From: halebsky@ssc.wisc.edu
Date: Thu, 8 Feb 1996 00:57:45 -0600
To: "socgrad@ucsd.edu"@ssc.wisc.edu
Subject: what has quant. meth. accomplished?
Michael Lichter asks, "What is a 'high-level statistical method'?
... Why does it matter?"

My response: High-level quantitative methods have either
(depending on your sentiment) come to dominate or are coming to
dominate American sociology.  Grants and awards are given,
taxpayer money is spent, articles are published in prestigious
journals (i.e, ASR), grad students are funded, courses are
required, jobs are obtained, etc. more readily if high-level
quantitative methods are involved.  Given this state of affairs I
think it is entirely appropriate to ask, "What has been
accomplished with such methods."

Michael goes on to accuse me of really wanting to "stop forcing
poor, innocent grad students to go through those horrible stat
classes."  Well, I am poor but not innocent!  More seriously,
though, I do admit that my enthusiasm for stats is minimal, and
one of the *reasons* my enthusiasm is low is precisely because I
don't see the payoff.  Hence my original question.  Perhaps there
were some results I just didn't know about.  Now, to be fair, I
do admit that high-level quantitative methods can be useful. 
However, their usefulness seems far out of balance with the
resources and awards associated with them.  

Pamela Paxton presents some interesting examples:

1) quantitative methods to measure democracy, which are then used
by USAid.  My response: Quantitative data can be generated from
anything, it's just a matter of giving something a number.  Once
you have quantitative data you can put it through the hoops of
high-powered stats, use up hundreds of hours of computer time,
etc.  The question is this: is there a crucial understanding
(leaving aside here and in the following examples the issue of
whether "understanding" really has much to do with solving social
problems) that could *only* have been arrived at through high-
level quantitative methods?  In this example, as in Pamela's
other examples, I believe the answer is no.  In this particular
case I would argue that a) deciding what constitutes "democracy"
is very qualitative, and b) having operationalized "democracy,"
the most important step is collecting data (that is, getting the
relevant information).

2) inequality in networks.  My response: Did it really take high-
level methods to figure out that disadvantaged people have
disadvantaged networks?   

3) "Fertility studies have also been helpful in moving some
developing countries toward lower birth rates."  My response: Do
developing countries really need exotic fertility studies (that
is, fertility studies involving sophisticated methods, extensive
computer manipulation, etc.) in order to move them toward lower
birth rates?

4) "Statistical information used in 'comparable worth' trials." 
My response: pretty much the same as 1) above: deciding whether
two jobs are comparable seems quite qualitative, and gathering
information (who gets paid what, what kind of work they actually
do, etc.) is more important than sophisticated manipulation of
that information.  


This will likely be my last post directly related to my original
question.  I really did want to know what quant. methods had
accomplished so I tried to avoid getting off on tangents. 
However, along the way there were a number of provocative
questions and issues which might profitably be pursued . . .

     * Dave Shafer asks, "What should the goal of sociologists
     be?" and says that he is "totally turned off" by the   
     mainstream journals.  Me too!

     * Michael thinks we should ask "Why should we bother
     studying sociology if it hasn't solved any of our collective
     problems?"  I agree--excellent question!


cheers,

Steve Halebsky
halebsky@ssc.wisc.edu