vol223


Fall 1998 9

The recent establishment of the California
Digital Library creates an unprecedented
opportunity to bring social science data to
users in a more user-friendly format as
well as making it available to a much
wider audience.

The California Digital Library was
established at the University of California
in the middle of last fall, in effect creating a tenth
campus library, this one entirely virtual.  In practice, it is
based at the UC Office of the President headquarters in
Oakland, California, it’s main manifestation in its initial
year appears to have been acquiring licensing agreements
with computerized databases of academic journals in the
fields of science and technology.  But with an anticipated
infusion of $3 million from state coffers this coming year,
plus another $1 million from the University of California
itself, the CDL, as it is called, will soon be looking for new
disciplines to cover, and new territory to conquer.
According to press reports, California has had a budget
surplus of $4.4 billion which can hopefully be partially
allotted to the libraries. 

Social Sciences are a likely future field of attention for the
CDL, and UC’s data archivists and librarians recently met
with the CDL’s newly installed Collections Officer to
discuss matters of mutual interest, including ways of
collaborating in possible future joint endeavors.

This paper, then, is part of a continuing, fairly new process
of rethinking how the UC system collects and provides
services to social science material in digitized formats.

Since the CDL bills itself as the 10th campus library
collecting everything in digitized form, there might have
been initial fears that collecting of digitized materials
would end on the individual campuses that make up the UC
system.  While legally and constitutionally the University
of California is one state-wide system, in actual
practice, there is almost no collection development (in
terms of library resources) at the system level, except for
licenses to databases negotiated system-wide and made
available via Melvyl, the UC catalog and database gateway;
and large purchases made in the past through “shared
purchases” system-wide, largely made up of big ticket
items and large microform sets. Each campus is strongly

independent, each headed by its own
chancellor and strong faculty senates;
campus libraries are no different, each
with its own collecting focus, depending
in a large part on the research
and instructional needs of the faculty on
the particular campus.

Despite being three years in the planning
stage,  the actual creation of the CDL may have caught
many campus librarians by surprise.  Any initial fears that
collecting of digitized materials on campus would
be displaced by the CDL collecting the same stuff for all
campuses soon gave way to the realization that there was
no way individual campus digitized collections would
disappear, nor would collection development cease.
In fact, since data archives and data collection development
vary across campuses, meeting individual needs not
necessarily duplicated elsewhere, the likely model that will
prevail is one more of collaboration with the CDL than one
of the CDL usurping and displacing campus
collection patterns. While it is relatively easy (and familiar)
for the CDL to negotiate licenses over journals in the social
sciences, it is a different kettle of fish for the CDL to enter
the field of collecting (and making available) raw data that
are the products of social science research.  It is not just a
matter of licensing, but also one of making it available to
users — and which users need to be defined — in an
appropriate format.

How the  CDL is structuring its own collection
development may be instructive here.  Currently, within its
science and technology collection focus, the CDL has
selected biotechnology and computer science as the two
areas where acquisition of digitized information in formats
other than electronic journals or monographs would take
place.  This is described in a March 8, 1998 announcement
on its listserv (CDLINFO-L) as  providing a “laboratory for
learning and planning for future CDL collections,” and for
“gradually [adding] other types of content” beyond
electronic journals or monographs. This is an ongoing
experiment, and it is too early to report any results. But
what is clear is that the CDL intends to acquire datasets
in biotechnology and computer science initially, although
what it would acquire in particular is not apparent.

In addition, the CDL plans to structure its collection in

The California Digital Library:
Implications for Social Science Data

Files Collections

by Daniel C. Tsang*



10 IASSIST Quarterly

three “tiers”, namely, Tier 1, material funded, in whole or
in part, by the CDL; Tier 2, material funded by two or more
campuses; and Tier 3, material that is paid for by only one
campus.  For the last tier, this implies the CDL would
provide a gateway, on its Web site apparently, to material
on that campus, even if, as is likely, only users affiliated
with that campus can use the material.  For tier 2, then,
material would likely not be accessible to those outside the
campuses that funded it; and for tier 1, since the
CDL funded the material, it would presumably be made
available to most or all campuses (a campus can opt out of
a particular acquisition).

Its current collection principles are that priority should be
given to digital format acquisition of those resources which
offer economies of scale by benefiting the most faculty
and/or students both locally and system wide. Also,
electronic materials should be selected based on increase of
access to the installed base of UC library collections
and build on the investments already made by the
university in digital resources.  If and when the CDL enters
the field of social sciences collecting and social science
data in particular, a major issue appears to me to be the one
of defining who would be the potential user population.  On
a traditional campus archive or data library, UC data
archivists and librarians have generally acquired material
only for the use of faculty and students on that campus.  In
addition, as I argued in an earlier IASSIST conference
paper1, collection development, in UC’s as elsewhere, has
generally, been reactive rather than proactive; i.e., datasets
are acquired in response to specific requests, rather than, as
with books, collected in advance of specific request, and
often without anticipation of actual immediate use  (e.g.,
through book approval plans).  Since that paper, however,
with the advent of the main source of social science data
now distributing, in effect, its entire newly acquired
collection every quarter or so in the from of Periodic
Release CD-ROMs, libraries are acquiring data without any
selectivity, in actual practice, and certainly not in response
to any specific request.  In addition, with the ICPSR’s
entire archive much easier to retrieve (with the demise of
distribution on round tapes), data archivists can potentially
replicate much of the collection on their own campuses,
although as of yet, there is no mirror site to the ICPSR
archive yet established.  Such a mirror site might be one
area the CDL might fruitfully consider, especially since its
mission appears not only to serve the campus-based
academic community but also the community at large,
although it is still unclear how that would work in practice.

The CDL, after all, is called the California Digital Library,
not the University of California Digital Library, strongly
implying it is collecting digital material to serve the entire
state, i.e., all the people of the state of California.  That was
the quid pro quo that apparently was necessary for the state
legislature to fund the CDL.  In announcing the creation of
the CDL last fall, UC President Richard C. Atkinson spoke

of creating “UC’s library without walls.”  It would be
a library allowing “scholars of all ages and interests to
range worldwide in their quest for knowledge, using the
Internet, the World Wide Web and a computer.”

If the target population is the scholarly community within
UC and beyond, that would be one thing.  Libraries are
used to collecting for scholars; the CDL could very well
mirror ICPSR’s archive (after becoming a full-
fledged member like the other UC’s that are members) by
paying them enough money so that they would not think
that they are losing money.  But since there is no physical
campus associated with the CDL, what does this mean?  Or
is the CDL membership to take the place of the individual
campus memberships? That does not seem likely, given
that existing campus archives and data collections have
constituencies they have nurtured and served for years; they
are not likely to disappear, at least not without a fight.  But
if the CDL promised access to all ICPSR data, and
provided front-end interfaces that facilitated the extraction
of variable-level data (it would need to write the software
etc.) of selected datasets, would that not make local
service points less necessary, or even impractical?  The
countervailing argument, of course, is that with all
secondary analysis of data, it is not sufficient merely to
make the data available; there has to be a variety
of services (metadata access; interpretation of metadata;
statistical consulting; etc.) that, because computer setups
vary across campuses, and even within, are best handled at
the local level.  

In addition, the CDL can perhaps better negotiate licenses
at the system-level with data vendors that provide data,
such as economic time series, of interest across the UC
system.

But given its mandate to serve the state, the CDL and its
pioneering vision, the CDL could very well do more than
just provide access, even improved access, to what
currently exists.  The CDL could well get involved in one
or more large-scale digitization projects, funded by industry
and government, to archive and make accessible datasets
previously not readily available, such as government data at
both the state, county and municipal  levels.  The challenge
here is for joint partnerships with communities local and
state; what is sorely needed is a collaborative effort to
make sure that government data does not go the way of the
main frame and that they are archived and preserved, and
eventually made accessible to users.

As the CDL develops, one potentially controversial area
involves intellectual property rights.  Who owns the
research that faculty have invested their time in?  The
University is now arguing that it does, and that faculty
who sign away rights to articles, for example, are just
making commercial journal publishers much more rich.
Administrators are now wondering why a university should



Fall 1998 11

have to pay exorbitant fees to access the journal output of
their own faculty, just because a commercial journal
published it. Richard Lucier, the CDL’s head, argues that
scholarly publishing must change, and that universities
must on their own, compete with the industry and ”publish”
on line the scholarly output.  

With data however, is it likely that individual faculty, even
at UC, will be willing to deposit their research data at the
CDL, if the University insists that it must?  The University
recently revised its policy on research; now, even if faculty
use a dataset gathered elsewhere for secondary analysis, it
must register with local Institutional Review Boards.  Local
boards could well insist that faculty deposit data thus
gathered (or originally collected data for that matter) with
the CDL, or the campus data archive.  Right now there is
no such provision or mandate, not least because researchers
are unlikely to be willing to part with their data, and there
is no common understanding of who owns what.  If
a faculty member leaves, he or she is allowed to take
research data he or she has gathered; thus far, I am not
aware that the university has insisted on ownership.  But a
case could well be made, given that every employee is,
upon hire, made to sign away most of his or her patent
rights to the University.

Involvement in large-scale projects is especially likely, and
necessary, if indeed, the user community stretches beyond
the scholarly community. If indeed the vision is to let
anyone access the information (and since the CDL is called
the California Digital Library, not the University
of California Digital Library, as originally envisioned), that
suggests that anyone, “of any age”, would have access to

the library.  That would well be a mammoth task, devising
a dataset (or more) that would be useful to such a mythical
user. 

There are many more issues one could raise, not least that
of digital archive maintenance, authentication, and dataset
updating, as well as version control.   At a minimum, the
CDL could provide a union list of what exists in existing
campus archives and collections, providing some
bibliographic control to an existing situation that is
anarchistic at best.  But I see it as doing more than that; it
can best make the use of data more appealing, by providing
the necessary tools to access data from the hard-to-find to
those most popularly requested.  As such the CDL can help
those of us in the data archive community by making, and
educating, more people to be sophisticated data users and
consumers.

In conclusion, the CDL is unlikely to be the sole digital
repository for California of social science data, but its
creation and expansion will likely spur collaborative
efforts with existing collections and archives as well as
create new ways to provide improved access to
these collections.

1.  Daniel C. Tsang, “Academic Libraries and Collection
Development of Nonbibliographic Machine Readable Data
Files,” IASSIST Quarterly, volume 12, number 3, Fall
1988, pp. 47-55.

* Paper  presented at the IASSIST Conference, May 21,
1998, at Yale University, New Haven, Connecticut.

http://www.yorku.ca/org/iassist/