Evaluating Methods for Preserving Confidentiality During the Release of Records Incorporating Address or other GeoDataThe use of a Geographic Information System (GIS) to understand spatial patterns of cancer is becoming more common within the public health field (see the International Journal of Health Geographics for countless examples), yet few registries have access to in-house expertise for maximizing new GIS technologies. In order to collaborate with a third party in performing such analyses, there are justifiable concerns regarding data release and potential spatial confidentiality violations of released data. Data need to be sufficiently masked to preserve confidentiality, while at the same time maintaining enough spatial information to allow sophisticated and accurate analyses. The development of robust but effective (relatively easy to use and affordable) techniques for maintaining the confidentiality of spatial data from registries, while maximizing collaboration for spatial analysis, is required. Moreover, a SEER-wide standard for distributing data for spatial analyses needs to be developed.This proposal shall concentrate on optimizing methods for developing datasets with geographical data that can be shared in an appropriate manner without losing spatial patterns of interest. A form of spatial regression called Geographically Weighted Regression shall be tested. Other commonly used spatial analytical approaches, ranging from smoothing surfaces to tests for aggregate level spatial autocorrelation shall also be tested.
