I think it is due to both nature AND nurture.

There aren't very many things that are due to only one or the other. Most things are due to a complex interplay of both internal (nature) and external (social) causal mechanisms. The controversy is in figuring out precisely what internal and external causes are involved and in figuring out how they interact with one another.

> Another study said testosterone is a part of aggression but not the cause for aggression. (whatever that means??)

What that means is summed by this little saying: Correlation doesn't imply causation.

Correlation is when two features are found to occur together. In this case, the finding that high levels of testosterone seems to be correlated with violence.

But the correlation doesn't tell us whether high levels of testosterone cause violence or whether violence causes high levels of testosterone. It could even be that they have a common cause and there is no direct relationship between them. Like how the weather and a barometer reading are correlated in virtue of having air pressure as a common cause.

There is also controversy with respect to how robust causal explanations need to be. An example of a robust genetic explanation is Huntington's. Huntington's is caused by repeats at the end of some chromosome or other (can't remember which one, sorry). This causal claim is robust because you can vary the environment considerably and yet if there are the repeats the person has huntingtons. An example of a non-robust genetic explanation is Schizophrenia. Identical twins are genetically identical. If one twin has schizophrenia there is only a 48% chance that the other twin will have schizophrenia, however. This genetic explanation is non-robust because environmental causal mechanisms are playing a more significant role than the genetic ones. If we know someone has the genetics 'for' schizophrenia it is still more likely than not that they don't have schizophrenia.

How does robustness relate to the testosterone case? I'd need to know the base rate for high levels of testosterone in the population to see what percentage of them had a history of violence. One would need to check the robustness of the correlation first and with respect to the causal claim... One would need to know whether testosterone levels were high PRIOR to violent behaviour or whether the violent behaviour preceeded the elevated levels.

END RANT