Weighting by number of homicides is actually reasonable. Since homicides are rare, variance of homicide rate is driven by number of homicides rather than by pop

Weighting by number of homicides is actually reasonable. Since homicides are rare, variance of homicide rate is driven by number of homicides rather than by population. Weighting by inverse variance is pretty common.