Lox - a rare compliment from me. How nice to see a correct use of "may" in place of the almost ubiquitous misuse of it for "might".

I have been puzzling over the insistence of the unholy alliance here that the prevalence of AIDS is statistically significantly linked, both as to incidence and as to growth rates to male homosexual behaviour and/or to being African.

In the case of Keith, he has indicated that his concern is solely correct statistical analysis. I find that hard to swallow (sorry if the imagery disturbs) in that the only places I have seen his pedagogic statistical bent exhibited are on this thread and on the thread about UK immigration.

It seems more plausible that the stated insistence is linked to hostility as such to male homosexuals and to Africans. In some cases (this is not a specific reference to Keith, but that in turn is not to waive the possibility of the reference including him) that that hostility is pretty baldly expressed as "Keep 'em out or lock 'em up" but it may be that it runs even deeper (and in less acceptable channels) than that.

Reverting now to statistics, Keith, you hypothesise, I think, that the rate of growth of AIDS (probably more correctly detected and reported AIDS) is linear.   I cannot be bothered to do the analysis, but if you take your figures over your number of years and then do a typical 2-tail Tukey test (or a Tukey-Neave test) on them, what is your degree of confidence found? If you took instead, the null hypothesis that the rate of growth was exponential, what degree of confidence would you find for that? I can't be bothered to dig my old copy of Ya-Lun Chou (or Sokolnikoff & Redheffer) out and do the sums (and anyway I haven't got the raw data) but given the shortness of the period of available data I would expect the two to be much the same. Tell me, O! statistical devotee, what does the statistical analysis show?   Depending on your answer, I may want to check your calculations and sources.