> where crime is more likely

This is the fundamental disconnect in your argument, because historical recorded arrest rates are not necessarily reflective of actual crime rates.