Robust point pattern inference from spatially censored data

Administrative data sources are increasingly used for spatial analysis and policy formation. For example, welfare-to-work programs have stimulated demand for spatial mismatch studies using ES-202 employment files. The increased resolution gained by geo-coding the address records in administrative files can be of enormous research value when the process under study resolves over small distances. Yet the resulting point-referenced data is problematic for inferential analysis. In particular, administrative data typically represents a sample of convenience thus posing serious validity problems for statistical inference. This paper proposes a robust estimation method for spatial pattern inference based on spatially censored data. The performance of the estimator is explored using simulated data and is also demonstrated on ES-202 data from North Carolina.