As already mentioned, the concept of
data integrity has a simple definition (it refers to the
accuracy or correctness of the data in a database), but
as a goal it has been formidable to achieve. In this
section of the study, we will mention some of the
technological innovations that the discipline of computer
science has devised for achieving and maintaining data
integrity. Our purpose in mentioning these technological
innovations (which are well known to computer scientists
and database professionals, but not to other audiences)
is not to provide an exhaustive review of them
(such coverage is available in standard database
reference texts, such as Date, 1995; Elmasri &
Navathe, 1994; and McFadden & Hoffer, 1991), but
rather to use them to point our attention to the
practical problems that create the need for procedures
for achieving and maintaining data integrity in the first
place.
First, data-management systems constitute a
major technological innovation in themselves. To bring
out their significance, we consider the seemingly
straightforward operations of inserting data into a
database, deleting data from a database, and updating
data in a database. The straightforwardness is illusory.
In practice, depending on how a computerized
data-management system is set up, there can be the
problem of deletion anomalies (where deleting one item in
the database unintentionally deletes other still-needed
items), the problem of update anomalies (where the
modification of data already in a database leads to
inconsistent values for the same item in different parts
of the database), and the problem of insertion anomalies
(where data for some business transactions actually
cannot even be inserted into the database). (For concrete
examples of these problems, see McFadden & Hoffer,
pp. 218-219). In the context of everyday operations in
business organizations, these anomalies were (and
continue to be) so extensive that they called for the
innovation of systems dedicated solely to data
management. The function of these data-management systems
is not to process data (for instance, applying
arithmetic operations to data in order to generate an
invoice), but instead is limited to the inserting,
deleting, and updating of data in a database. Today, this
innovation is reflected in the division of labor between
computer programmers (whose work is commonly associated
with third-generation procedural languages such as COBOL,
FORTRAN, and C) and database analysts, designers, and
operators (who typically perform their work with
commercial database software packages such as ORACLE,
DB2, IMS, IDMS, and Access). In this division of labor,
the intention is for the latter to keep and maintain the
data to be processed by the former. Before the innovation
of data-management systems, a computer program itself
(such as a COBOL program) was responsible for (what would
today be the data-management systems tasks of)
defining data structures and managing the operations of
inserting, deleting, and updating data the very
same data that the computer program itself was also
processing; the multiplicity of computer programs in a
business organization, of course, typically led to a
multiplicity of databases and different (often
incompatible and unshareable) arrangements by which data
were stored and maintained, which in turn destroyed data
integrity. The division of labor between data processing
(now the province of third-generation procedural
languages) and data management (now the province of
data-management systems) has been an innovation that has
significantly improved this situation and is now so
widespread that it is taken for granted.
Given that there is this division of labor,
there remains the practical problem of how (on the
data-management side of the division of labor)
insertions, deletions, and updates can proceed without
damage to the accuracy or correctness of the database.
Here, some major technological innovations contributed by
computer science are three major ways of designing a
database. Hierarchical, network, and relational
architectures have different ways of categorizing and
arranging data, where data-management systems based on a
relational architecture are today the most popular and
therefore, arguably, the most important. Relational
database management systems have, in turn, benefited from
technological innovations targeted at them, such as
mathematically proven forms of particular designs for
enforcing proper storage and maintenance of data
(commonly known as "first normal form,"
"second normal form," "Boyce/Codd normal
form," etc.). These "normal forms" and
"normalized databases" pertain to particular
ways of categorizing or rearranging a business
organizations database into an interrelated set of
two-dimensional tables so as to protect data integrity in
the face of daily insertion, deletion, and updating
operations.
A third technological innovation also contributing to
data integrity is entity-relationship diagrams,
sometimes called semantic models (Date, p. 347).
Here, the motivating problem is the practical one of how
a database analyst can systematically diagnose complex,
large-scale business situations so as to identify all of
the attributes of the situation that an accurate or
correct database would need to include data on.
Entity-relationship diagrams were innovated to address
this problem. This type of modeling or diagramming is
accompanied by sophisticated rules for how to depict
business "entities," their
"attributes," and the "relationships"
among the entities. For database designers,
entity-relationship diagrams and the sophisticated rules
governing them serve to facilitate the task of diagnosing
or analyzing complex, large-scale business situations,
where the resulting blueprint would be useful for setting
up a database using a relational, hierarchical, or
network architecture.
Fourth, there is the technological innovation of business
rules, sometimes called integrity rules or integrity
constraints. Whereas there are rules that govern, for
instance, what a correct relational database design would
be or what a correct entity-relationship diagram would
be, there are also rules that pertain to the correctness
of stored data values in reflecting the actual business
situation that concerns a given data-management system.
Typically, business rules involve "data
dependencies," which are relationships among the
data items themselves. For instance, a business rule
could state that the numerical value for the salary of an
employee must fall within a specific range, where the
minimum and maximum of the range depends on the code
indicating the employees job classification. In
this way, business rules provide a means of enforcing
conditions pertaining to the real-world meaning of the
data in a data-management system. Whereas the rules of
entity-relationship diagramming would pertain to all
situations for which such diagrams are made and whereas
the rules of relational database architecture would
pertain to all situations for which such an architecture
is used, business rules are specific to just the actual
business situation that a given database represents. The ad
hoc nature of business rules therefore might appear
to pose a difficulty to their systematic implementation;
however, the feature of automated business-rule
enforcement has been successfully innovated by computer
science and is widely available in all commercial
database softwares.
There are more technological innovations than just the
four, mentioned above, that the computer-science
discipline has contributed to meeting the challenge of
achieving and maintaining data integrity. However, these
four are sufficient to address of purpose of revealing,
first, what the practical problems of data integrity
involve and, second, how difficult and challenging the
goal of data integrity has been to achieve. To review,
some of the major facets of the challenge are that (1)
every instance of inserting, deleting, and updating data
in a database has the potential to threaten its accuracy
or correctness, (2) data integrity can be more easily
jeopardized if data management is not separated from data
processing, (3) the complexity of a business situation
reproduces itself in the complexity of the task of
identifying all of the attributes that an accurate or
correct database would need to include data on, and (4)
the numerical and other values stored in a database must
not violate the real-world meanings of the real-world
entities and relationships that these data values are
describing.
The discussion, so far, has been about the design and
operation of a single database. The management of
data in interrelated multiple databases raises an
additional level of problems. To address these problems,
there is what a computer-science approach would consider
to be the mandatory "enterprise data model"
(McFadden & Hoffer, p. 37, Zachman, 1987). It is, in
essence, an organization-wide entity-relationship model,
where the emphasis is on the organization-wide data
relationships rather than the details specific to any
single location or department within the organization.
Certainly, local-level entity-relationship diagrams for
individual departments within a business organization can
always be made, and localized data-management systems
could then be built, based on these localized blueprints.
However, when such localized data-management systems
exist, data items often turn out to be defined in ways
that are incompatible or unshareable across the different
localized systems. The purpose of enterprise data models
is to prevent or remedy this problem. In practice,
however, there is little known evidence of corporations
that have been able (whether owing to technological or
organizational obstacles) to draw up and implement
enterprise data models. This problem is magnified
dramatically when there is a merger of different business
organizations (such as one banks merger with
another bank), resulting in the presence of databases
whose respective designs were developed in completely
independent fashions. A major challenge that the
discipline of computer science has been tackling is how,
whether within an organization or across organizations,
the data from one database can be accurately mapped to
another database (cf. Wybo, 1994, Wybo & Lee, 1994).
In summary, the goal of achieving and maintaining the
integrity of a database has been a major challenge.
Computer science is to be credited for the major
technological innovations it has contributed towards
achieving this goal. Because data integrity remains a
goal difficult to achieve, and because information
systems (including data-management systems) also have
organizational components and not just technological
ones, there is a place for an organization-science
approach to data integrity in addition to the existing
computer-science approach. In the next section of this
study, we explain the particular organization-science
approach we will be taking.
Organization science is diverse,
including positivist methods as well as interpretive
ones. The former take a natural-science approach to
research, while the latter delve into the rich
understandings that human beings in everyday life have of
themselves and their organizational settings. The two
schools of thought are sometimes believed to be mutually
exclusive; however, Lee (1991) has made the case that the
two schools can each contribute to the same research
stream. Of the many different types of
organization-science approaches available, we choose
hermeneutics as one that is suitable to our
investigation.
We offer three reasons for this choice. First,
hermeneutics has already developed a track record as a
viable and productive research approach in organization
science (Lee, 1991; Phillips & Brown, 1993) and other
management-related disciplines, including marketing
(Arnold & Fischer, 1994; Thompson, Locander &
Pollio, 1990), accounting (Boland, 1993), and information
systems (Davis et al., 1992; Lee, 1994; Myers,
1994; Winograd & Flores 1986). Second, because it is
interpretive, hermeneutics would be attractive for,
first, drawing attention to the role of people and their
procedures in a data-management system and, second,
complementing the positivist computer-science focus on
hardware, software, and database design. Third, and most
important, hermeneutics is directly relevant to the task
of achieving and maintaining data integrity; the data in
a database, whether numbers or words and whether printed
on a piece of paper or flashed on a computer screen,
constitute the very subject matter text
that hermeneutics has always considered to be its domain
of study.
Hermeneutics arose as the study of the interpretation
of texts, such as the Bible and the ancient Greek
classics. Some of the questions that motivate
text-interpretation as a field of study in itself are
(Tice & Slavens, 1983, p. 293):
how are we to grasp the
meaning of each others expressions and actions
once they are recorded or to the extent that we do
not have the advantage of dialogue? Supposing,
further, that the items in question are ancient texts
or texts from another language and culture. By what
principles are we to grasp their meaning?
Furthermore, scholars have broadened the hermeneutical
conception of what can be considered a text. Hermeneutic
scholarship had originally conceived of text as referring
only to the words, numbers, and other symbols that human
beings write, print, or otherwise inscribe, but
contemporary hermeneutical scholars have extended the
conception of text to include individual actions, group
behaviors, and even social institutions, all of which, as
"text analogues," have meanings that can be
"read" and, as in the case of any text, be
interpreted (Ricoeur, 1981; Taylor, 1979). Hermeneutics
therefore qualifies as a suitable and promising approach
for studying not only how people handle the data (that
is, the text) in a data-management system, but also how
people understand themselves, their respective actions,
their collective behaviors, and even their organization
(all of which would be "text analogues").
In the subsection that immediately follows, we draw on
a distillation of hermeneutic concepts whose
effectiveness in research on information systems has
already been demonstrated. Whereas we will use the same
five hermeneutic concepts that Lee (1994) distilled from
the work of Ricoeur and others distanciation,
autonomization, social construction, appropriation, and
enactment we will additionally elaborate on the
concept of enactment. In the subsection after that, we
will reprise the value that the organization-science
perspective of hermeneutics has for research on data
management.
2.1 Five hermeneutic concepts
Where his interest rests in examining the richness
possible in communication that uses electronic mail, Lee
(1994) explains hermeneutic interpretation as involving
five concepts (distanciation, autonomization, social
construction, appropriation, and enactment) that he
distills from the work of others (emphasis is added):
- In explaining how
the interpretation of text occurs, Ricoeur (1981)
argues that it is unnecessary to refer to the
authors intentions, the texts
originally intended audience, or the texts
originating culture. A document typically becomes
separated from its author, its originally
intended audience, and its originating culture,
whereupon we can say that the text takes on a
life of its own. To illustrate Ricoeurs
position, the Bible offers an example. For
centuries, people have read the Bible and found
it meaningful, even when they have had no
knowledge of its authors, original audiences, or
originating cultures. Another example is computer
output.
Bolands 1991 study shows that
even computer-generated numeric documents, though
authorless, can be rich in meaning to its
readers. Distanciation refers to the
separation, in time and distance, that occurs
between a text and its author, its originally
intended audience, and/or its originating culture
and society. Autonomization refers to the
texts taking on a life of its own, despite
the distanciation. "By appropriation
Ricoeur emphasizes that interpretation is
to make ones own what was initially
alien as it actualises the
meaning of the text for the present reader"
(Boland, 1991, p. 445, emphasis added). Along
these lines, Ricoeur adds, "the intended
meaning of the text is not essentially the
presumed intention of the author..., but rather
what the text means for whoever complies with its
injunction" (p. 161).
-
- Thus, the meaning
of a text can refer, but is not restricted, to
what it means to its author or what the author
had in mind. There is, in addition, what Ricoeur
calls the texts non-ostensive reference
or what I call the already existing social
construction, socially constructed reality,
or world, which the meaning of the text is a
manifestation or artifact of. Because the term
"socially constructed" has become
commonly used (in my view, incorrectly) as a
synonym for "subjective" in its
pejorative sense, my meaning for this term
requires clarification. For this purpose, a good
illustration of a socially constructed reality is
Euclidean geometry.
-
- Euclidean geometry
does not exist in the physical world of nature.
It is, strictly speaking, a fiction. People who
carry knowledge of it can come and go, but the
object called Euclidean geometry persists. Like a
physical object, it can retain the same form
across the different individuals who encounter
it. In this sense, it is objective, not
subjective. It is an objective, socially
constructed reality.
- Furthermore,
suppose I am reading a Euclidean research paper.
Its meaning is not restricted to the Euclidean
argument that its author is making, but also
involves the entire socially constructed
apparatus that comprises Euclidean geometry
its axioms, theorems, symbols, and logic,
all of which transcend what is in the paper and
all of which were in existence prior to the
writing of the paper. The paper itself is just
one possible manifestation or artifact of this
apparatus. Upon grasping this socially
constructed reality (the papers
non-ostensive reference), I become Euclidean
myself and I become able to identify
inconsistencies and suggest improvements in the
paper. I could thereby transcend the
authors own understanding and even get to
understand the author better than the author
understood himself
This refers back to
Ricoeurs notion of appropriation,
which is taken a step further: not only can I
appropriate the text or what the author had in
mind, but also, the text and the socially
constructed world behind it can appropriate me.
Upon this appropriation, I am no longer an
independent individual exercising free will (if
indeed I ever was), but I become an agent of the
socially constructed world of Euclidean geometry.
Other individuals, placed in the same situation
as I, would be likewise appropriated and
transformed. We could then (and only then) engage
one another in a dialogue about this Euclidean
research paper, collectively enact some
meanings for it, and after some time,
effort, and perhaps struggle reach a
shared understanding of it. Moreover, rather than
being agents under the total control of the
socially constructed Euclidean world that has
appropriated and transformed us, we could still,
as a group, eventually render change in this
world, perhaps by contributing to its collection
of theorems or by offering resolutions to logical
inconsistencies or gaps that we identify in it.
-
- Other examples of socially
constructed worlds standing behind a text are
the world of science standing behind an article
in the MIS Quarterly, the world of
American civilization standing behind an article
in Newsweek, and the world of spirituality
standing behind the Bible. Even when different
readers bring different socially constructed
worlds to the same text, the text becomes
meaningful for a reader only with one or another
world (non-ostensive reference) standing behind
it.
-
- Most pertinent of
all to this articles topic e-mail
there is the world of the organization
standing behind any e-mail communication. An
organization is a social construction with an
existence independent of the people who, at the
moment, are populating it and the buildings that
are housing it. The socially constructed world
of an organization is what would persist even
if there were complete turnover in its personnel
and even if it were to relocate to different
buildings in a different city. And just as the
meaning of a text can include but also go beyond
whet its author had in mind, the meaning of an
e-mail communication can include but also go
beyond what its sender had in mind. Consider an
instance in which I, as a manager, am reading
some e-mail from a subordinate about a problem
involving some office politics. By grasping the
already existing, socially constructed world of
the organization standing behind this e-mail
communication or rather, by letting
this world appropriate me as I am reading the
e-mail I could even get to understand the
problem better than the e-mails author
understands it. Upon this appropriation, I can
perform what Weick (1969) would call my enactment
of meaning for the e-mail
Just as Lee then proceeded with his argument by using
the hermeneutic concepts of distanciation,
autonomization, social construction, appropriation, and
enactment to show how, in his case study, richness can be
achieved in communication that uses electronic mail, we
will proceed with our argument by using the same five
concepts to show how, in our case study, data integrity
can be achieved by a human being playing the role of a
key component in a data-management system.
Of these five hermeneutic concepts, we will elaborate
on enactment. First, in his description, Lee provided
only a brief description of it, concentrating instead on
social construction. Second, with regard to the topic of
this study data integrity enactment
provides the occasion during which data integrity, as the
result of human action, may take place. Third, enactment
turned out to have a prominent role in the situation that
we observed for our case study. An elaboration of the
concept of enactment is available from Weick, who states
(1969, p. 27, emphasis in the original): "actors themselves
create the environment to which they adapt." In his
terminology describing the environment of an
organization, Weick explains that he prefers the term
"enacted environment" to "external
environment":
Instead of discussing the
"external environment," we will discuss the
"enacted environment." The phrase
"enacted environment" preserves the crucial
distinctions that we wish to make, the most important
being that the human creates the environment to which
the system then adapts. The human actor does not
react to an environment, he enacts it. It is this
enacted environment, and nothing else, that is worked
upon by the processes of organizing.
The terminological distinction between "external
environment" and "enacted environment" is
significant. "External environment" refers to
the objective, socially constructed reality and the
objective physical reality, both of which a human being
encounters as "given." Still, the same
objective reality can be interpreted by different people
in different ways, where the meaning that a person
constitutes for it would be this persons
"enacted environment." (Weick states [p. 70]:
"enactment refers to the constituting of an
environment by actors.") The term "enacted
environment," compared to "external
environment," gives recognition, first, to the
central role of the human being (an enacted environment
must be enacted by someone) and, second, to the
consequent need for outside observers (such as
researchers interested in data management) to interpret
the observed human beings perspective in order to
account for the environment that this person is enacting.
2.2 The additional value of an
interpretive, organization-science perspective for
research on data integrity
Before we present our empirical investigation, some
additional remarks on our hermeneutic approach are in
order.
Explicit in the interpretive, organization-science
approach of hermeneutics is the notion that text does not
function like a conduit through which meaning would
somehow be transported from the texts author to the
texts readers. Rather than being a passive
recipient of whatever it is that a data-management system
delivers, the reader of the data is an active producer of
meaning. Meaning is not "in" the symbols that
make up numbers or words, but instead is a result of what
the numbers or words signify in the persons enacted
environment. In other words, the data of a database
(whether numbers or text and whether printed on a page or
flashed on a computer screen) might very well be
incorrect and inaccurate; however, through the way
that a skilled manager reads the data and enacts her
environment, the data can come to acquire integrity.
Given this way of conceptualizing data and people, we
see that the interpretive organization-science approach
of hermeneutics is valuable in that it draws attention to
previously unconsidered factors people and their
procedures with which a data-management system can
improve or achieve data integrity. This is new territory
waiting to be explored, compared to the earlier factors
hardware, software, and database design
already well researched by the computer-science approach.
A hermeneutic approach to research on data management
is neither unique nor unrecognized. Taking a hermeneutic
approach, Hirschheim, Klein, and Lyytinen (1995) have
already advanced what they call a "rule based school" for data
modeling. In this study, we complement their work; we
observe a manager, Carmen, and we interpret from her own
perspective how she brings integrity to the data in a
data-management system.
In this section of our study, we
present evidence demonstrating that a human being is not
a passive recipient of the data that a data-management
system delivers to her, but rather, is capable of the
task of acting on the data so as to contribute
significantly to the goal of data integrity. After
presenting a description of the setting, we present
evidence from three situations in which a manager brings
integrity to the data with which she works.
Urban Midsized University (UMU) offers programs in the
liberal arts, sciences, and professions. UMU has
approximately 30 years of experience with computerized
administrative systems. The student accounts office is
part of the general university accounting office. It is
responsible for student accounts, including assessing
student tuition, insurance, and user fees as well as
crediting student accounts through individual payments,
grants, fee waivers, and various other payments to
students.
The university fee structure is determined by local
government regulation and varies with academic program,
class standing, and immigration and citizenship status.
Eligibility for and amounts of various student financial
assistance are also determined by these factors. Although
rules governing tuition and fees are relatively stable,
rules governing financial assistance can change quite
often. In addition, students drop or add courses, change
immigration status, and incur various fees and fines on a
regular basis, all of which can affect the amount of
financial assistance for which they are eligible.
In addition to determining the current balance of
student accounts, the student accounts office must ensure
that fees are correctly charged and that any changes in
student status or enrollment are reflected in these
balances. To do so, managers in the student accounts
office work closely with the university registrars
office and rely heavily on data from both accounting and
registration systems to perform their job.
Various software, computerized files, manual
databases, and automated and manual procedures make up
the offices "system" for processing
accounts. One major problem experienced by this system is
the minimal level of computerized support for enforcement
of business rules among the data. In all business rules
(as defined above, in section 2), the permissible values
for a data item in the database are dependent on the
values of other data items in the database. (Numerous
other problems also exist, such as the inability to
reconstruct account histories and the absence of any
explicit entity-relationship model. In this study, we
focus on the data-integrity problems related to business
rules.)
One example of a business rule would be "the
salary of an employee must not be greater than the salary
of the manager of the department that the employee works
for" (Elmasri, 1994, p. 640). In this case, the
validity of a particular value for the attribute salary
of an employee would be dependent on the value of the
attribute salary of that employees manager. With
regard to the UMU student accounts office, one important
business rule (whose enforcement is not computerized) is
"the value of a partial tuition waiver for a student
who is a citizen or resident alien must be zero,"
because government policy designates these waivers for
international students only. In this case, the values
allowed for the tuition attribute of a student are
dependent on the value of the students citizenship
attribute (which itself can take on the values of
citizen, resident alien, or international student visa
and which can change over the course of the
students studies).
There are numerous other examples of business rules in
the data of the student accounts office. For instance,
tuition fees and eligibility for various forms of
financial assistance are dependent on a number of student
characteristics and their interactions. The consequences
of not respecting these dependencies include
unintentional over- or under-billing of student accounts,
disbursing undeserved fee reimbursements, and improperly
applied fines and service fees. Improperly disbursed
financial assistance is ultimately charged to the budgets
of departments or individual researchers by external
funding sources.
Because the design of the computerized portion of the
systems used by the student accounts office does not
enforce these business rules, one group of users can
proceed unaware of the implications or effects of its own
actions on other groups. For example, a department may
award a student a full tuition grant while the
fellowships office may award the same student a partial
tuition waiver. The student would then have an undeserved
credit in his or her account. The error could be
corrected by reducing the departmental grant by the
amount of the tuition waiver, but sometimes this
correction does not occur until the student claims his or
her credit. Later, another university office could
process a change in the students immigration
status, making the student ineligible for the tuition
waiver. The next round of corrections could result in the
removal of the tuition waiver and, therefore, a negative
balance in the students account, which in turn
could result in the students being assessed a late
payment fee.
The current system provides little computerized
support for enforcing these dependencies among attribute
values. Support that is provided is in the form of
supplemental batch mode COBOL programs, whose execution
must be individually requested and that consider only
some specific part of the total set of dependency
relationships. Thus, erroneous or invalid data can be
entered into the system and not detected until the
execution of one of these validation programs is
requested. Requests for execution of these programs are
generally motivated by incorrect student account
statements or student attempts to claim reimbursements.
In summary, there are major problems in achieving
and maintaining integrity in the database of the student
accounts office. Although some data standards exist
(student and course numbers) so that the identities of
certain entities are consistent across all systems, no
automated enforcement of business rules is provided.
UMUs central administration is considering
replacing the entire student records system with a new
"integrated" system at a projected cost of over
$10 million and six to ten years.
We interviewed a manager, Carmen, in the student
accounts office at UMU. We videotaped an interview with
her and we transcribed it. To encourage the replicability
of our results, we provide (below) both our transcript
data and our step-by-step interpretation of Carmens
situation and how she brought integrity to the data with
which she was working. This will allow other researchers
to perform studies in comparable ways, where their
findings could then serve the scientific function of
corroborating, refuting, or building on the current
study.
At the beginning of the interview, Carmen sat at her
desk, which was crowded with stacks of paper. A computer
monitor, keyboard, and mouse also competed for space on
the desktop. We opened the interview by asking Carmen to
explain how she handles the students accounts.
"Lets talk about whats on my desk,"
she began.
3.1 The process of achieving and
maintaining data integrity as a human activity: an
instance where a number, "3,133," tells a story
A form that Carmen often handles is titled "UMU
Tuition Fees Covered by Internal Funding" (or
referred to simply as "internal funding").
Carmen explained,
The thing I do is I deal with a lot of
departments. Departments usually send me forms. Forms
that look like this. These are called internal
funding. ...What this means is that departments grant
students scholarship awards. Students that they think
are outstanding... What happens is they have a
research grant account which they want the money
debited from and we credit the students
account.
What they do is something like this. They send
me a form. [Carmen holds one up. It is an
internal funding form. She points to a place on it.]
This is the students name. And this is the
student number. We work by student number. Student
number tells us everything about the student...
[Carmen points to a field on the paper form she is
holding up.] So they put in the student name and
the student number for us. And they say, ok,
[Carmen points to another part of the form] this
is a scholarship award. [Carmen points again] and
this is [the department of] medicine. [Carmen
points again] and this is the account that
its coming out, to credit the student. This is
their research account... And [Carmen points
again] they tell me its the fall deposit.
And they give me an amount. But this amount is not
always consistent.
I dont take their word for it. I have to,
to verify it... So what happens is that as soon as I
get this [form] I go into the student record
account. Like this. [Carmen works the keyboard
and mouse.] ...So a student walks in [to the
office] and says, I want you to check my account,
so we look into it. [Carmen randomly picks up an
internal funding form from her desk and points to a
field in it.] I punch in this [student]
number...
[Carmen uses the mouse to click to a screen in
this students record.] What she is telling
me, for the fall they [the department] want to
pay 3,133. [Carmen points to a field on the
monitor.] Im looking [at the] fall
assessment...
[Carmen points to a field on the monitor.] So
the fall assessment says 3,885.58. This... this is an
international rate. Ill go back. Ill show
you. [Carmen clicks back to another screen.] ...He
is a full time international student, graduate level.
Hes taking 15 credits. But how do I know
hes international? [The screen does not say
so.] Well, you have to know the rate, for citizens
and international. The rate for citizens is usually
900 full time... And international is like 10,000...
[Carmen looks at the internal funding form.] So
I question her. She tells me she only wants 3,133.
Usually the department pays the whole tuition
[which is 3,885]. So I say why is she only giving
me 3,133? Ok, this student is receiving a partial fee
waiver on top of this. This is a perfect example of
where I run into complications.
What happens at this point is department
usually sends a form [the internal funding form]
and they tell me, ok, pay off all the students
tuition. And the student receives [in addition]
a partial fee waiver. And this is governments
money. The government. They are giving to the
department, they are giving the fellowship
[office] so much money and they distribute to
departments. And this helps to pay, this is a
reduction on their fees. Its always for
international students because their fees are so
high... Citizens, never, because their fees are so
low.
[We question Carmen as to whether its a hard
and fast rule that you can always tell, by the fees
charged, if the student is international or a
citizen.] ...I would have to go back to fellowship
[the fellowship office], I dont know how
that came about. Basically... its been like
that for many years. I know where the money is coming
from, its coming from the government... They
distribute it among the departments and they say,
well, whoever you think [deserves it]...
Students have to be teaching assistants, they have to
be working at UMU, have to be completing their
degree... They have to apply for this, the students.
They dont automatically get it. There are a
limited amount of students that receive it... [Carmen
points to the paper form on her desk] and this
particular student is fortunate enough to have the
department pay, but is also getting a reduction in
fees.
Now the problem is see, the department
is saying, ok, theyre paying off the tuition.
The tuition is 3,885. Now I could put through 3,885.
Now this partial fee waiver comes about ok,
753. If I were to put in 3,885.58 [Carmen points
to a field on the monitor] and then receive a
partial fee waiver, I would have to credit back to
the department because its not the
students money. The department expects the
money to go back to them...
For Carmen, "3,133" was not just a number.
"3,133" told a story. To identify the main
points in the lengthy quotation, above, we provide an
outline of the story as follows. The story began with
Carmens own examination of 3,133. "[Carmen
looks at the internal funding form.] So I question
her. She tells me she only wants 3,133. Usually the
department pays the whole tuition [which is 3,885]. So
I say why is she only giving me 3,133?" The
reason behind this seemingly odd behavior of this other
person, Carmen found out, was that the department was
also awarding the student a partial fee waiver, worth
753. Together, the scholarship award (3,133) and the
partial fee waiver (753) equaled the tuition and fees
charged to the student (the fall assessment, 3,886).
Based on this, we can interpret one of the meanings that
this number evoked in Carmens eyes was that
"the amount of a scholarship for a student, even
when it is less than the fall assessment, is considered
valid if the amount of the scholarship plus amounts from
other payment sources equal the amount of the fall
assessment." For Carmen, this meaning would be a
matter of common sense; for a database analyst, this
meaning would be labeled a "business rule."
Whereas the outline of the story about Carmens
reaction to the number, 3,133, appears to involve just
the straightforward application of a rule, we can also
look more closely at the story to uncover rich details
about how the reaction arose in the first place.
Specifically, when we were interviewing Carmen and,
again, when we were reviewing the videotape, we wondered,
exactly who is this "she" whom Carmen was
questioning ("So I question her")? And who is
this "she" who was even talking to Carmen
in our presence ("She tells me she only wants
3,133")? Whoever this person was, she was not
physically present in the room when we were conducting
the interview and she was not named or otherwise
mentioned in the internal funding form. So, where did
this person come from?
With the help of the hermeneutic concept of
"enactment," described above in section 2, we
can explain that Carmen, in her creation of a meaning for
the number 3,133, also literally created or enacted
the presence of this other person and willfully inserted
this person into the environment, there this person took
a place alongside the internal funding form,
Carmens desk, and us (the interviewers). Though not
physically visible to us, this person was present for
Carmen (it turned out to be a person from the department
who filled out the internal funding form). Indeed, this
person was sufficiently real and objective for Carmen (1)
to participate in a dialogue with her ("So I
question her. She tells me she only wants 3,133."),
(2) to notice that her behavior was odd ("why is
she only giving me 3,133?"), and (3) even to be
sufficiently motivated by the enacted persons odd
behavior to proceed to verify the validity of
"3,133." With regard to the hermeneutic concept
of enactment, it is important to emphasize that this
person was not already "there," objectively
existing and just waiting to impinge upon the choices and
decisions that could be made by Carmen. Rather, this
person materialized only upon Carmens enactment of
her.
As Weick states (p. 28), "the [enacted]
environment is put there by the actors within the
organization and by no one else." This is exactly
what we observed in Carmens enactment of the other
person and Carmens willful placement of this person
into Carmens own data-integrity environment.
At the beginning of our interview with Carmen, this
enacted person was present nowhere in the room and was
not available to be observed by anyone not even
Carmen. If asked, neither Carmen nor anyone else could
have noted or vouched for this other persons
presence at the beginning of the interview. The same is
true for the meaning embodied in Carmens business
rule (which we interpreted, above, as "the amount of
a scholarship for a student, even when it is less than
the fall assessment, is considered valid if the amount of
the scholarship plus amounts from other payment sources
equal the amount of the fall assessment"). At the
beginning of the interview, this business rule was
present nowhere in the room and was not available to be
observed by anyone, including Carmen herself. So, where
did this rule come from? Like the person with whom Carmen
participated in a dialogue in our presence, the business
rule materialized only upon Carmens enactment of it
and her willful insertion of it into the environment,
taking a place alongside the internal funding form, her
desk, us (interviewers), and even the person who Carmen
enacted.
In summary, "3,133" prompted Carmen to enact
(1) the presence of another person ("she") who,
in speaking with Carmen, behaved in a way that Carmen
found odd (that is, her giving Carmen only
"3,133," rather than 3,886), and (2) the
business rule that served to make this other
persons behavior sensible to Carmen. Hence,
"3,133" was not a mere number that Carmen
simply read or "inputted" from a form she was
reading and then keypunched or "outputted" to a
computer she was operating. Carmen was not a passive
recipient of "3,133," but was an active
component herself in the student accounts offices
data-management system, where she actually verified the
integrity of the "3,133."
3.2 The process of achieving and
maintaining data integrity as a human activity: an
instance where 4,290 = 3,100.93
As we continued with the interview, we witnessed
additional instances of Carmens enforcement of
business rules. In one instance, an internal funding form
indicated "4,290" as the amount that a
department was awarding to a student, but Carmen
responded by changing it into "3,100.93." In
this instance, we again observed Carmens enactment
of (1) from her perspective, the presence of other people
and (2) also what we would describe as some business
rules.
[Carmen enters a students number into the
keyboard and she quickly clicks through several
screens.] I happened to catch it on time. [Carmen
turns to a paper form for this student.] See here,
the department put down for the deposit, 4,290. [Then,
Carmen points to something that was written on the
form by hand.] I wrote this in. I wrote this in
myself. The original said 4,290. Thats why I
cannot take their word for it. I go into his account
this was the fall of last year. [Carmen
points to a field on the monitor.] The fees came
out to 3,853.93. What they were trying to do was pay
this, plus health insurance. Which is $440 for health
insurance. Which comes out to 4,290. So they have the
right concept. Theyre telling me theyre
paying the whole fees, at the time. So I look and,
ok, that sounds good. Now in the meantime, while I
was processing this, I notice [Carmen points to a
field on the monitor] that this student got a
partial fee waiver. [Carmen points to the field
on the monitor.] You see that? 753 credit. So I
couldnt possibly do 4,290. He also got exempt
from health insurance. [Carmen points to another
field on the monitor.] International health
insurance, zero. Hes not paying 440. He got 753
credit. So what I did what I took this amount [Carmen
points to a field on the paper form for this student]
and I reduced it to 3,100.93.
[We ask Carmen: How did you get 3,100.93?]
In all, I have to come out to 3,853. [She
points to this number appearing in a field on the
monitor]. Thats what my goal is.
[Carmen points to different parts of the screen.] So
I reduce 440 [the exact figure is 436.07] from
this [that is, from the 4,290], I put in 753,
and [she points to the paper form] the
department I just took from their account $3,100. So
were saving I mean, I would have taken
out 4,290. Instead, I only took out 3,100.93...
coming out of their research account.
As in the first instance above, where the number
"3,133" tells a story, we can look closely at
this instance to uncover similarly rich details about how
Carmen brought about data integrity. First, just as
Carmen in the previous instance had enacted another
person, "she," and proceeded to have a
conversation with this person, Carmen in this instance
enacts a group of people, "they," who literally
talk to her ("Theyre telling me
theyre paying the whole fees, at the time")
and whom she hears ("So I look and, ok, that
sounds good"). These people were not present at
the beginning of our interview. They acquired a presence
only upon Carmens enactment of them and her willful
insertion of them into her environment.
Second, in explicitly drawing attention to
"3,853" as her goal ("In all, I have to
come out to 3,853. Thats what my goal is.")
and in walking us through her activity, Carmen revealed
her use of several related (perhaps, hierarchical)
business rules. We can express them as follows:
- "the amount of a scholarship for a student
must not be greater than the students fall
assessment";
- "the amount of a scholarship for a student
must not be greater than the students fall
assessment, less any incorrectly charged fee for
health insurance"; and
- "the amount of a scholarship for a student
must not be greater than the students fall
assessment, less any incorrectly charged fee for
health insurance, less any partial fee
waiver."
At the beginning of our interview with Carmen, these
business rules had the same status as the people she only
later enacted: these rules were present nowhere in the
room and were not available to be noticed or observed by
anyone not even Carmen. These business rules, just
like the group of people with whom Carmen had a
conversation, materialized only upon Carmens
enactment of them and her willful insertion of them into
her environment.
Although Carmen enacted these rules, she did not
create them randomly. Instead, it was through her
accumulated experience at UMU (in particular, her being
socialized into and becoming an agent of the socially
constructed world of the student accounts office)
that Carmen developed these business rules and
"programmed" herself to enforce them. The
result of her effort of achieving data integrity was the
transformation of "4,290" into "3,100.93."
In other words, Carmen actually determined
"4,290" as having the meaning of
"3,100.93"! The student received an award
of $3,100.93 and the students department saved the
difference, $1,189.07.
3.3 The process of achieving and
maintaining data integrity as a human activity: an
instance where the meaning of a number changes over time
In the collection of computerized and manual databases
in the student accounts office, a set of data that is
consistent today might not be consistent tomorrow. For
instance, a change in a students
full-time/part-time attribute (such as when a student
drops a course and falls below the pre-established,
course load based definition of "full time")
would invalidate the value that Carmen had previously
entered (753) for the partial-fee-waiver attribute (since
only full time students can be eligible for the full
753). In other words, consistency is affected by time.
This means that the data-integrity efforts required of
Carmen could be never ending. From the following portion
of our interview with Carmen, we will draw material to
illustrate how Carmen accounted for the time-dependent
nature of the integrity of her data:
The problems I run into... [Carmen turns to
a paper form on her desk and runs her finger down
it.] This person, from the time she told me to me
to put in 2,894, I put it in. Then [after this
person is contacted by Carmen, as Carmen explains
later], she tells me: cancel the scholarship, the
student withdrew from the university. Lets look
him up. [Carmen works the keyboard and mouse to
enter the students id number.] He was
charged 3853. I mean hes credited. Now
theres a $50 charge for withdrawing, ok? [Pointing
to a field on the monitor], look at what happened
here. In September 30, I put in 2,894, which is a
scholarship award [the source of which is funds
from a department]. Then he also received a
partial fee waiver [the source of which is
governmental] on top of it. So at the time [thumbing
through some forms on her desk] I always do
"print screens" of the account, always.
Because if I dont, the department could [say]:
"Why did you give him 2,894?" Because the
student withdrew, we only have a $50 charge in his
account. [The computer system only provides a
current account balance without a history of the
transactions that led to that balance. It shows $50.
It does not show $2,894 anywhere.] Well, I have
back up. He wasnt [withdrawn] at the
time. [Pointing to her print-screened hard copy,]
he was charged 3853.93, when I processed it. So I
have to back myself up in order to say why I credited
the student. ...So I tell him [pointing to the
print-screened hard copy] ok, the student was
registered and now [pointing to a field on the
monitor] I see the student has withdrawn. Now [pointing
at the monitor] you wont see this, ok, but
he did have a huge credit of $3,000 sitting there. So
what did I do, I took it all back. I have to reverse
it. I did this in December. I caught this in
December.
[We ask Carmen: How did you come across it?]
...well I happened... OK, student must of
withdrawn in November. Thats what happened.
Because students come down [to the student
accounts office], and I didnt see it, ok?
And he didnt ask for a refund, because I guess
he figured the money wasnt his. Ah, what
happened is, first of all, I read the statements, so
I got my statements in November. I saw he had a huge
credit of $3,000. I called the department. I said,
"why does this student have a huge credit? He
withdrew from the university." And they said,
"oh yes, you can cancel it." But they never
called me. Sometimes they call me. Sometimes they
dont... In this case, they never called me. I
called her. And she sent me a memo [pointing to
it], saying, see, look at the withdrawal... [Flipping
to some other papers in the same stapled stack for
this student], if you look at credit ahead of the
time [points to a field on the paper].
[We look at it. The credit is 2,744.]
Yeah, this was dated November 11th.
[We ask: So do you do a routine scan of accounts
that have large credits?]
Whats happening is, where the
inconsistencies come about is that I always have to
look into the students accounts all the time...
When you deal with fee, student fee, facts change
automatically. This is a perfect example. I processed
this in September. Come November, he withdrew from
university. The problem is, he withdraws from the
department and the department, you know, paperwork
just keeps staying and staying and things dont
get done automatically... In November, we found out
that the student has $2,000 sitting in his account
which wasnt his and I had to reverse
everything. Partial fee waiver, internal funding.
[We ask: How did you find that out, again?]
Statements.
[We ask: What is a statement?]
Ok, a fee statement. Students receive their
bill, ok? Like this [holding one up]. They get this
in the mail, saying you owe so much money on
such-and-such a date... [Running her finger down
this statement,] ah, here, UA [which is a code
indicating that the student is receiving a
scholarship award]. This one doesnt get
mailed out to them. Why? Because of this little thing
[pointing to the top of the statement page]
that says UA. I inquire upstairs. They pull out these
particular students for me. [That is, they pull
out these physical statements from all the others, so
that these will not be mailed out.]
[We ask: They print those out in a separate batch
and give them directly to you?]
Yes. This is the only way I can verify to see
where I put my adjustments. The ones that are ok, I
mail them out. The ones that have problems, I put
them aside and I correct them.
This portion of our interview with Carmen reveals how
the validity of "2,894" (a scholarship award,
involving a transfer of funds from a departments
account to a students account) was time-dependent.
In September, "2,894" was consistent with other
data; from Carmens perspective, a salient fact was
that the scholarship award, "2,894" did not
exceed the total fees, "3,853.93," charged to a
full-time student in September. In December, the very
same number, "2,894," was no longer consistent
with other data; from Carmens perspective, it was a
scholarship awarded to a student who was no longer
registered. In walking us through the activity in which
she detected that "2,894" changed from being a
valid value at one time to being an invalid value at a
later time, Carmen demonstrated to us how she enacted
her environment and, in doing so, brought about data
integrity.
In the instances examined in subsections 3.1 and 3.2
above, Carmen enacted people and rules. In this instance,
Carmen enacts something new. She enacts a history.
Specifically, she enacts an environment with a particular
sequence of historical events: (1) a student acquired
full-time status in September, (2) the student dropped
out in November, and (3) the department, which had
originally made the award to the student, remained
unaware of the students change in status. Carmen
enacted this environment as a result of her coming across
the number, 2,744, in an un-mailed student fee statement.
How was it possible for this one number to lead Carmen to
enact an environment having a history of these three
events?
First, for Carmen, the symbol "2,744"
appearing on the fee statement had two possible meanings:
it could have meant a credit of 2,744 dollars for the
student, but it also could have meant that something in
the student accounts database was incorrect. We observe
that, in effect, Carmen recognized the latter possibility
by using a rule that we interpret to be "a large
credit in a students account in the middle of the
semester should be investigated as a sign of an
error." As a result of this data-integrity activity,
Sheila determined that, in effect, the "2,744"
appearing on the fee statement did not really mean 2,744
dollars; in other words, "2,744"¹ 2,744.
Second, in pursuing the possibility of there being an
error, Carmen checked this students account and saw
the number 50 on the computer monitor. For Carmen, the
number "50" not only signified a withdrawal
charge for the student, but also indicated that there
were previously recorded figures that the computer system
(due to its lack of a transaction-logging capability) no
longer stored or displayed. Interestingly, Carmen herself
had devised a means to test the integrity of this portion
of her enacted environment. By referring to
print-screened hard copies that, as a matter of routine,
she had made much earlier, she was able to verify not
only that there were previously recorded figures, but
also what those figures were: a scholarship of
"2,894" and the original charge of
"3,853.93," both in September. In effect,
Carmen inserted into her environment a student who had a
full-time status in September and then later (in
November, she theorized) withdrew. Also, we can say that,
for Carmen, "50" in December means
"2,894" and "3,853.93" in September.
Third, the combination of coming across the numbers
2,744 (a possible credit or a possible mistake) and 50 (a
sign that the student had a full time status but later
withdrew) motivated Carmen to suspect that the department
was unaware of the students change in status,
prompting her to contact the department ("I
called the department. I said, why does this
student have a huge credit? He withdrew from the
university") which then resulted in her
receiving verification from the department for her to
proceed with the overall scenario she was enacting
("And they said, oh yes, you can cancel
it")
In Carmens data-integrity activity, which was
her reaction to her reading of "2,744" in a fee
statement, we can observe the functioning of several
business rules. We can interpret them as follows:
- "a scholarship award to a student [e.g.,
"2,894"] must not exceed the total fees
charged to the student [e.g.,
"3,853,93"]";
- "a scholarship award to a student can be
valid only if the student is registered";
- "a scholarship award to a student who has
withdrawn is not valid"; and
- "if the amount of a credit on a fee
statement for a student is huge
[why does this student have a huge
credit?], the amount is not valid,"
where "huge" would be gauged by
Carmens own experience and expertise.
The following rule, though not necessarily a business
rule, still supported Carmen in her data-integrity
activity:
- "the occurrence of a withdrawal fee for a
student [that is, 50] necessarily
indicates earlier occurrences of other activity
in the students account when he or she was
still registered [e.g., 2,894 and
3,853.93]."
In tracking the validity of "2,894" over
time, Carmen invoked all of these rules in her overall
enactment of an environment that was the setting to a
particular sequence of historical events. With the help
of this enacted history, Carmen did not accept an
incorrect number (2,744) at face value, but instead made
the correction of reversing the scholarship award,
thereby recovering $2,747 for the department that had
awarded this amount to the student before he withdrew
from the University.
3.4 The five hermeneutic concepts
and the process of achieving and maintaining data
integrity
In our examination of Carmens data-integrity
activity, we mentioned the hermeneutic concept of
enactment explicitly in the three instances. However, as
described earlier in this study, enactment takes place
along with distanciation, autonomization, social
construction, and appropriation. To clarify how the
organization-science approach of hermeneutics informed
our examination, we now make our usage of the remaining
four hermeneutic concepts explicit.
First, all the data that Carmen was dealing with,
whether they appeared on paper forms or her computer
monitor, had undergone distanciation and autonomization.
Here, we will focus on "3,133" (from §3.1),
"4,290" (from §3.2), and "2,894"
(from §3.3). Immediately upon their being inscribed in
the internal funding forms, the symbols
"3,133," "4,290," and
"2,894" all commenced their separation in time
and distance from their authors (the people who wrote
these numbers in one or another internal funding form),
but still had the capacity to be or become meaningful to
a reader even in the absence of their originating
authors. With regard to the human process of achieving
data integrity, it is significant that the symbols
"3,133," "4,290," and
"2,894" all took on a life of their own, since
this autonomy allowed their meanings to change and, if
necessary, be corrected as indeed we witnessed in
the cases of "4,290" (the correct meaning of
which Carmen determined to be "3.100.93") and
"2,894" (whose meaning was correct at the time
it was entered in September, but no longer when Carmen
checked it in December).
Second, to bring out the significance of social
construction, we ask the reader of this paper to
consider what the symbols "3,133,"
"4,290," and "2,894" would all mean
to a person who is a stranger to UMU, to the UMU student
accounts office, and to the UMU student accounts
offices internal funding forms. Of course, if
divorced from this context, the symbols
"3,133," "4,290," and
"2,894" would all be meaningless. For a
stranger, these symbols would be nothing more than a
string of digits. Carmen, however, was no stranger to the
UMU student accounts office. Instead, she was a member
and agent of the socially constructed world of
UMU, including its student accounts office. In her
reactions to "3,133," "4,290," and
"2,894," this socially constructed world
imposed itself in the form of the many business rules we
observed in §3.1, §3.2, and §3.3. (We emphasize that,
whereas we have chosen to label them as "business
rules," they were simply common sense or standard
operating procedures for Carmen.) These rules were not
random; Carmen did not simply make them up. Rather, they
were already a part of the objective, socially
constructed world of the UMU student accounts office that
Carmen encountered as "given."
Third, Carmen appropriated the text, including
the symbols "3,133," "4,290," and
"2,894." She did not simply accept their
meanings at face value; instead, she tested their
correctness by applying business rules and, in this way,
demonstrated the process of achieving data integrity as a
human activity. At the same time, one can also say that
the text succeeded in appropriating Carmen by requiring
her (perhaps, even conscripting her) to act, again and
again, as an agent of the socially constructed world of
the UMU student accounts office where she was enforcing
its rules.
One may attempt to discuss the five concepts
separately, but in reality the five concepts work
together and are difficult to separate. For instance,
when appropriating "4,290," Carmen
relied on a business rule imposed upon her by the socially
constructed world of the UMU student
accounts office. With this, she proceeded to enact
a meaning for "4,290" that even transcended its
meaning to whoever first wrote it in an internal funding
form; this also served to further the autonomization
of the text, "4,290," which in turn illustrated
to us its distanciation from the intentions of the
person who first composed this number. Finally, this
resulted in Carmens correction and transformation
of the symbol "4,290" into
"3,100.93." Taken together, the five
hermeneutic concepts in this scenario illustrate an
instance of the process of achieving data integrity as a
human activity.
Our case study is useful for
demonstrating that a human being can play a
significant data-integrity role as a part of a
data-management system. Whereas the UMU student accounts
office could stand to benefit by the adoption of some of
the standard computer-science innovations mentioned in
§1 of this study, the rather primitive state of
data-management technology in the student accounts office
is nonetheless useful for revealing to us, as outside
observers, the human capacity to cope with unintegrated
data and, where necessary, to bring about integrity in
the data.
The technological innovations that the discipline of
computer science has contributed to the specific
data-management task of safeguarding data integrity are
invaluable and irreplaceable. These technological
innovations (which include data-management systems in
themselves, different architectures for structuring the
data in a database, semantic or entity-relationship
models, and automated business-rule enforcement) will
always remain key components in any computerized
information system that manages the integrity of the data
in a database.
Still, as impressive as the computer technology has
been, it has not been enough. Whether due to
technological or organizational reasons, data integrity
remains an elusive goal for many, if not most, business
organizations.
There are few known examples of corporations that have
drawn up and implemented truly enterprise-wide data
models. The result is that, in most business
organizations, the data in a departments local
data-management system are often incompatible or
unshareable with the data in other departments
respective data-management systems. In a world without
resource constraints, an organization could draw up a
truly enterprise-wide data model and reengineer all of
its existing databases so as to conform to the new model.
However, because all organizations have resource
constraints, a more practical alternative would involve
something other than a total-technology solution. Such a
solution would involve recognizing that some (though not
all) individuals are capable of shouldering a major share
(but not all) of the burden of the data-integrity process
and that data integrity can sometimes (though not always)
require an individual to be assigned a major operational
role in it. Such a solution would involve, on the one
hand, implementing some of the technological innovations
of computer science and, on the other hand, treating
managers, like Carmen, as system components to be built
right into the design of an overall (not only
computerized) data-management system, where these human
beings would have carefully specified roles in the
process of bringing about data integrity.
Simply specifying such a role in the form of a job
description and then hiring someone to fill it,
unfortunately, would not work. We saw that, in the case
of Carmen, she was already thoroughly socialized as a
member of the UMU student accounts office, where it was
largely this already existing socially constructed
world (including the business rules that it was
imposing) that was acting through Carmen as its agent.
And it was only as a member of this socially constructed
world that Carmen, in specific instances, was able to enact
a suitable environment for achieving and maintaining data
integrity (that was, among other things, populated by
people that were invisible to us, the interviewers). With
her enacted data-integrity environment, Carmen proceeded
to verify or, when necessary, change the data she was
working with. How does such socialization into one or
another socially constructed world take place and how
might it be promoted, particularly with regard to the
process of achieving and maintaining data integrity as a
human activity? What are the circumstances under which a
manager could successfully enact a suitable
data-integrity environment (and which should be promoted)
and the circumstances under which a manager would fail
to enact a suitable data-integrity environment (and which
the design of the overall data-management might be
structured to help avoid)?
For future theory-oriented research in data
management, perhaps the most enticing question of all is:
what balance or division of labor should there be between
computer-science researchers and organization-science
researchers with regard to the development of innovations
still needed for achieving and maintaining the integrity
of data in business organizations? For future
practice-oriented research on data management, the
corresponding questions are: how should the formal,
technological procedures of systems analysis and systems
design be expanded so as to recognize human beings not
just as "users" of output but also as system
components, and what metrics might be developed to help
assess the different contributions to data integrity that
would be made by, or expected from, the
"computer-science components" (hardware,
software, and database design) and the
"organization-science components" (people and
their procedures)?
Finally, there are ramifications for the approaches
that future organization-science research can take. The
hermeneutic approach in this study is but one
interpretive organization-science approach; there are
numerous other interpretive organization-science
approaches and they have already been successfully
deployed in other domains of research on information
systems (Mumford et al., 1985; Nissen et al.,
1991; Lee et al., 1997). Furthermore, interpretive
approaches represent but one research domain within
organization science; there are also the positivist
approaches, examples of which have actually predominated
in the scholarly journals in the information-systems
discipline. All organization-science approaches have the
potential to contribute innovations still needed to
achieve and maintain integrity in the data of all
organizations.
Our investigation illustrates the
classic Davis & Olsen conception (1985, p. 53) that
the salient components of an information system are not
only hardware, software, and data, but also people and
procedures. The starting point for our research was the
state of the art of data management, which recognizes
data integrity to be a computer problem. Our case study
supports the conclusion that data integrity is not only a
computer problem requiring computer-science solutions,
but also an organization problem requiring
organization-science solutions.
Arnold, S. J. & Fischer, E.
"Hermeneutics and Consumer Research," Journal
of Consumer Research, Volume 21, June 1994,
pp. 55-70.
Baskerville, R.
"Risk Analysis: An Interpretive Feasibility
Tool In Justifying Information Systems
Security," European Journal of
Information Systems, Volume 1, Number 2,
1991, pp. 121-130.
Baskerville, R.
"Information Systems Security Design
Methods: Implications for Information Systems
Development," Computing Surveys,
Volume 25, Number 4, December, 1993, pp. 375-414.
Boland, R.J.
"Information Systems Use as a Hermeneutic
Process" in Information Systems Research:
Contemporary Approaches & Emergent Traditions,
edited by H.-E. Nissen, H.K. Klein, & R.
Hirschheim. New York: North-Holland, 1991, pp.
439-458.
Boland, R.J.
"Accounting and the Interpretive Act," Accounting,
Organizations and Society, Volume 18, Number
2/3, 1993, pp. 125146.
Date, C.J.
An Introduction to Database Systems. Sixth
Edition. Reading, Massachusetts: Addison-Wesley,
1995.
Davis, G.B., Lee, A.S., Nickles, K.R., Chatterjee, S.,
Hartung, R. & Wu, Y
"Diagnosis of an Information System Failure:
A Framework and Interpretive Process," Information
& Management, Volume 23, Number 5, 1992,
pp. 293-318.
Davis, G.B. & Olson, M.H.
Management Information Systems: Conceptual
Foundations, Structure, and Development.
Second Edition. New York: McGraw-Hill, 1985.
Elmasri, R & Navathe, S.B.
Fundamentals of Database Systems. Second
Edition. Redwood City, California:
Benjamin/Cummins, 1994.
Hirschheim, R.A., Klein, H.K & Lyytinen, K.
Information Systems Development and Data
Modeling: Conceptual and Philosophical
Foundations. New York: Cambridge University
Press, 1995.
Lee, A.S.
"Integrating Positivist and Interpretive
Approaches to Organizational Research," Organization
Science, Volume 2, Number 4, 1991, pp.
342-365.
Lee, A.S.
"Electronic Mail as a Medium for Rich
Communication: An Empirical Investigation Using
Hermeneutic Interpretation," MIS
Quarterly, Volume 18, Number 2, 1994, pp.
143-157.
Lee, A.S., Liebenau, J. & DeGross, J.I.
Information Systems and Qualitative Research.
New York: Chapman & Hall, 1997.
McFadden, F.R. & Hoffer, J.A.
Database Management. Third Edition.
Redwood City, California: Benjamin/Cummins, 1991.
Mumford, E., Hirschheim, R., Fitzgerald, G. &
Wood-Harper, T.
Research Methods in Information Systems.
New York: North-Holland, 1985.
Myers, M.D.
"A Disaster for Everyone to See: an
Interpretive Analysis of a Failed IS Project,"
Accounting, Management and Information
Technologies, Volume 4, Number 4, 1994, pp.
185-201.
Nissen, H.-E., Klein, H.K. & Hirschheim, R.
Information Systems Research: Contemporary
Approaches & Emergent Traditions. New
York: North-Holland, 1991.
Phillips, N. & Brown, J.L.
"Analyzing Communication in and around
Organizations: A Critical Hermeneutic
Approach," Academy of Management Journal,
Volume 36, Number 6, 1993, pp. 1547-1576.
Ricoeur, P.
- Hermeneutics and the Human Sciences, J.B.
Thompson (ed.). New York: Cambridge University
Press, 1981.
Taylor, C.
- "Interpretation and the Sciences of
Man," in Interpretive Social Science: A
Reader, edited by P. Rabinow and W.M.
Sullivan. Berkeley: University of California
Press, 1979, pp. 25-71.
Thompson, C.J., Locander, W.B. & Pollio, H.R.
"The Lived Meaning of Free Choice: An
Existential-Phenomenological Description of
Everyday Consumer Experiences of Women," Journal
of Consumer Research, Volume 17, Number 3,
1990, pp. 346-363, 1990.
Tice, T.N. & Slaven, T.P.
Research Guide to Philosophy. Chicago:
American Library Association, 1983.
Weick, K.E.
- The Social Psychology of Organizing, .
Reading, Massachusetts: Addison-Wesley, 1969.
Winograd, T. & Flores, F.
Understanding Computers and Cognition: a New
Foundation for Design. Norwood, New Jersey:
Ablex Pub. Corp., 1986.
Wybo, M.D.
"Maintaining Heterogenity in a Shared Data
Environment: a Case Study," Working Paper
94-07-09, Faculty of Management, McGill
University, 1994.
Wybo, M.D. & Lee, A.S.
"Data Integration as a Human Activity:
Implementing Partial Integration in a University
Accounting Office," Working Paper 94-07-09,
Faculty of Management, McGill University, 1994.
Zachman, J.A.
"A Framework for Information Systems
Architecture," IBM Systems Journal,
Volume 26, Number 3, 1987, pp. 276-292.
Zuboff, S.
In the Age of the Smart Machine: the Future of
Work and Power. New York: Basic Books, 1988.
|