College and Research Libraries


Letters 
To the Editor: 

The recent article by Professor Julie Hallmark entitled "Scientists' Access andRe-
trieval of References Cited in Their Recent Journal Articles" (College & Research Libraries, 
55 [May 1994]: 199-209) reported potentially interesting results, but her methodology 
may not give them adequate support. I believe that the following points should be 
clarified. 

Hallmark selected nineteen journals from which to take the sample of articles, 
journals recommended by scientists and chosen through independent evaluations; the 
evaluations are attributed to Katz's Magazines for Libraries, 1989, which is general and 
dated. The list of titles chosen is small and does not adequately represent the disciplines. 
For example, the list for Biology omits basic titles in genetics and molecular biology, 
that for Chemistry consists of only three ACS journals, and Physics is represented by 
only the parts of Physical Review. Also, why are Nature and Science excluded, but PNAS 
included? Were impact factors or times cited considered? The titles on the list may be 
prestigious, but the groups seem narrow for the disciplines they represent. 

The timing in this study is another concern. The sample letter is dated March 23, 1992, 
and the articles chosen for the sample were published during the last six months of 
1991. Does this mean that an author could have been asked about OJ;le reference in a· 
paper published nine months earlier thus written as much as twelve or eighteen months 
ago? How is the reliability of the author's recall taken into account? Another related 
concern is whether the scientists had cited the reference in. previous work, or if instead 
the citation was new to them. If it was previously cited, then it might be even more 
difficult to recall with accuracy how it was originally accessed or retrieved. 

Some further methodological concerns. The scientists were given choices about access 
and retrieval of their references. How were these derived? Were they a result of 
pretesting, or drawn from another reliable(?) instrument? Also, what value was there 
in prefixing the sample to contain 60 percent citations from 1980-1991,35 percent from 
1979 or earlier, and 5 percent foreign language? Hallmark has already restricted her 
research to cited journal articles; with monographs, conferences and patents excluded 
this formula is more restrictive still. On the practical side, if a citation to a 1978 paper 
in German was randomly chosen, how would this be categorized? And when the data 
for access and retrieval are tabulated, ranges of percentages reflect how these articles 
were discovered and obtained, but these describe the aggregate set of scientists, not the 
individuals. Some may find most references by talking to colleagues while others may 
not find any this way. In order to see patterns of individual use and discovery of 
references, one must ask scientists how they find several of their references. 

The identification of user needs is such a constantly popular and elusive topic that 
librarians have asked for better studies than have been done to date, studies with 
methodologies that yield valid and transferable results. This report suggests several 
ideas about scientists but these may not be justified. Because the author achieved both 
a high response rate and many comments, perhaps she could provide a more detailed 
interpretation of her data. 

CHRISTINE S. SHERRATT 
Assistant Science Librarian 
Massachusetts Institute of Technology 

85