On Wed, 27 Oct 1999, Jim Clark wrote:
> Hi
>
> On Tue, 26 Oct 1999, Michael Sylvester wrote:
> > is it still revelant to to use subject variables like gender and ethnicity
> > in the randomized block design?
>
> The answer would be: "it depends."
>
> (1) It depends on what you mean by "use." A common problem with
> students is confusing the blocking or matching factor in a
> Randomized Block design versus the inclusion of an individual
> differences variable (e.g., Gender, Ethnicity) as a factor in a
> Factorial design. Matching factors involve individual
> participants in one group being matched with individual
> participants in other groups (e.g., twins, litter mates, cases
> matched for some individual difference measure such as IQ or
> depression level). So variables like Gender and Ethnicity do not
> make good blocking variables, although they can be included as an
> additional factor along with some other within-subjects factor
> (i.e., as part of a randomized block design).
>
> (2) It also depends on what you want to find out from your study
> and whether it is reasonable to ask about the consistency of
> results across gender, ethnicity, or whatever. If you are
> interested in whether the effect of Factor X (the within-subjects
> factor) is the same for Men and Women, for Canadians and Greeks,
> or whatever, then it is hard to answer this question without
> including the different groups.
>
> Best wishes
> Jim
I asked this because we keep on hearing about that famous aspirin study
that concluded that low dose aspirin taken daily could ward off heart
attacks. Dr.Judith Reichmann,a famous Gynecologist,who has published
on menopause pointed out that this study did not include one single
woman because it was thought that heart attacks were the province of men.
She refers to that study as an example of gender bias in medical research.
Michael Sylvester
Daytona Beach,Florida
