Sun, 7 Aug 1994 14:43:40 -0700 for <socgrad@UCSD.EDU>
 by KUHUB.CC.UKANS.EDU (PMDF V4.3-8 #5489)
Date: Sun, 07 Aug 1994 16:43:34 -0500 (CDT)
Date-warning: Date header was inserted by KUHUB.CC.UKANS.EDU
From: 1k1mgm@KUHUB.CC.UKANS.EDU (Christopher Gunn)
Subject: Re: tell me it's not true!
To: socgrad@UCSD.EDU
>I recently had a conversation with the PI on the grant that I work for
>about publishing the results of the research on the grant.  He told me
>that to get published in the "major" journals for ANY field (this would be
>Nat'l Journal of Public Health for him, but I'm in Soc so I'm asking if
>this is true of ASR or AJS), the results of the research MUST BE
>CONFIRMATORY!!  He said that unless you came up with a hypothesis,
>methodology, etc. that could be confirmed, you wouldn't get published, at
>least in those journals. 

I'm interested in this discussion but I'm not sure I understand it.
As you are using the word, is the opposite of 'confirmatory' to be
'exploratory' or 'negative'?  I think I'd tend to see the issue as a
sort of 2x2 table:

                            Positive Results    Negative Results
Exploratory                        1                   2
Confirmatory                       3                   4

Is your PI saying that (1) and (3) could be published, or (3) and (4),
or only (3)?

If the distinction is confirmatory-exploratory, I'd wonder if the
reported bias isn't one more of style than of substance, i.e., recasting
a 'see what I found!' approach into something more dignified. :-)

>Now this severly disturbed me.  I have always felt that research that 
>*eliminated* variables was just as important as the research that *found* 
>variables (i.e. combinations that "worked" with the dependent var).  

This is a problem throughout science.  There have been several
recent articles in _Science News_ about the issue of all the stuff
sitting in drawers because it isn't sexy enough to publish.  However,
asserting the truth of the null hypothesis is tricky in any discipline
and especially so in social science, where experimental controls
are looser in both the formal and operational sense.

It seems to me that there are a lot of possibilities for a
technological solution to this business, however.  University
libraries may never want to pay $1,600/year for _The Journal
of Failed Experiments_, but it wouldn't be that tricky to have
negative results on-line in Gopher-space for reference purposes.
Might not help people get tenure, but it would help the
various disciplines as entities.

Christopher Gunn                Molecular Graphics and Modeling Laboratory
1k1mgm@kuhub.cc.ukans.edu       University of Kansas
Phone: 913-864-4428 or -4495    Malott Hall
                                Lawrence, KS  66045