5.5 Point data: Tests for clustering
Points on a map can be evenly spaced, randomly distributed or exhibit clustering. Clustering is important because it may provide useful clues about patterns of disease causality and risk. It is possible to test for spatial clustering – the aggregation of health events in a particular area. Clustering tests consider all of the points in a study area and summarise the pattern across the whole data set . Because they consider all points in a data set, such tests are sometimes called global tests. Some such tests are designed to work with case data only, which show the locations of individuals diagnosed with a specified disease. Other tests are designed to be used with case data and data on disease-free controls. There are many tests for clustering in point data, but only three will be described here.
Activity
View the presentation below which introduces three tests for clustering in point data and read the suggested reference material.
Then download the attached Zip file which contains a dataset and instructions for the implementation of a cluster testing exercise using childhoood leukaemia data in either ArcGIS Desktop or Pro.
References (Essential reading for this learning object indicated by *)
Although it misses some more recent developments, notably the spatial scan statistic, this article provides an idea of the range of spatial cluster tests available:
Centers for Disease Control and Prevention (1990) Guidelines for Investigating Clusters of Health Events – APPENDIX. Summary of Methods for Statistically Assessing Clusters of Health Events 39 (RR11), 17-23 Available at http://www.cdc.gov/mmwr/preview/mmwrhtml/00001798.htm
The TerraSeer statistical adviser at http://www.biomedware.com/files/documentation/clusterseer/default.htm provides a useful overview of some of the key clustering tests and when to use them (follow the links to ‘products’, then select ‘clusterseer’ and look for ‘adviser’). kj