Social Science Statistics Blog

« January 10, 2006 | Main | January 12, 2006 »

11 January 2006

Applying Spatial Statistics to Social Science Research

Drew Thomas

Spatial Statistical methodology is beginning to gain popularity as a methodological tool in the natural and social sciences. At Harvard, Prof. Rima Izem is leading the way towards the use of these techniques across many disciplines. This semester, Prof. Izem debuted her Spatial Statistics seminar, which met Wednesday afternoons in the Statistics Department.

Of those topics discussed in the seminar, lattice data analysis proves to be invaluable to the analysis of well-defined electoral districts. The principle of lattice data is that our land area can be divided into mutually exclusive, complete and contiguous divisions; interactions between the divisions can then be analyzed through various covariance methods.

A full understanding of spatial interaction may prove to be valuable to electoral analysis. Determining the interdependence of districts through means other than traditional covariates may suggest the presence of a true "neighbor effect." How one determines the covariance of districts may prove to be more art than science, but the depth of work yet to be done in this field should give many opportunities for meaningful investigation.

Posted by Andrew C. Thomas at 1:41 AM