Lead crime and teenage pregnancy.
There is a very impressive time series and cross country correlation between lead exposure in early childhood and violent criminal behavior in ones late teans and early 20s if one is male or getting pregnant in one's teens if one is female ( the teenage boys are probably involved in that too ). There is also a clear correlation between lead in kid's hairs and impulsiveness and anger management issues. How about data for more than 1 person but less than a whole country ?
The lead hypothesis suggests that the rise in crime in the US in the 60s and the decline in the 90s should be concentrated in big cities where there are lots of cars per acre and high concentrations of lead (or maybe lead where food is grown is what matters ???). I definitely have the impression that this pattern is seen in the data.
A competing explanation is drugs with the crime wave due to the heroin then the crack epidemics. Epidemics affect many people in one area and pass (here immunity comes from seeing the damage done). A change in the geography of crime miight follow a change in the geography
of easy access to addictive drugs. There is a difference between plant products and synthetic meth. Also drugs used to be snuggled mostly mixed with legal goods so they entered in big port cities. Now(I think) they mainly enter empty border regions with few drug sniffing dogs per acre. How to control for this ? I think overdose deaths are not a good measure as emergency treatment improved and it helps one's chances to be within a mile of an emergency room with naltrexone.
Still if not a fruitful area for actual research, this seems to me to be an excellent area for uninformed speculation.