Many healthrelated variables show spatial distributions. Fields such as human genetics, population genetics, demography, epidemiology and public health furnish spatial data and this project is concerned with appropriate statistical techniques for analyzing them and inferring underlying spatial processes from observed spatial patterns. We propose to develop techniques for characterizing spatial data detecting transition zones and defining clones and patches. We shall also develop test of various types of hypotheses for spatial data analyzing choropleth maps of gene frequencies testing boundaries of regions, analyzing spatial data as distances, and experimenting with restricted randomization procedures. Significance tests for the difference between two correlogram will be explored. Inferences from gene frequency surfaces to processes will be investigated by means of analysis of regression residuals and by studies of stimulated multiple migrations. These techniques will be applied to a large data base of gene frequencies and to anthropometric variables from Europe and elsewhere. We shall isolate and identify components due to selection, migration, isolation by distance, and genetic drift by combined considerations of geographic variations patterns and spatial correlograms. Detailed analyses will be made for six countries in Europe and for several smaller scale data sets at the village and tribal level. In the European data we shall also study the relation between genetic differences and language differences. An ethnohistorical data base will be completed which will furnish quantitative information on gene flow and language boundaries in Europe for the past 3000 years. This data base will be tested against the gene frequencies to see whether the latter reflect the putative ethnohistorical relations.
Showing the most recent 10 out of 24 publications
