---------- Forwarded message ----------
Date: Tue, 2 Jul 1996 11:47:03 -0230
From: Dorothy Milne <dmilne@MORGAN.UCS.MUN.CA>
Subject: Guidelines for Journal Usage

Al Henderson's comment on "all the trouble caused by usage
studies" (in his posting on June 25th) caught my eye and begs for a reply,
all the more because he has misquoted and misinterpreted the results
of a paper I published on the subject.  So, here goes.

 --- [ Al wrote ] -------------------------------------------------------
<Writing in the Journal of Academic Librarianship (5:66, May 1979), Melvin
<Voigt argued that research use cannot be determined by circulation
<statistics, particularly where open stacks permit patrons to stand among
<the shelves browsing, reading, checking citations, etc.

This may have been a fair comment in 1979, but there are a number of more
recent studies which have shown that when the total use of journals
is measured (browsing, reshelving), it correlates quite well with circulation
statistics.  It is far less clear that using reshelving statistics alone
would correlate well.  In the posting that elicited Al's reply, the
library had both reshelving and circulation data. A good first step
for them would be to see if the two sets of data correlated well. If
not, they should rely on the circulation data.

<A study by Dorothy
<Milne and Bill Tiffany (Serials Librarian 19,3/4, 1991) indicated that
<many patrons did not cooperate with their reshelving methodology.

First of all, we did not use a reshelving methodology.  We used a method
that asked patrons to tick a tag whenever they made any use of a journal
issue - any use that would otherwise lead them to order the article by
ILL. So - browsing an issue counted as a use, as did reading the article,
photocopying the article, or borrowing it.  What we found was that
users failed to tick about one-third of the time.  However, since there
is no reason to think that the ratio of ticking to not-ticking would
vary from one journal to another (i.e. that physicists failed to
tick more than historians), this problem was resolved by adjusting the
use figures upwards by 1.5.  Since the decision on which journals to
cancel depended on a listing of titles in order of the intensity of
their use, the method produced usable and repeatable results.

<The infamous Pitt study used a sample that indicated low usage of journals
<while interviews indicated that faculty members systematically browsed all
<new issues.
 A number of studies have shown that _actual use_ of journals by faculty
 members and the use that faculty members _say_ they make of journals are
 quite different.  Our own results showed almost no correlation between
 the two.  In my view, the only reliable results come from measuring
 actual use.

<There are also arithmetical complications. Suppose your dept. heads browse
<all major journals systematically. A weekly such as NEJM will show 50 uses
<per patron while monthlies show twelve and bimonthlies show six.

 In my view, the only meaningful basis for cancellation/retention of
 journals is the cost-per-use of a title, based on the total number of
 uses per year.  If one measures use, estimates the annual total of uses,
 and then calculates an estimate of the cost-per- use  (i.e. annual
 subscription cost / annual no. of uses), the distinction between
 weeklies, monthlies, quarterlies is taken into account.  This is
 neither complicated nor difficult to do.

<A "scientific" study of usage would probably use some other method to
<assure a given level of reliability.

 Our methodology was scrutinized by some very annoyed mathematicians and
 physicists (annoyed because their journals were slated for cancellation).
 They were itching to find deficiencies in the method.  They came back
 apologetic and said they found no methodological problem. (The method
 was devised by a Ph.D. scientist in the first place.)

 As for reliability, we have checked our results against ILL requests
 over the past 8 years and so far have found no major errors.  A
 recent collection evaluation based on a recent citation analysis in
 chemical instrument analysis offered an interesting confirmation that
 our cancellations (based on our use studies) have indeed targetted the low
 use titles and spared the high use titles.

 I would agree that no estimate of usage will ever be "scientifically
 accurate" - this sort of human behaviour is too difficult to analyse
 with total precision. This is why our method was based on "estimates"
 which then rank the journals in sequence. A better question, in my
 opinion, is what is a library's best approach to getting reliable
 information?  As far as I know, no better method has yet appeared in the
 library literature.

 The commonest errors that some cancellation projects make is to
 fail to correct for different lengths of journal runs, to fail
 to estimate use for a total year for all titles (to correct for
 the weekly/monthly/quarterly distinction), and to fail to
 calculate the cost-per-use.  Cost alone and number of uses alone
 do not yield results which make economic sense.

 The purpose of the cost-per-use estimate is to aid in the decision
 of whether it would be cheaper to supply the information by ILL
 or from an electronic source.

<Best wishes, Al Henderson, Editor, Publishing Research Quarterly
<70244.1532@compuserve.com

Publishers are not overjoyed to have libraries judge their journals
by the actual use they receive.  Publishers' focus on "quality" has
included paper and printing quality, binding quality, and quality
of an editorial board.  They have not, however, concentrated on
the "usability" or relevance of the information they publish.  I
sympathize with them, since this is a difficult thing to identify
and promote.  None the less, I get the impression that publishers
have focussed more on getting more _quantity_ (more articles, more
issues) rather than on keeping costs down by publishing articles
which the readership really wants to see.  Thus, their journals
are swollen with articles that will never be read, and the costs
of this system have spiralled out of sight.  There is little that
libraries can do in this environment than observe user behaviour
and act accordingly,

Cheers,

          Dorothy Milne

````````````````````````````````````````````````````````````````````````
  Dorothy Milne                      e-mail: dmilne@morgan.ucs.mun.ca
  Science Collections Librarian
  Memorial University,               voice:  [709]-737-8270
  St.John's, Nfld., Canada           fax:    [709]-737-2153
''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''