Goal: To use spatial statistics tools to learn more about overdose death rates in the contiguous 48 states.
Problem: I'm not sure which tools I should use. I don't have county-level data, just rates for the entire state.
Data: Five years worth of death rates for each state and the District of Columbia. I also have state data pertaining to poverty, unemployment, that I can use as exploratory variables.
What I've done: I've run spatial autocorrelation, exploratory regression, and I was going to do geographically weighted regression but decided against it because my sample size was too small. I've been running these spatial statistics tools on the lower 48 U.S. states and DC for a selected year (49 observations).
Any suggestions? Flags? Glaring issues? Are there any tools or strategies that make sense in particular? I already know that rates seem to be significantly worse in some states. Just mapping the rates by year makes that evident.
Issues? "Ecological Fallacy" always comes to mind when dealing with statistics at such a scale.