Comparing Esri ArcGIS and SAS Geocoding Approaches: Test case with 3,238 Wisconsin addresses

Area of Science:

• Environmental Epidemiology
• Geospatial Analysis
• Public Health Research

Background:

• Environmental exposure assessment in cohort studies relies on linking residential addresses to spatial data.
• Different geocoding platforms may yield varying results, impacting data comparability.
• The Cohorts for Environmental Exposures and Cancer Risks (CEECR) consortium uses both ArcGIS and SAS for geocoding.

Purpose of the Study:

• To compare the geocoding accuracy and data comparability of ArcGIS Geocoding and SAS GEOCODE Procedure.
• To assess the impact of different geocoding methods on linking survey data to environmental and socioeconomic factors.
• To evaluate the suitability of these platforms for CEECR consortium studies, considering offline use and confidentiality.

Main Methods:

• Geocoded 3,238 Wisconsin residents' addresses using both ArcGIS and SAS GEOCODE.
• Linked geocoded data to eight environmental and socioeconomic datasets at various geographic scales.
• Compared geocoding accuracy (point, street, zip, census tract, block group, county) and concordance of linked exposure measures.

Main Results:

• ArcGIS showed slightly higher specificity in address matching (97.7%) compared to SAS (95.9%).
• Both platforms demonstrated high concordance at county (99.6%), census tract (96.5%), and census block group (94.7%) levels.
• Correlations for linked exposure measures (socioeconomic status, air pollution, cancer risk) were very strong (ICC ≥98%) at census tract and block group levels.

Conclusions:

• ArcGIS and SAS GEOCODE are highly comparable for geocoding and linking environmental/socioeconomic data for the majority of addresses.
• SAS GEOCODE is a cost-effective option for researchers already using SAS, offering similar results to ArcGIS for most applications.
• Minor discrepancies in geocoding accuracy, particularly in rural areas, should be considered in spatial exposure assessment.