Evaluation and Appraisal of University Administrative

Computing Datasets

by Mark Conrad

'

Data Archivist

Pennsylvania Stale University

Introduction

In September of 1990 the Pennsylvania State University

Archives in cooperation with Management Services, a

division of the university responsible for administrative

computing services, began a two year grant project

funded by the National Historical Publications and

Records Commission (NHPRC Grant #90-095). The
objectives of this project are to appraise, preserve, and

make available electronic records created, stored and

used in the Management Services Division of Penn State

University; to develop ongoing procedures for the

appraisal of administrative computing data in the future;

to develop protocols for the use of the data by institu-

tional and outside researchers while enforcing restrictions

on access for privacy and confidentiality purposes; and to

provide recommendations based on the project for the

preservation of archival data from on-line administtative

database systems.

The project was divided into four phases. The first phase

lasting two months was used for the orientation of the

data archivist to the operations of the University Ar-

chives, the University Records Management Program,

and Management Services. Phase two, the phase we are

currently operating in, is used for the appraisal of

datasets. This phase is scheduled to last 18 months.

Phase three and four are scheduled to last the final four

months of the two years. In phase three recommenda-
tions will be made for the identification and preservation

of future archival datasets and protocols will be devel-

oped for the research use of the datasets. In phase four

reports on the project will be prepared and circulated. In

actuality, much of the work scheduled for phases three

and four has already begun and is being carried out

concurrently with the appraisal process.

This paper will focus on the second phase of the project

I will discuss the appraisal process as it is currently being

carried out, difficulties encountered and lessons learned

to date.

The Appraisal Process

For the purposes of this project the appraisal process will

be confined to a finite number of datasets from the

university's administrative computing mainframe. We
will not be examining electronic records from other

mainframe, mini or microcomputer systems.

Rather, the appraisal process will be limited to some
3,000 datasets recorded on "history" tapes. History files,

in Management Services' parlance, are usually copies of

master files at a particular point in time (often the end of

a semester or academic year). The datasets would be

very difficult to recreate if they were destroyed because

they are copied from files that are constantly updated.

These files are kept for possible reuse ^.

The datasets date from the late 1960's to the present.

These datasets contain information for fourteen areas of

administrative responsibility at the university. The areas

are: Accounting, Payroll, Bursar, Student Aid, Agricul-

ture, Planning and Analysis, Budget and Resource

Analysis, Management and Systems Engineering,

Admissions, RegisD-ar, Testing Services, Development,

Graduate School, and Physical Education.

Each area has a data steward who "develops the coding

structure of the data, insures the data's accuracy, deter-

mines the frequency of updating, and establishes data use

and protection requirements."' The data steward is

usually a senior administrator, such as the Registrar, or

that persons' designate.

The appraisal process begins by selecting a data stew-

ard's area of responsibility. The data archivist must have

the written permission of the data steward to examine the

datasets under his/her control. In some cases datasets are

jointly "owned" by more than one data steward. In such

instances the disposition of the datasets must be dis-

cussed with all interested parties. In starting the ap-

praisal process I chose data stewards that had only a few

history tapes to test the procedures developed for the

appraisal process and the database system to track

appraisal information.

Having chosen an area of responsibility, the next step is

to identify those datasets that belong to the data steward

from the (Management Services) Tape Library listing of

the history files. Once identified the datasets are grouped

by common dataset name. This is done because datasets

with the same name usually contain the same types of

data and can be initially evaluated as a group.

The next step is to locate as much information about the

datasets as possible. I am to find the procedure and

Summer 1991



program that created or used the information, any

documentation, a file description, a record description,

record counts, or any samples of input or output. Rec-

ords Management Program retention schedules are

checked to see if similar records in another format have

already been scheduled. If similar records have been

scheduled for destruction and the datasets do not have

some additional value by virtue of their being in elec-

tronic format and thus more manipulable, the datasets

may be recommended for destruction. (Junk is junk no

matter what the format.)

While the search is on for information about the datasets,

some of the tapes are read and printouts are made of a

sample of records from each file. This serves several

purposes. Firstly, it verifies whether or not the tapes are

still readable. Some of these tapes have been in storage

for a very long time under less than ideal environmental

conditions. Secondly, the dump can be used for compar-

ing what's actually on the tape with what the documenta-

tion says should be on the tape. If the two don't match,

other documentation must be located or the file may be

recommended for disposal. A file is of no value if a

determination cannot be made as to where one field ends

and the next begins or as to what a particular value in a

field indicates.

Having gathered as much information about the datasct

as possible, the next step is to interview the data steward

and/or a contact designated by the steward about the

datasets under his/her con&ol. A standard list of ques-

tions has been developed to help the data archivist gather

all the information necessary to make an informed

appraisal decision.

Those questions are:

Do you have documentation for these files?

Do you have samples of input and output for these

files?

Where did the data come from?

What was it used for?

Is it still being used?

Is it updated?

How often?

Are the records maintained in another format?

Has the other format been scheduled for retention or

disposal?

Are there any requirements for the retention of this

data that you are aware of?

Are there any restrictions on the use of this data that

you are aware of?

At this point the data archivist should have enough

information to begin making the appraisal decision. The

decision-making process is not that different from the

process for more traditional records. It is certainly not

very different from the process used by archivists

working in the electronic records programs in govern-

ment archives.

Does the dataset have legal, evidential, or

informational value?

Is this dataset unique?

Is it the most desirable format for keeping the

information?

Is the data hardware/software independent?

If the answer to all of these questions is yes, all datasets

that share the common dataset name and structure will be

recommended for accessioning by the Archives. Reten-

tion schedules arc developed for all datasets, regardless

of their status, in concert with the data steward. Manage-

ment Services, the Records Management Program Staff

and the University Archives/Records Management
Advisory Committee.

Once the decision has been made that a group of datasets

are archival, each dataset is read and a data dump is

obtained. As with the sample of datasets read previously

this is to verify that each dataset is readable and the data

is valid. File structures and record descriptions change

over time so the data archivist must insure that data from

each dataset is adequately documented so that a re-

searcher or the archives staff can use it. Assuming the

datasets are readable and understandable, copies are

made of each dataset and the relevant documentation.

The datasets are accessioned by the University Archives

and the data archivist turns his attention to the next set of

files.

Problems Encountered

You may have already gathered that one of the biggest

problems has been locating adequate documentation for

many of the datasets. The record description for many
files seems to change on a regular basis. Often the

documentation is not updated to reflect these changes or

conversely when the documentation is updated previous

versions are discarded despite the fact that files still exist

that were created using the previous documentation. It is

not unusual to find a number of files with different file

(ASSIST Quarterly



structures, the same name, and one set of documentation

that may not match any of the files. In talking with other

archivists working with electronic records, I have been

assured that Penn State is not alone in this predicament.

Management Services is currently exploring the possibil-

ity of recording documentation for a dataset directly onto

the first label of the tape a dataset is recorded on. As
long as the documentation is copied along with the

dataset whenever the dataset is transferred to new media,

the proper documentation should be available for the life

of the dataset

Another related problem occurs when trying to appraise

an older dataset Often there are no employees still

working in the office that used the file who remember
what the file was used for. Sometimes the office itself no

longer exists! The turnover on a university campus and

the restructuring of administrative units can make it

difficult to find someone who can tell you how a file was

originally used or what dataset replaced the one you are

evaluating. The only solution to this problem is to carry

out the appraisal early in the life cycle of a dataset.

Lessons Learned
One of the most important lessons we have learned is

that the shorter the time lapse is between the creation of a

dataset and its appraisal the easier it is to identify archi-

val datasets and insure their preservation. The archivist

can interview all the players involved in the creation of

the records to better understand why the records were

created and under what circumstances. Documentation

can be evaluated to insure it adequately explains the data

so that it will be useful to future researchers. Datasets

that are identified as archival can be marked for special

handling to insure the data will still be readable 10, 25, or

100 years from now.

Another lesson we have learned is that the cooperation of

the administrative computing center is essential to the

success of an electronic records program. The archivist

needs to understand how data is manipulated at the center

to meet the informational needs of the institution. The

administrative computing center personnel must have an

appreciation of the potential value of the data beyond the

purposes for which it was originally created. Any
archives considering implementation of an electronic

records program would be well advised to begin building

relationships with their institution's administrative

computing center(s) now.

The most important lesson we have learned is that more

records are being stored in electronic format all the time.

If we do not identify and preserve the archival datasets a

large portion of our institutional memory will be lost. At

Penn State we have begun the process of insuring these

valuable records will be preserved, we encourage other

institutions to join us, and we are happy to share informa-

tion about our project.

Footnotes

^Pennsylvania State University. Management Services

Division. Standards and Procedures Manuals (on-line

manual).

' Pennsylvania State University. Administrative Policy,

AD-23.

References

Pennsylvania State University, Management Services

Standards and Procedures Manual #3, Chapter 5, Section

1. (on-line manual)

Pennsylvania State University Administrative Policy,

AD-23, p. 1.

'Paper Presented at lASSIST 91, 15 May 1991, Edmon-
ton, Alberta, Canada.

Summer 1991