>In message <F277OUIHNkjgJKVApwO0000cbab at hotmail.com>, jan carowan
><jancarowan at hotmail.com> writes
> >a public health expert such as Richard Levins may ask: does a significant
> >statistical difference in the characteristic of interest between the
> >contrast classes prove that the criterion used for setting up the groups
>is
> >indeed causally significant?
>
>Well, call me thick, but I still don't understand what you mean.
>
Take the example of blood types as hidden variables:
If you are ignorant of hidden variables (as the technical term goes) of blood type, rhesus factor, etc. your prediction of the outcome of a transfusion can only be statistical in nature ("The overall failure rate is X%). If you know these variables however you can predict case by case outcomes in a blood transfusion.
In the case of Aids contraction, you seem to me to be ignoring or downplaying the hidden variables while resting with very imperfect statistical knowledge: if you are gay or a drug user or African the chance of the random event of Aids is significantly higher than it is for "most people." Such "knowledge" is not causally relevant, but it does lead to invidious categorization.
> >
> >>Indeed there is information useful for action. Namely it is unnecessary
> >>to promote undue alarm and fear amongst straight people who are unlikely
> >>to get Aids.
> >
> >Straight women are unlikely to get Aids?
>
>Well, on the evidence from Britain's 20 000 cumulative total of Aids
>cases, 252 contracted through heterosexual contact suggests that yes,
>straight women are not very likely to get Aids (unless they fall into
>another high risk category).
You seem to me to be defining straight woman as one who precisely does not also fall into a high risk category. This seems to me to be essentialist, invidious categorization.
Warm regards, Jan
_____________________________________________________________________________________ Get more from the Web. FREE MSN Explorer download : http://explorer.msn.com