A study of US counties shows that counties with high arrest

A study of U.S. counties shows that counties with high arrest rates for armed robberies (with a firearm) also have high proportions of conservative Republicans. The authors conclude that Republicans commit more armed robberies than non-Republicans.

What is the unit of analysis under investigation in the research project? What mistake does the researcher make in the conclusion? What is a more likely explanation for the association between party identification and robbery arrest rates on a county level?

Solution

Unit under analysis is the number of arrests.

The mistake is that they are assuming that the total proportion of arrest is same through out. But there is a possibility that the non-republicans may not get caught.

More likely conclusion is that most of the republicans get caught after the armed robbery