; 748; 704; 228). Mean ρ was lowest among street-matched, complete, zip-coded, unedited and urban addresses, and addresses with North American Datum of 1983 or World Geodetic System of 1984 coordinates. In mixed models restricted to vendors with minimally acceptable match rates (A-C) and adjusted for address characteristics, within-address correlation, and among-vendor heteroscedasticity of ρ, differences in mean ρ were small for street-type matches (280; 268; 275), i.e. likely to bias results relying on them about equally for most applications. In contrast, differences between centroid-type matches were substantial in some vendor contrasts, but not others (5497; 4303; 4210) pinteraction < 10-4, i.e. more likely to bias results differently in many applications. The adjusted odds of an address match was higher for vendor A versus C (odds ratio = 66, 95% confidence interval: 47, 93), but not B versus C (OR = 1.1, 95% CI: 0.9, 1.3). That of census tract concordance was no higher for vendor A versus C (OR = 1.0, 95% CI: 0.9, 1.2) or B versus C (OR = 1.1, 95% CI: 0.9, 1.3). Misclassification of a related exposure measure – distance to the nearest highway – increased with mean ρ and in the absence of confounding, non-differential misclassification of this distance biased its hypothetical association with coronary heart disease mortality toward the null.ConclusionGeocoding error depends on measures used to evaluate it, address characteristics and vendor. Vendor selection presents a trade-off between potential for missing data and error in estimating spatially defined attributes. Informed selection is needed to control the trade-off and adjust analyses for its effects.

Error and Bias in Geocoding School and Students’ Home Addresses

“…The collective findings of Zandbergen and Green (2007) nonetheless differ from those based on a previously described 5% random sample of 2,608 street addresses from the Environmental Epidemiology of Arrhythmogenesis in WHI (EEAWHI) (Whitsel et al 2006). In that study, we found that the fraction of participants’ addresses determined to be < 100 m from the nearest highway was relatively constant across mean positional errors of 150–600 m, a finding driven by the counterbalance of approximately equal false positive and negative rates over the same range.…”

“…They found a 200–500 m range of mean positional errors in their study of 126 Orange County, Florida, public school addresses, a somewhat higher range than that associated with geocodes assigned by four commercial vendors to a larger variety and number of street addresses in the 48 contiguous U.S. states (Whitsel et al 2006). In both studies, however, the ranges exceeded commonly used thresholds for identifying those at greatest potential risk of traffic-related exposures, raising due cause for concern.…”

The environmental epidemiology of atrial arrhythmogenesis

“…Indeed, the ability to reliably and accurately link residential addresses with spatial coordinates and statistical tabulation areas using geographic information systems software may be critically important because geocoding error has the potential to distort downstream interpolation of residence-specific ambient pollution exposures and definition of socioeconomic contexts within which pollution exposures occur. These distortions may, in turn, lead to biased estimates of association and contextual effect modification [41, 42]. …”

“…A fully Bayesian hierarchical framework in which site-specific regression parameters are represented as random effects with prior distributions reflecting anticipated variation among sites meets this need [49]. The advantage of this framework is that uncertainty in exposure measurements can be represented within it by prior distributions derived from the above-referenced participant address geocoding analyses, kriging analyses, and random-effects meta-analyses of the bivariate association between personal and ambient pollution concentrations [41-44, 46]. Additional sources of uncertainty—namely those associated with co-pollutants and PM chemical components—can be incorporated in a similar manner.…”

Influence of Demographic and Health Survey Point Displacements on Distance-Based Analyses

“…In this locational inaccuracy setting, much of the previous literature has explored properties of the introduced error (Bonner et al, 2003; Cayo and Talbot, 2003; Ward et al, 2005; Rushton et al, 2006; Whitsel et al, 2006; Zhan et al, 2006; Goldberg et al, 2007; Kravets and Hadden, 2007; Zandbergen, 2007, 2008, 2011; Zimmerman et al, 2007, 2010; Zandbergen and Hart, 2009; Zandbergen et al, 2011) as well as its impact on subsequent spatially-based analyses (Burra et al, 2002; DeLuca and Kanaroglou, 2008; Mazumdar et al, 2008; Jacquez and Rommel, 2009; Zimmerman et al, 2010; Zinszer et al, 2010), where the errors are unintentionally introduced due to incorrect geocoding. A number of other studies have investigated the locational error that is intentionally introduced (Armstrong et al, 1999; Leitner and Curtis, 2004, 2006; Kamel Boulos et al, 2006; Olson et al, 2006; Wieland et al, 2008), such as with DHS point displacement.…”