.png){kind=icon}
Return
Estimating maximum acceptable positional errors for crime distribution pattern
.png){kind=icon}
PRE
AI
DOI：10.1016/j.compenvurbsys.2026.102464.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Geocoding converts textual crime addresses into geographic coordinates and serves as the foundation for most quantitative spatial analyses in criminology. However, the resulting geocoded coordinates often contain positional errors, which can distort the spatial pattern of the data. The impact of these positional errors depends not only on their magnitude but also on the spatial scale of crime analysis. However, a standardized measure integrating both factors is currently lacking. To address this gap, we introduce the Ratio of Mean Positional Error to Average Unit Edge Length (ROML), a standardized measure expressing positional error relative to size of spatial unit. Using crime data from Cincinnati, Washington D.C., and Chicago in the U.S., we evaluated the effects of ROML across fourteen spatial scales, including neighborhoods, census tracts, census block groups, and census blocks, as well as regular and hexagonal grids. Results show that as the ROML increases, the impact on the spatial patterns consistently increases, indicating that the spatial crime patterns become more biased. Notably, when the positional errors were twice the average unit edge length, the spatial patterns derived from data containing positional errors were significantly different from the true spatial patterns. This threshold is uniform across the three cities, four crime types, and fourteen spatial scales. This standardized threshold offers a benchmark for evaluating the effects of positional errors on spatial distributions of crime.
Keywords:
Geocoding uncertainty
Positional error
Crime distribution pattern
ROML
S index
Journal
IF:
8.3
Papers:
1.6K
Citations:
8.3K
Organization
Cited Papers
No cited papers available