RE: Calculating the Cost per Article in the Current Subscription Model

[Phil Davis <pmd8@cornell.edu>]

Calculating the Cost per Article in the Current Subscription Model -- A
guide to the perplexed

For public distribution of my spreadsheet model, I was perhaps too scant
in annotation and description.  Below I will describe and provide more
justification for the data elements used in this spreadsheet.

Total FTE

This variable is not used in the calculation of cost per article -- it is
only used to provide an independent variable for graphing purposes,
allowing one to see patterns in the data.  For example, how do larger ARL
institutions compare to smaller ones?

2003 Serials Expenditure

This figure comes directly from ARL for the 2002-2003 academic year.  It
is used to estimate the amount of funds that are dedicated to purchasing
serials.  Serials include journals, databases, magazines, encyclopedias,
newspapers, etc.

2003 Article Hits

This is the number of article hits in ISI Web of Science for each ARL
institution using the corporate author search string listed in column J
(ISI Institution Search Notes).  Errors in ISI indexing will underestimate
this number, and hence overestimate the cost per article.

2003 Serials Exp. Devoted to Scholarly Journals

This figure is estimated by multiplying the 2003 Serials Expenditure with
Assumption 1.  It reduces the serials expenditure figure by assuming that
not all serials funds go to purchasing scholarly journals.  A line-item
case study at Cornell has indicated that approximately 50% of our serials
budget go to purchasing journals.

Estimated Scholarly articles published

Since ISI does not index the universe of published articles, this variable
estimates the real number of published articles by multiplying 2003
Article Hits by assumption 2 (an estimate of what percentage of the
article universe is indexed by ISI).  The default figure is 92% which is
generated by an idealized model [1] and may be artificially high,
resulting in a higher per-article cost.  Users are encouraged to reduce
this figure and see what happens.

Estimated Scholarly articles adjusted by first-author

Many articles have more than one author residing at more than one
institution.  In order not to double-count articles for more than one
institution, we adjusted the number of Article Hits by assumption 3.  
This figure (61%) was established by doing a sample of Cornell-authored
hits to see if a Cornellian was the first author.  In other words, of 100
article hits that come up in ISI for "Cornell", 61 of them had a
Cornellian as the first author.  This figure is assumed to be
generalizable to other ARL institutions.

Cost per Article Published

This figure is simply the Serials Expenditures Devoted to Scholarly
Journals divided by the Estimated Scholarly articles adjusted by
first-author.  In other words, the adjusted estimated number of articles
being published by your institution divided by the adjusted estimated
amount of money your library spends on scholarly journals.

Clear as mud?  Thought so...

Notes:

[1] Estimating the Percentage of Article Indexing for ISI Web of Science

ISI Web of Science indexes a total of 8,769 journals (SCI + SSCI +AHCI).  
If we assume 20,000 scholarly journals, we can state that ISI only indexes
44% of scholarly journals.  But this is not the same as saying that they
index 44% of the article literature since some journals are much more
prolific than others (i.e. compare the number of articles published in a
weekly science journal with a quarterly humanities journal).  If we assume
that ISI indexes the more prolific journals (in rank order) and use a
logarithmic model, we find that the nearly 9,000 "core journals" they
track index an amazing 92% of the article literature (log 8,769 / log
20,000).

However, we all anecdotally know this figure to be much too high, and I
haven't received anything more accurate from the staff at ISI.  Lowering
this indexing percentage down to say 70% increases the estimated number of
articles produced by an institution and thus lowers the cost per article
in the subscription model.

.

Philip Davis, Life Sciences Bibliographer
Mann Library, Cornell University, Ithaca, NY 14853
(607) 255-7192 ;  (607) 255-0318 fax
pmd8@cornell.edu
http://people.cornell.edu/pages/pmd8/