I am trying to test to see if there is a "universal" trend in homicide across multiple nations.

First, I utilized squared semi-partial correlation coefficient by estimating two models (1) a model fitting nation and time dummy variables and (2) a model fitting just the nation dummy variable. Subtracting the R^2's from both models gives me the squared semi-partial correlation coefficient (or the approximate amount of variation shared among nations net of nation-specific explanations). After finishing that, I simply estimated an F-test to see if the model with versus without the time fixed effects were significantly different. The issue with this test is that the degrees of freedom are so high because of pooling across nation and time, that you cannot trust the p-value the F-test provides. But the editor of the journal requires an explanatory test.

thereofre, the second test I conducted was a basic group trajectory model. This was a reviewer from previous journal submissions suggesting. But the results are like no shit duh. Nations with high homicide have a high homicide, nations with a medium level of homicide have a medium homicide, and nations with low homicide have a low homicide. Duh. Not helpful and the current journal editor is not a fan of the test.

I am out of ideas of how to test if multiple trends of homicide across the nation are similar across time. While I do not believe they are, there are theories of crime that expect these trends to be the same/converge over time. There has to be a way to test that I cannot figure out or am just not thinking of it. Either way, it's driving me insane and any help would be greatly appreciated.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.