Is Taylor's "compromised need" pseudoscience?

If you've read my blog at all, you probably know I'm a Taylor (1962,
1968) href="http://ejournals.library.ualberta.ca/index.php/EBLIP/article/view/650">groupie.
In fact, in a recent  href="http://scienceblogs.com/christinaslisrant/2009/06/librarian_basics_the_reference.php">post
I talked about going from a visceral need to a compromised
need.  This is a central idea in library science. So when I
saw this article in my feeds today, I had to pounce on it:




title="ctx_ver=Z39.88-2004&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.jtitle=Journal+of+the+American+Society+for+Information+Science+and+Technology&rft_id=info%3Adoi%2F10.1002%2Fasi.21129&rfr_id=info%3Asid%2Fresearchblogging.org&rft.atitle=Compromised+need+and+the+label+effect%3A+An+examination+of+claims+and+evidence&rft.issn=15322882&rft.date=2009&rft.volume=&rft.issue=&rft.spage=1&rft.epage=6&rft.artnum=http%3A%2F%2Fdoi.wiley.com%2F10.1002%2Fasi.21129&rft.au=Nicolaisen%2C+J.&rfe_dat=bpr3.included=1;bpr3.tags=Research+%2F+Scholarship%2COther%2CInformation+Science%2C+Library+Science">Nicolaisen,
J. (in press). Compromised need and the label effect: An examination of
claims and evidence Journal
of the American Society for Information Science and Technology,
1-6 DOI: href="http://dx.doi.org/10.1002/asi.21129">10.1002/asi.21129




Let's look at this paper, its claims, and discuss it a bit, shall we?




As a reminder, compromised need is what comes out of the information
seeker's mouth or is typed by her hands when interfacing with an
information retrieval system (here, an information retrieval system can
have a librarian as the interface - and that librarian can be there in
person or connected via some electronic means - or can be a web search
page or research database search page, or even a book index).
 The idea is that what actually comes out might be very
different from the actual need because there are labeling problems, you
might not know what you need or how to describe what is
needed, and because you change what you say based on what interface
you've got and what you think
the system can do with your input (see for example, my comps reading
from Wolfram (2008) in which the searches were different for two
systems, with similar google boxes).



Nicolaisen starts by talking about the importance of this concept - the
compromised need - and how it wasn't really used for much until the
1980s, when researchers started to use cognitive and psychological
research in LIS.  Apparently though,  this theory has
never been validated as such and tested to see if it holds water.
 It's basically been taken at face value and reference
training for librarians has changed accordingly.  His point in
this article is to compare Taylor's claims to empirical
studies that track reference questions received to see if there is
support for "compromised need."



In describing the claims, I think Nicolaisen says some things that seem
obvious, but do not match with my experience or what I've seen
in articles on evaluating reference service in the public library. The
first of these is that this compromised thing makes sense for areas
outside of one's expertise but makes no sense for a known-item search
or someone with a "verificative need".  He says:

He seems very skeptical (the way this is written) and questions how
often this happens. But actually, there are many instances when this is
indeed the case.  For example, when the information need is on
a sensitive subject or if the patron doesn't have any faith that the
information system can respond to that request.  He lists a
pile of references in which this is taken as a given, and found none in
which this idea is questioned.  Indeed, in the literature
reviews everyone apparently relies on a study by Ingwersen that
essentially had a sample size of 2 - which is ok for qualitative work,
but it's not, by definition, generalizable.



When looking through the evidence provided in the studies he
reviewed, he found that only a very small proportion of the
questions required extensive interviews and likewise very few of the
questions changed from the initial question after the reference
interview. He ends the article by describing what's needed for
induction - going from some observations to a universal statement -
including large sample size, works in different settings, no
conflicting information.  Further he calls this pseudo
scientific because it faces unresolved problems and it is accepted
without question and testing (whoa... them's fightin' words)



I think Nicolaisen's sample of literature is weak, to be honest. Many,
many public libraries have evaluated reference service and time and
again they've shown that failure to do a proper reference interview
leads to poor results.  No interview -> few questions
are answered correctly -> there is something in the interview
process that allows the system to make a better match with the need
than without it. The patron could have perfectly specified the need -
but if it is not understood by the system, then they won't get the
answer.  How about studies of search engine logs?
 Clearly the needs are imperfectly specified because the
system returns documents matching the terms, and yet the user enters a
new search.



 I was one of the student investigators on a study by Kaske
and Arnold of virtual reference services.  We did a typical
Hernon and McClure study and it was the same old thing - librarians who
asked what we needed and checked back to see if what they gave us was
appropriate were the only ones who successfully answered the question.
 It's not that we weren't saying what we needed, it's that it
couldn't be interpreted correctly most of the time in isolation. Just
because interviews were not performed does not mean that the questions
did not require an interview! It is often the case that someone will
come in for a specific book, but really have a much bigger problem, and
that you can only address this bigger issue through an interview.



Of course my experience and the myriad studies in the state of Maryland
do not contradict what Nicolaisen found in the studies he looked at;
however, I think he picked the wrong studies. He was only looking for
studies that specifically looked at before and after statements of
information need. It's not that it's not studied and questioned so it's
pseudoscience, it's that it is part of all of the studies that we do in
certain areas of our field.  



My arguments are somewhat confused, but basically:

1) studies showing importance of reference interviews to answering
patrons' questions are relevant to this topic

2) reports that interviews aren't required do not say if/how the
patron's actual problems were solved or if the patrons were satisfied
with the service

3) people put 2-3 words into a search engine - that's it - there's no
way that can perfectly specify their information need



And I'm going to stop writing now, as I stopped saying anything new a
while ago!