Related Experiment Videos
Positional error in automated geocoding of residential addresses
Michael R Cayo1, Thomas O Talbot
1Geographic Research and Analysis Section, Bureau of Environmental and Occupational Epidemiology, New York State Department of Health, 547 River Street, Room 200, Troy, NY 12180-2216, USA. mrc02@health.state.ny.us
International Journal of Health Geographics
|December 23, 2003
Summary
Automated geocoding introduces positional error, increasing with lower population density. Using residential property parcel data significantly reduces this error, especially in rural areas, offering a more accurate alternative for geographic information system applications.
Area of Science:
- Environmental Health
- Geographic Information Systems (GIS)
- Spatial Analysis
Background:
- Public health research increasingly utilizes Geographic Information System (GIS) technology.
- Accurate residential address geocoding is crucial for assessing environmental contaminant exposure.
- Automated geocoding using street centerlines introduces positional error.
Purpose of the Study:
- To evaluate positional error in automated geocoding of residential addresses.
- To assess how geocoding error varies across different population densities.
- To compare automated geocoding with an alternative method using residential property parcel data.
Main Methods:
- Assessed positional error for 3,000 residential addresses by comparing geocoded points to true locations via aerial imagery.
- Analyzed geocoding accuracy across rural, suburban, and urban population densities.
- Evaluated geocoding using residential property parcel points as a reference.
Main Results:
- Positional error increased as population density decreased; 95% error was 2,872m (rural), 421m (suburban), and 152m (urban).
- The alternative method using property parcel points showed significantly reduced error: 95% within 195m (rural), 39m (suburban), and 21m (urban).
Conclusions:
- The level of geocoding error can impact research outcomes and must be considered.
- Residential property parcel data offers a more accurate alternative for geocoding when traditional methods yield unacceptable error levels.