Regression Analysis for comparison of crime rates across a river

#1
I need a bit of help refining a regression comparing crime rates of counties separated by a river. My current equation looks like:

(crmrte1-crmrte2)^2=A0+rivA1+popdiffA2+cnty1+cnty2+st1+st2

Anyone have any ideas on a better regression design? Currently I have data on population, crime rates, and location. I also have data on law enforcement employment for certain counties.
 
#2
It's hard to say what the best regression design would be without seeing the data. There are model selection methods (e.g. Akaike Information Criterion method) that can help select your model.

A very simple way to select your model is to find the one that grants the biggest adjusted R^2 value.

What are you using to model the data?
 
Last edited: