Previous by Date Index by Date
Threaded Index
Next by Date

Previous by Thread Next by Thread

Statistical measures

Forwarded message:
Date: Mon, 06 Oct 1997 10:12:36 -0700
From: David Farrell <>
Subject: Statistical measures

This in response to the brief LibLicense exchange of some weeks ago
re. usage statistics.  We would welcome any responses to this report.

David Farrell
University of California, Berkeley

To:  Liblicense Reflector
From:  TAsk Force for Web-based Use Measures

As a follow-up to the comments recently posted here, we are posting the
following brief report of the work of our task force.  We formed at the
JSTOR meeting at 1997 ALA midwinter, Washington.  Our goal was to identify
statistical elements and models that could be used by JSTOR and other
information providers to develop their statistical reporting mechanisms. 

JSTOR is working with these guidelines and they have been discussed with
other providers.  Britannica Online and SilverPlatter are close to
releasing new versions of their statistical packages and several of our
suggestions are incorporated in those releases.  Both plan additional
improvements in the future and our task force has offered to act as "beta
test reactors" to ensure that we have products that meet our needs as well
as theirs. 

Any comments or suggestions on our work would be appreciated and may be
forwarded to:

Web Statistics Task Force 
David Farrell, Berkeley, Chair
Jim Mullins, Villanova
Dave Perkins, CSU-Northridge
Sue Phillips, Texas
Camille Wanat, Berkeley
Kristen Garlock, JSTOR, ex-officio



The Task Force will identify elements and develop models for measuring and
evaluating use of Web-based information resources, for managing
collections, services and budgets.

Specifically, the Task Force will

        --Identify units of measurement
        --Explore the capability of vendors and systems for Web-based
	  products to record and measure use
        --Devise analytical models and reports formats for evaluating and
          applying use measurements

The Task Force will work directly with JSTOR to assist in the development
of a use measurement and reporting capability for its archival database.
The scope of the Task Force's inquiry will also address other Web-based
information resources to which use measurement might be applied.  These may
include, for example, Encyclopedia Britannica and Encyclopedia Americana;
Project MUSE; SilverPlatter; GaleNet; Cambridge Scientific and society
publishers (IO, AIP, SIAM, etc.) The following are elements the TF would
like resource providers to measure and report, as appropriate to their
system hardware/software and the nature of their product(s).


1.  ABSTRACTING & INDEXING SERVICES (e.g., EconLit and the A&I portion of a
mixed database such as ABI/Inform)

        a. Number of queries
                By database
                By IP address/locator (to subnet level)
        b. Number of sessions ("logins"); Average length of sessions
        c. Number of turnaways
                Due to contract limits (e.g., requests exceed simultaneous
			user limit)
		Due to server limits (e.g., system overload or failure)
	 d. Number of citations retrieved: marked or selected, downloaded,
			emailed, printed
	 e. Usage levels, per time period
		Hours of use for a.-c.
                   By day, month, year
                   By time of day
		Peak and/or average amount of simultaneous use
		Total hours of server downtime by month

2.  FULLTEXT MONOS/SERIALS  (e.g., Reference works like Britannica Online,
		JSTOR, and the full text/image portions of mixed databases
		like ABI/Inform)

        a. Number of queries
		By database or title (of journal)
                By IP address/locator (to subnet level)
                By Browse and Search modes
        b.-c. as for 1, above
	d. Number of items retrieved: marked or selected, downloaded,
			emailed, printed
		Other (Image & AV files, ads, reviews, etc., as appropriate)
                Other (ads, reviews, etc.) (as appropriate)
        e. Usage levels, per time period
		Hours of use for a.-c.
                   By day, month, year
                   By time of day
		Peak and/or average amount of simultaneous use
       	  Total hours of server downtime by month



We will not request statistical information that reveals names or other
confidential information about our users; will confidentiality be protected
by keeping IP address monitoring to the subnet level?  Are there other
areas of potential violation?


We would like to see aggregated statistics that will give us a context in
which to analyze statistics for our own institutions.  The aggregate could
be compiled by the resource provider (e.g., stats from an anonymous
selection of institutions similar to our own), or it might be an aggregate
we would select (e.g., stats from all campuses in the UC consortium,
presented either anonymously or not, as desired by the participating


Preference for resource provider to maintain a website (updated monthly)
which our institutions can access on demand.  We don't want stats emailed
or available only on special request. Prefer that raw stat data be
presented in a tabular form that can be downloaded and run through a local
spreadsheet; plus a graphical display.
© 1996, 1997 Yale University Library
Please read our Disclaimer
E-mail us with feedback