Irving Scheffe wrote:
> I'm quite confident that, if you examine any
> data you want, you'll emerge with the conclusion
> that a 4.5 to 1 ratio is, in fact, "huge." If you
> feel otherwise, do the study and refute us.
>
> An even more interesting question (than
> the "hugeness" of the 4.5 to 1 ratio) would be this:
I don't think a 4.5 : 1 ratio is a priori evidence for a _huge_ difference in
any data set unless you bring to bear all sorts of other info about
distribution, variability and context.
What about:
Group A Group B
0 0
0 1
0 0
This provides an even larger ratio of means. Is it huge? Is it small?
> If you go into numerous departments and cull out
> comparable groups composed only of men, with 4.5 to 1 citation
> ratios, what kind of salary differences do you
> observe? In other words, in the world at large,
> what kind of salary differences do you tend to
> observe between comparable groups of scientists
> with 4.5 to 1 citation ratios?
In my experience I'd expect a difference close to zero if other factors such
as age were controlled for. However, I work in a different environment. I've
spent a great deal of time in multidisciplinary research environments, for
example. Is 50 citations in vision science better than 8 in mathematical
models of memory (for an example within a single discipline department)? The
simple answer that the citation data don't, without several layers of
statistical or contextual support, allow a conclusion of a huge difference to
be drawn (in my opinion).
> You sound like someone who is fully prepared to have
> gender discrimination issues be "performance based."
I have no idea what you mean here.
I commented that salary increases etc. should be determined by a range of
publicly available performance criteria. The gender discrimination issue runs
much deeper, IMO (I've tried not to comment on this because I think it gets
very off topic). If gender discrimination existed I'd expect it (based on my
experience in a different University system) to be more complex. For example,
if you want to discriminate against someone just load them up with teaching
and administration: their research will suffer and their promotion prospects
be damaged.
Thom
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
