On Wed, 14 Mar 2001 13:19:36 +0000, Thom Baguley
<[EMAIL PROTECTED]> wrote:
>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?
Thom, it is not clear whether the differences in our positions
are that great.
First, I quickly agree that if two groups of professors have 0 and 1
citation, we can ignore the usefulness of citation counts in setting
their salaries. <G>
I think you must have misread my post <especially the
part you snipped>
My clause "any data you want" referred to the external *normative
data* used to help interpret the 4.5 to 1 ratio at MIT,
*not* to any citation data you want.
Let me "unsnip" the comment you had made that I was responding to.
>If you wish to infer that difference is "huge" that requires more than an
>observation of a difference - you need to explain and support that inference
>in some way. My point was that the baseball case allows you to state a
>difference, but an inference about the size of the difference requires further
>support. In the baseball case that support comes almost entirely from
>knowledge of baseball. (Was I really _that_ unclear?).
In other words, you were asking whether the MIT
data have supporting external normative information
to support the notion that the observed MIT performance
ratio is huge. My comment was referring to that normative
data. It invited you to gather such data and see for yourself.
It would be useful.
>
>> 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.
The notion of "comparable groups" was intended to imply that age and
research area *was* controlled for, a priori. Age was controlled
for in the groups we compared at MIT.
To be clear, are you asserting that, for example, if I went to several
biology departments where salaries are public, and divided the male
faculty, stratified by age, into subgroups with high and low citation
counts, that I would NOT tend to find differences in salary between
the high and low groups?
>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).
Different subareas of biology have different citation rates, no doubt.
The key question in assessing gender discrimination is NOT whether
perfect criteria are employed, or even if they exist. The key question
is whether the criteria that do exist are employed differentially
across gender.
There is no evidence that the MIT report writers, or their successors,
had any intention of analyzing any performance data, let alone a
sophisticated analysis of performance data.
Hopkins, the chief complainant, describes herself as a molecular
biologist, along with several other members of the comparison groups.
Citation counts for subareas tend to move up or down according
to the popularity of the area. This, in turn, often contributes
to market value. This removes part of the justification for
"controlling for relative citation counts" within a field.
A key empirical question relevant to MIT is whether, *in a faculty
composed entirely of men*, one group with several Nobel prize winners
and many times the citation count, twice the publication count,
and vastly higher general recognition would be expected
to have higher salary than another group with no Nobel prize
winners, equal seniority, but half the publication count
and about 20-25% of the citation count.
While we're at it, Thom, I want you to try a simple thought
experiment. Imagine the genders were reversed at MIT. Imagine
that the senior female biologists had higher salaries,
and included 2 Nobel prize winners and a third shortlisted person,
4.5 times the citations, twice the number of publications, and
vastly greater general public recognition, relative to
male biologists *of the same age and seniority*. Now, imagine
the MEN complained about their lower salaries. Are you seriously
suggesting they wouldn't be accused of sexism and jealousy? Are
you seriously suggesting you wouldn't be outraged at this apparent
failure to accept a huge performance differential? I cannot
predict your reaction, but I can confidently predict what the
reaction of the feminist community would be.
>
>> 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 meant that evaluation of gender discrimination should
include explicit attempts to evaluate performance and
factor out the influence of performance on salary.
Failure to do so penalizes men in many cases, since, in
some areas of science, they radically outperform women.
Not surprisingly, failure to evaluate performance is a staple
of nearly all feminist analyses of gender discrimination.
Implementation of performance evaluation will not necessarily be easy
or noncontroversial.
Note that the 9 signatories at the recent MIT conference adopted a
number of positions with no implementation details whatever.
>
>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.
There is no evidence that this happened at MIT. Indeed, the
MIT complainants expressed explicit annoyance about being left OFF
committees.
Defenders of reverse discrimination have countless theories
about how "gender discrimination" "might" be occurring to
inhibit the progress of women, but virtually no data to support any of
their conjectures. Indeed, often the wild conjectures involve the
notion that the phenomena are not amenable to empirical analysis.
What we have at MIT Biology is (a) evidence that the junior women
biology faculty do not believe they have experienced gender
discrimination (read the MIT paper and note how this inconvenient fact
is twisted), (b) evidence that the senior men outperform the senior
women by a huge margin by every objective criterion (Nobel prizes,
major prizes, grant money, publications, citations), and (c) evidence
that the senior women feel less powerful and influential than
the senior men. One obvious interpretation is that (c) is a
justifiable consequence of (b). It is difficult to verify
this, because MIT refuses to release any data.
You can, in interpreting (c), ignore (b). That
is what proponents of reverse discrimination want you to do.
However, this is not rational, nor is it in the best
interests of science.
In the long run, I am far less interested in the specific
events at MIT than I am in the notion that future complaints
of "gender discrimination" MUST include performance data
in the evaluation. Performance can be quantified. We do it
all the time at tenure and promotion meetings, grant evaluation
committees, etc. Yet many purported studies of "gender discrimination"
ignore performance.
Let me make one more point. Much has been made by several
contributors here of the notion that the huge performance
differences noted at MIT might be due to "random variation,"
and that this needs to be assessed. [The notion is, to anyone
familiar with the performance data, absurd and irrelevant.]
Now, suppose two groups of equally senior men and women show salary
differences and minimal performance differences. Suppose that every
time this happens, women complain and deans increase their salaries.
Is this fair?
Not necessarily, unless men have a similar right, invoke that
right, and achieve similar
results. What will happen is that, in faculties where men and women
perform equally, women, following the lead of their MIT compatriots,
will end up never earning less than men, but men will often earn less
than women. Strangely, discussions of "gender discrimination" seem to
ignore the notion that men can be discriminated against.
To return to your original point -- I agree that external criteria
establishing the "hugeness" of the MIT differences would strengthen
the argument. The more data, the better.
Jim Steiger
--------------
James H. Steiger, Professor
Dept. of Psychology
University of British Columbia
Vancouver, B.C., V6T 1Z4
-------------
Note: I urge all members of this list to read
the following and inform themselves carefully
of the truth about the MIT Report on the Status
of Women Faculty.
Patricia Hausman and James Steiger Article,
"Confession Without Guilt?" :
http://www.iwf.org/news/mitfinal.pdf
Judith Kleinfeld's Article Critiquing the MIT Report:
http://www.uaf.edu/northern/mitstudy/#note9back
Original MIT Report on the Status of Women Faculty:
http://mindit.netmind.com/proxy/http://web.mit.edu/fnl/
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
