there is a negative correlation between respiratory disease, and cardiovascular disease and neoplasm. Mortality due to respiratory disease is consistently above average, and mortality due to cardiovascular disease and neoplasm consistently below average. This fact is not visible in the analysis of differences in mortality levels.
Geographic interpretation of the cluster age patterns is customary in an analysis of this kind. The distribution of provinces by cluster membership is mapped in Figures 3-3a, b, c, and d. Geographic labels are assigned to all six female clusters, but to only three male clusters. As the maps show, the geographical territory covered by the age clusters is often not clearcut, especially with large clusters. The geographic designation is assigned based on the area in which a cluster is dominant (e.g., female clusters 1-410).
Two small female clusters, 5 and 6, have clear geographic designations. Cluster 5, "Urban, Far East and North," represents areas situated on the frontiers of Russia with an extreme climate, poor social infrastructure, and high migration. Cluster 6 has five of seven members belonging to the autonomous republics of North Caucasus; the sixth is the neighboring Kalmytskaya autonomous republic. In fact, this cluster contains nearly all autonomous republics of southern Russia that are not clustered and therefore classified as outliers.
The most homogeneous male clusters are rural clusters 5 and 6. Since the age patterns of mortality of these clusters are similar, we can aggregate their members in one cluster stretching from northwest to southeast across the center of European Russia, and therefore labeled as "Middle European." In contrast, the aggregation of two pairs of female clusters with similar age patterns (urban clusters 1 and 2, rural clusters 3 and 4) has no clear geographic interpretation.
In general, only relatively small clusters have a distinct geographic designation. The difficulty of defining geographic identity in classifications of this sort is common rather than unusual. Problems in assigning a geographic designation to specific age patterns of mortality arose in the creation of the Coale-Demeny models (hence we have the compromise model West), as well as with the U.N. models (United Nations, 1982; Heligman, 1984); models produced by the Organization for Economic Cooperation and Development (OECD) did not even assign geographical labels (Organization for Economic Cooperation and Development, 1980; Waltisperger, 1984).
The fact that age mortality profiles are similar across a wide range of geographical areas points out that, among factors that influence the mortality patterns of Russian provinces, universal factors are more important than those linked with specific regions. This homogeneity may reflect uniformity of lifestyles, behavioral patterns, attitudes, values, and quality of life among ethnically similar Russians.