After an initial attempt to create a dual kernel density map with CrimeStat did not work, I created a similar dual kernel map by using two different rasters. I created a kernel density map of population density, and normalize both the Population and Crime rasters by their respective maximum values to give them comparable scales. Then, I subtracted the crime raster values from the population raster values to obtain the relative risk of crime for each pixel. The map shows the variation in risk across the city by showing the standard deviation, or average distance from the mean.