Real Evidence? Forget about it! We have mathematical models.

What do you do when the real world fails to cooperate and instead gives you evidence contrary to expectations? Well, naturally you ignore the evidence and use complex speculation instead.

The only sustained objection in litigation worse than relevance is speculation. A sustained relevance objection often means the cross examiner has nowhere to go with his questioning. Speculation on the other hand often points to a witness's underhanded attempt to distract or mislead the jury. And speculation is what a gratuitous mathematical model is.

Experts get to speculate sometimes, but not when there is perfectly good evidence available -- in this case epidemiological evidence. Why turn to modeling when you can turn to real data, such as was done by Talbott? Because the real data in this case won't help you sell the goods.

And so, here we go again with not social scientists but mathematics professors! That's right. Now the mathematicians have been recruited to show how magical [abstract] all this talk about circumcision is.

This study concludes with not so famous last words,

... an intervention will fail if steps are not taken to prevent the majority of men increasing their sexual activity due to overestimation of circumcision's benefits.*

There you have it, the great unknown variable. And as has been seen, circumcised men begin to approximate behavior of their uncircumcised brethren in a more permissive environment, thereby raising their infection levels to the average of their communities.

*Edit: I just noticed that the authors' conclusion is in fact contradicted overstated in the abstract by this observation: 

These benefits are lessened with increasing sexual activity in men who have been circumcised, with complete negation of the intervention occuring once a lower bound of 40% of men increase their activity.

Yeah, so forget a "majority of men increasing their sexual activity." You need a mere 40% to negate the benefit. But again, we already knew circumcision fails as an HIV public health measure from the epidemiological evidence.

Apologies for the heavy linkage.

Comments

[SunkenShip]

Great post. I too wonder why the heck they rely on mathematical models when they don't translate to real life at all!

Mathematical models predicted a catastrophic increase of HIV infections in Brasil, guess what? It never came to pass. Mathematicians should stay away from the pro-circ lobby.