Isn't that the point? People will have risky sexual behavior whether circumcised or not, so if circumcision reduces the likelihood of AIDS transmission in people with risky sexual behavior, it's a net win.

EDIT: the downthread comment clarified the parent comment for me.

As I said even further downstream, if the study was done naively, where we simply took circumcised vs. uncircumcised people and compared their HIV rates, I would agree 100%.

But the original study was taking a population of adult uncircumcised males in africa and circumcising 50% of them randomly, and then tracking their HIV infection rates. That should control for upbringing. From the studies:

"In these studies, men who had been randomly assigned to the circumcision group had a 60% (South Africa), 53% (Kenya), and 51% (Uganda) lower incidence of HIV infection compared with men assigned to the wait-list group to be circumcised at the end of the study. In all three studies, a small number of men who had been assigned to be circumcised did not undergo the procedure; likewise, a small number of men assigned to the control groups did undergo circumcision. When the data were reanalyzed to account for these occurrences, men who had been circumcised had a 76% (South Africa), 60% (Kenya), and 55% (Uganda) reduction in risk for HIV infection compared with those who were not circumcised."