Future research must seek to identify the optimal size of a geographic unit for characterizing health effects. Some evidence suggests that the size of the geographic unit matters for capturing and potentially intervening on residential factors. A study in Michigan found that large racial differences in prostate and breast cancer survival in large geographic areas (15 federal house legislative districts) were markedly reduced and sometimes eliminated in smaller geographic units (110 state house districts and 212 neighborhood areas).74 Interestingly, there were three neighborhood areas, with high and persistent poverty for whites, where blacks had better survival than whites.
