<V
DESIGN AND IMPLEMENTATION OF DECISION
SUPPORT SYSTEMS IN THE PUBLIC SECTOR
John
C.
Henderson
David
A.
Schilling
September 1984
CISR WP #118
Sloan WP #1550-84
Center for Information Systems Research
Massachusetts
Institute of Technology
Sloan School of Management
77 Massachusetts Avenue
Cambridge, Massachusetts, 02139
DESIGN AND IMPLEMENTATION OF DECISION
SUPPORT SYSTEMS
IN
THE PUBLIC SECTOR
John
C.
Henderson
David
A,
Schilling
September 1984
CISR WP £118
Sloan WP #1550-84
©
J.
C.
Henderson,
Forthcoming
in
D,
A.
Schilling
1984
MIS Quarterly.
Center for Information Systems Research
Sloan School of Management
Massachusetts Institute of Technology
ABSTRACT
This paper examines the implications of utilizing decision support systems
(DSS)
in
the public sector based on a
community mental health system.
programming)
process.
DSS developed and implemented
for a
The DSS includes a multiple objective
(goal
model and encompasses a multiple party decision
allocation
The experiences and insights acquired during the development and
implementation are relevant
to public sector decision support, in general.
The
importance of a DSS as a process-support aid rather than simply providing
answers
(i.e.,
a product-oriented aid) and the interaction of system architecture
and the chosen design strategy are key insights.
between
model-oriented
appropriate.
requires
typical
support.
and
data-oriented
In particular,
DSS
does
not
The public sector decision maker's concern with
the ability
spread sheet
to
operate
in
a
higher dimensional
model and there
is
a critical need
the distinction
appear
to
be
issues of equity
framework than the
for
communication
1.
Introduction
Developing
and
challenging task.
As
implementing decision aids
Lamm
public
the
in
sector
is
a
(1980) and others have pointed out, the political
process tends to promote those that survive or win, not those seeking truth.
Often, the essential benefit of a decision aid, a valid model,
is
that most threatens (he survival of the public decision maker.
the very element
It is
not surprising
that Brili (1979) notes, "Designing a solution to a public sector problem
is
largely
an art."
Hammond
aids
unless
(19S0) suggests that
explicit
attention
is
it
may
given
to
not be sufficient to provide decision
how these
Without effective learning support,
learning.
aids
Hammond
support
(1980),
effective
Pressman and
Wildavsky (1979) and others predict dysfunctional consequences are likely to
resuit
from our policy making processes.
experirneiital approach
Although
Hammond
a necessary condition for learning,
is
lie
argues a quasinotes that the
strong quasi-rational model of inquiry represented by the application of manage-
ment science techniques has had positive impact on public sector decision
making.
For example,
multiple objectives and
management science models can
when combined with the
help to externalize
results of quasi-experiments
provide an enhanced learning environment.
The need
and
to facilitate
organizational
access to decision aids as well as support individual
learning
explicitly
is
addressed
in
the
decision
support
systems literature (Sprague and Carlson, 1982) and (Alavi and Henderson, 1981).
As Keen and Scott-Morton
strategy for
develops a
decisions.
(1979), Alter (1980) and others note, the basic design
DSS begins with an
tool
for
the
user
analysis of the decision process and adaptively
to
learn
about and cope with semi-structured
Experience
ease
DSS design has
in
of. use (at least
directed
1975) or prototyping (Keen and Gambino, 1981) are
towards
approaches assume there
achieving
is
a
to
the decision process.
enhanced (perhaps even made possible) by developing an
learning
curve,
design
be significant user and analyst learning both
will
with the characteristics described above.
along
These
characteristics.
these
terms of the technology as well as with regard
learning
Design methodologies
by an intermediary), and adaptability.
such as middle-out (Ness,
explicitly
also indicated the importance of flexibility,
the
system
is
initial
in
This
system
As both the user and analyst move
adapted
to
support
their
evolving
information and learning needs.
The trend
in
public sector applications of
seems consistent with
this
perspective.
management
science techniques
Public sector planning models have
evolved from those that focus on efficiency to those that attempt to describe
and account for conflicting objectives (ReVelle et
multiobjective models
in
ai.
areas such as fire station location (Schilling et
1980), policy patrol scheduling, (Saladin, 1980) and
al.,
1978;
ai.,
water resource management
(Major and Lenton, 1978) are some recent illustrations.
(Henderson et
The application of
1977).
Recursive frameworks
Cohon and Marks, 1973) have been proposed
that use a
multiobjective planning model to establish system parameters and then disaggre-
gate these solutions using heuristic and simulation models
their
impact on system operations.
This iterative approach
with the adaptive design concepts proposed by
order to evaluate
in
is
quite consistent
DSS researchers.
Research on the application of decision support systems
in
the
public
sector has also emphasized the need to address both the problems of conflicting
objectives as well as the need to better support the traditional data analysis
efforts of the policy analyst.
Hammond
(1980) notes that both forms of decision
•3-
Providing these types of decision aids
aids are necessary.
adaptive
mode
is
This paper
system
the
in
the objective of
many current resea r ch
investigates the impact
DSS was designed and implemented.
embedded
friendly,
efforts.
implementing a decision support
Section 2 describes the specific setting for which
the public sector.
allocation model
of
a user
in
in
the
Section
DSS and
3
discusses the muiticriteria
the overall
DSS
Section 4
design.
elaborates on the organizational decision process, while section 5 presents the
implementing the DSS and discusses
results of
its
impact during the
two
first
years of actual use. Filially, section 6 provides some concluding remarks.
2.
The Mental Health System
lesearch
This
implemented
Board.
for
focus
will
the Franklin
en
a
decision
support
system
and
designed
County (Ohio) Mental Health and Retardation
The Board oversees forty contract agencies which provide required
community-wide mental health services. The nature of the decision process
allocation decisions
is
critical to the
Board
in this
environment.
for
There must be
opportunities for various constituencies, representing diverse interests, to have
influence
en
complex programmatic and financial decisions.
Unfortunately,
within this realm of complexity, the decision makers are often untrained.
They
are chosen based on the constituencies and values they represent, rather than on
their
knowledge of the problem
They serve
in
a voluntary
time constraints.
It
is
area, or
expertise as planners or decision makers.
mode, meeting infrequently and
little
typically, under severe
wonder that decisions often reflect the relative
power of a special interest group rather than some overall
goals and priorities.
set of
community
n-
MHR
The Franklin County
Board, faced with increasing demand and an
eroding resource base,! began an effort to improve the quality of their budget
planning and allocation process.
They identified a need
to clarify goais and link
these goals to a comprehensive model for mental health delivery.
They sought a
budget process that would provide Board members with a better understanding of
how specific allocations affected program
level
and overall community mental
health system goals.
As a starting point, they chose the Balanced Service System (BSS) model as
the fundamental conceptualization of a mental health service system.
is
a
model
of
mental
dysfunctioning
used
by
the
Joint
The BSS
Commission
on
Accreditation of Hospitals (JCAH) to generate standards for community mental
programs.
health
dimensions:
In
its
basic
form the BSS model consists
The service function
(crisis stabilization,
of
two primary
growth, and sustenance)
and the service environment (protective, supportive, and natural).
The function
indicates the nature of the service while the environment describes where the
service
is
provided.
cells of this
Each of 200 possible service types are assigned
two-dimensional matrix.
examples of the type of services
for a
in
Figure
each
cell.
1
to
one of the
depicts this matrix and includes
This model satisfies requirements
comprehensive mental health framework and also provides the basis
externalizing Board goals.
As
will
be discussed,
for
the goal structure addresses
both specific program areas
(e.g., a particular cell in
and systemwide goals
the need to balance service delivery across a range
(e.g.,
the service delivery matrix)
of service environments).
'The Board's allocations budget is approximately 20 million dollars. However, projections for budget cutbacks and inflation are significantly reducing
these resources while various need assessments indicate increasing demand for
service.
Insert Fig u re
.About
1
for an
The Board also recognized the need
began an effort
direct
link
adequate decision
develop a decision support system that would:
to
between Board goals
allocation decisions,
(2)
(as
formulated
new
using
the
They
aid.
provide a
(i)
model)
BSS
and
provide a means to better understand the tradeoffs
between goals and the impact altering goal
easily incorporate
Here
means
priorities, (3) provide the
to
restrictions, policies, or cost and service parameters into
members.
decisions, and (4) provide training tools for Board
Given these needs, a DSS design and implementation effort was undertaken.
3.1
The following sections describe the resulting DSS and
its
impact.
DSS Framework
One
of the basic concepts of
successful
public
is
the need for flexibility and adaptability
As many public sector researchers note
(Keen and Scott-Morton, 1979).
1979).
DSS
must be able
decision aids
sector
to
(Brill,
accommodate
unanticipated changes both to the structure of embedded models as well as to
the nature of the user interaction.
Achieving these system characteristics
fundamental goal of the DSS designer.
provided through a modular design.
is
consistent with that proposed
in
is
a
This flexibility and adaptability can be
The system framework employed (Figure
Sprague and Carlson (19S2).
It
2)
consists of
three basic components; model management, data management, and information
management and provides
for
a
user
friendly
interface.
Each component
is
decoupled as much as possible and consists of a set of well-defined processing
modules.
This modularity minimizes the
thereby
allowing
most
changes
number of system interdependencies,
to
be
relatively
localized
and
-6-
Further, various processing modules were written
straightforward.
analysis-oriented
level,
language (SAS).
This
language
provides
in
a
high
many data
processing-oriented macro statements and parameterized routines which sub-
reduce the time required
stantially
components.
In
to
generate or modify particular system
cases where this language did not meet specific needs, the
module was written
in
command
Fortran or a macro
language.
About Here
Insert Figure 7
While initial prototyping efforts focused on the development of a mathe-
matical model, the eventual success of the DSS depended on effective, inte-
grated software environment for each of the component systems.
discussed
in
Section
DSS generators
made
in
the
5, this
As
will
be
would suggest appropriate system characteristics
for
as well as give rise to questions on the validity of distinctions
DSS
concerning model-oriented versus data-oriented
literature
decision support systems.
To provide
component
description of each
a background for these remarks, a brief
provided.
is
The model management component focuses on the generation and execution of the allocation
definitions,
format
with
for
system
structure
model execution.
the system
A model
model.
generation module translates variable
and parameter estimates
an
appropriate
This module also provides a means to interface
transaction data base.
model can be accomplished by
into
fairly simple
Relatively extensive changes to the
adjustments to the model generation
module.
The
package.
model
execution
However, the
capabilities of the data
module
flexibility
utilized
IBM's
MPS
programming
of the model generator combined with the
management component permit
ate linear programming software.
linear
the use of any appropri-
component
ol
the
component included processing module"-
to
interface with
each
with
a?
Finally,
system and provided
DSS,
the
for interactive dialogue with the user.
model management
host
trie
operating
This aspect, termed
"system control." enables much of the operation of the model management
component
to
be relatively transparent to the user and provides the means
to
integrate this component with other parts of the system.
The model management phase of the DSS generates
solution data base.
an
appropriate
As the DSS research notes, the model must be embedded
information
system by merging
is
this solution
From
in
is
important
to
note
that
this
to
delivery
(e.g.,
These application data bases
the subsequent information processing modules.
this
component decouples the generation and
management component
programs from changes
second
database selected application data bases
execution of the model from the generation of management information.
fact, the role of a data
in
order to creaie an integrated
are extracted for use by the application programs.
in
the
data base with various other data bases
this solution
create significant efficiency
of
provide the foundation for
to
variable labels, historical data trends, etc.)
solution data base.
The purpose
system.
delivery
component, data management,
It
a large, very detailed
to isolate
primary data sources
(in
It is, in
changes to application
this case,
changes to the
allocation model).
The data management component also provides the means
analyze data
stoi
ed
later, this capability
in
the system transaction data base.
proved necessary
for the successful
As
to access
will
and
be discussed
implementation of the
DSS. A high level language (SAS) provided efficient processing of large
?The transaction data base contained over 500,000 records.
files? as
-s-
well as
thie
ability to quickly
adapt parameter calculations
for buth
changes
in
problem structure and specific data sources.
The information processing component creates a wide range
reports.
number
bases.
To achieve adaptability and
of applications
this
flexibility,
of managerial
component consists of a
programs that operate on extracted applications data
This structure permits modifications to a particular program or report to
be localized and therefore greatly simplifies the adaptation of the information
generation process.
paths
(Schilling
reports.
et
This component uses visual representations such as value
al.
1982)
and bar graphs
to
augment
traditional
tabular
The system allows easy manipulation of both the representation form as
well as the particular format via a friendly user interface environment.
3.2
The Model
The complex and
utility of a
political nature of the allocations decision highlighted the
rnodcl-based decison support system.
The complexity arose not only
from the great variety of allocations decisions required, but also from
interrelationships.
These issues were addressed by formulating a
linear
their
program-
ming resource allocation model.
The selection of an appropriate model structure was influenced by several
considerations.
First,
the presence of
lay
decision
makers and other non-
technical users favored a model structure which was intuitive and, therefore,
easy to understand.
in
Secondly, due to the prototyping/evolutionary approach used
system development, the model had
to
be capable of extensive elaboration.
Thirdly, the chosen structure should address the multiple objective nature of the
decision problem, namely
Finally,
it
the competing Balanced Service System categories.
was important that the model help strengthen the behavioral
link
between
newly adopted BSS framework and the decision maker's existing
the
perceptions of system- wide needs.
response to these desired characteristics, a goal programming model
In
structure was selected.
Goal programming has been used and tested
in
a wide
variety of multiple objective decision problems with sophisticated user:, as well
as
Such a
novices.
development.
In
straightforward
model structure
can
respond
well
to
an
evolutionary
addition, the multiple B5S objectives couid be represented
fashion
using county-wide
service
needs as goal
in
a
By
levels.
directing the Board's attention towards balancing these services, the behavioral
link
between the BSS framework and a Beard member's current cognitive model
could be improved.
The model formulation follows
discussed
detail in
in
in
programming structure and
Henderson and Schilling (19S1).
model are not germane to
manner
a classic goal
this paper, a brief
is
While the details of this
overview
provided so that the
is
winch the DSS and the model evolved over time can be discussed.
The primary decision variables reflected the amount
of dollars from each
funding source allocated to each service type provided by each agency.
were four different sources of funds to be accounted
for, resulting in
There
over 500
Besides the budget constraints which limited total dollars available
variables.
from each funding source, restrictions were specified on the percentage increase
and decrease that any agency's budget
might change.
county-wide funding level for each service was limited
Similarly,
in
the
shift.
insure
that any allocations shifts would be politically feasible.
defunding an entire agency or
to
total
amount that
These agency and service funding restrictions were included
might
difficult
the
implement.
service
The Board
in
a single year
specifically
it
to
For example,
would be extremely
chose a strategy that would
spread major funding level changes over several planning periods.
-10-
Legal restrictions were incorporated which addressed the legislative and
contractual
stipulations
of
specific
funding
example,
For
sources.
federal
regulations require that the proportion of federal funds to public funds must be
no more than three to one.
Consistent with a goal programming approach, constraints were included
that
measured goal deviations and created an objective function minimizing the
weighted deviations from the BSS goal
levels.
Since none of the goal levels was
attainable given current or foreseeable funding levels, the deviations were
The
one-sided.
tools
weightings of the deviational variables served as the
priority
identifying
for
group conflict and consensus formation as well as the
mechanism by which the group could examine alternative
allocation patterns.
While goal programming has seen numerous applications,
several potential pitfalls.
Of most concern
in this
bounds are added to the model
This activity often
solutions.
in
occurs when
an attempt
the
model
This problem can be particularly troubling
unrealistic.
it
application
of solution manipulation, discussed by Harrald et al. (1978).
arbitrary
all
In
in a
is
the possibility
such a situation,
force acceptable
to
is
nonetheless has
too
simplistic
and
prototyping imple-
mentation effort, where both the DSS and model evolve from a simple, first-cut
system.
In
order
to
changes were subjected
members.
inhibit
to
arbitrary
manipulation,
all
extensive discussion and debate with Board and staff
Changes were introduced only
if
a consensus opinion existed that the
modification was a fundamental policy elaboration.
For example, during
development, two basic model improvements were made.
that the
model ignored differences
residence.
To rectify
proposed structural
this
in
It
initial
became apparent
services based on client age and area of
inadequacy, constraints were added
to
insure
that
each client age group and geographic area received at least a minimum level of
"11-
funding.
another instance,
In
'supplemental') were required-
was determined that some services (termed
when (and only when) other
basic services
mod.il
represented
changes
were not attempts
evolutionary
enhancements
decision maker learning (see section
to
In
worth noting that these modifications are
model
the
to
5).
contrive
but
solutions
which
were
Both of
Constraints were then written to reflect this observation.
purchased.
these
it
fact
in
resulted
from
support of these conclusions
it
is
in
the model three years
often 'process.'
The means by which a
still
present
later.
u..
The Decision Process
In
decision
the public sector, the key
is
word
is
reached can often receive more attention than the decision
designing and implementing a D5S, issues of process
common
may be
perception
among
users
is
that
some
In
A
become paramount.
their decision
of
itself.
making power
For example, one of the byproducts of a model-based DSS
sacrificed.
that decision criteria must be
made more
explicit.
Attention
is
is
then directed
toward the mechanism by which these criteria are established and applied. This
externalization
often
represents a
major change
for
public
sector
decision
makers (Hammond, 1920).
The likelihood of successful implementation
of resultant
change
is
increased as the magnitude
To
decreased (Keen and Scott-Morton, 1979).
minimizing unnecessary process modifications
Franklin County,
is
is
very desirable.
In
the case of
the planning and allocation process involved group
making throughout.
this end,
decision
There was strong commitment among Board members
planning process which utilized an interacting group to obtain a consensus.
Board members
felt
to a
The
such an approach was both politically feasible and enhanced
12-
To avoid the
the opportunity for debate and compromise.
groups,
such as dominance and rutting (Gustafson et
pitfalls of interacting
al.
1973),
an estimate-
discuss-estimate procedure was used to generate goal weights, (Gustafson et
al.,
1973; Delbecq and Van de Ven, 197i).
This process calls for each committee
DSS
analysis.
member
to
review the results of a
Each member then assigns importance points
(the
sum of which
The distribution and mean
equals one hundred) to the various goals. 3
of the
collective votes were tabulated and fedback to the group to stimulate discussion
and promote conflict resolution.
importance points occurred.
be
effective
(Delbecq et
in
estimating
1975).
al.,
Following
this
debate, a second allocation of
This vote-discuss-vote sequence has been shown to
parameters and
for
facilitating
The average weights produced by the second voting were
used as priority weights on the deviationai variables
model.
group consensus
The model was then solved
to
in
the goal programming
generate an allocation pattern, which was
presented to the group and the vote-discussion-vote cycle was repeated.
This group process
model.
It
technique
is,
for
is
the solution technique
determining
makers and
appropriate
the
it
is
weights
on
the
objectives.
to initiate the
As the implementation proceeded,
the
comfortable with interpreting the goal weights.
to directly link
changes
in
Its
easily understood by non-technical deci-
This simple
blends easily into the existing group process.
format provided an effective means
makers
the goal programming
essentially, a multi-party extension of a simple, iterative search
relatively unsophisticated structure
sion
for
DSS prototyping
decision
effort.
makers became quite
The DSS allowed the decision-
weights with changes
in
allocation.
At latter
^Introductory training sessions emphasized the underlying assumption of an
interval scale implicit in the averaging of these important points.
13-
stages
in
the process, minority opinions,
input
from
to further support group debate.
members, were analyzed
of the group
constituences
other
average weights based on a subset
i.e.,
outs de
(originally
the
;
Later,
was easily
process)
incorporated.
5.1
Results
This implementation represents a single data point and, hence, results are
quite tentative.
and
its
critique
However, the study represents an actual DSS implementation,
usage over a three-year period provides a significant opportunity
Two major
DSS concepts.
critical relationship
emerged from
insights
this
study: (1) the
between DSS and the more traditional MIS functions and
the characteristics of third generation
DSS technology,
to
particularly
(2)
DSS technol-
ogy applicable to the public sector.
One general
trend
in
DSS research centers on
the appropriate role of the
end user/decision maker and the ability of the DSS to reduce his/her dependence
on the existing MIS function.
Rockart and Flannery (1981) found
independence to be a major factor behind end user computing.
gulf
between DSS builders and traditional MIS designers
prototyping or adaptive design by DSS builders.
stress features such as
priority
is
This potential
widened by the use of
This design process does not
documentation and ease of maintenance that receive high
from traditional MIS organizations (Keen and Gambino,
Another aspect of
this
based DSS can be built with
requirements.
this desire for
independence
little
or
is
1982).
the notion that effective model-
no attention
to
Alter (1980) suggested a taxonomy of
the "data processing"
DSS
that distinguishes
between "data-oriented" DSS and "model-oriented" DSS. The former emphasizes
storage and access to data while the latter emphasizes formal problem represen-
14-
tations and solution procedures.
simplified,
many DSS
is
over
and current DSS technology reflect
this
While Alter recognizes this distinction
applications
For example, many DSS generators provide primary strength
dichotomy.
modelling with
little or
no capability
in
in
data management.
This research does not support the notion that
DS5 design
continue to
will
be independent of the MIS function. Specifically, the distinction between modeloriented
DSS and data-oriented DSS does not appear
implemented
in
this
study was conceived as model-oriented and
ment efforts emphasized the modeling aspects
of
experience demonstrated that the capability to
link
transaction
data
base
was
critical
throughout
speculate that successful
DSS applications
DSS
processing
to
the
basic
data
will
systems
the
the
initial
the
develop-
And
system.
yet,
the model to the large
prototyping
effort.
generate requirements
in
The DSS
appropriate.
organization.
We
to link the
This
DSS
implementation significantly altered both the data definitions and the data flow
associated with the Board's transaction data systems.
This resulted
in
interdependencies between the DSS user and the MIS organization.
implementation served as a catalyst
to
The structure of the model became
the basis for redesign of the data collection activities.
it
DSS and ensure
The DSS
generate the commitment necessary to
implement a data administration function.
institutionalize the
increased
While this served to help
reliable input data for the allocation model,
also created the need for end-users to work closely with the MIS organization.
As the Board expectations
became more
organization.
sensitive
for data quality increased, the credibility of the
to
the
data
base
maintenance efforts
Thus, the on-going success of the DSS
the effectiveness of the MIS organization.
became
of
the
DSS
MIS
directly linked to
-15-
Hammond
(1980).
Keen (1980) and others have noted the
traditional
reliance of public sector analysts on descriptive data analysis to support the
This implementation emphasized the need for the public
policy analysis process.
DSS
sector model-based
to
provide for descriptive data analysis as well.
had the system been unable to easily respond to
this
Again,
data intensive analysis, the
implementation effort would have suffered.
The study suggests that DSS may provide increased opportunities
innovations
in
the MIS function.
requires different
Much
management and
like
the introduction of
for
new products
technical practices, the design and imple-
mentation of a DSS requires approaches that differ from the more traditional
MIS practices. Yet,
if
successful, the
DSS creates an ever increasing dependence
between the DSS end-user and the MIS function.
the public sector
This seems particularly true in
where the use of such systems may result
precedence setting
in
policies.
A second major
DSS technology,
public sector
insight relates to the characteristics of third generation
particularly as they
may
Future
apply to the public sector.
model-based DSS generators must address at least three needs:
easy incorporation of an equity dimension, enhanced data analysis capabilities
and increased communication capabilities.
As the Franklin County implementation proceeded, system modifications
centered around both the ability to change the model and an ability
information processing and basic data management modules.
changes
in
the model structure focused on issues of equity.
In
to alter
many
cases,
As Savas (1978)
points out, there are a variety of conflicting ways to operationalize notions of
equity.
While
Initially, the
some
model did not explicitly operationalize equity
relationships
indirectly
created
solutions
which
relationships.
were
more
16-
"equitable," they were not explicitly formulated to do so.
minimum
constraints which required
have their origin
levelc of services offered by- agencies
may
the equity notion of equal outputs.
in
However, as the DSS evolved, the board sought
in
For example, legal
the allocation of funds.
to explicitly insure equity
For example, constraints forcing distribution of funds
between geographical areas were added.
These efforts
geographically isolated providers received at least a
to ensure that small,
minimum
allocation repre-
Many
sented the Savas' equity concept of equal inputs per unit area.
discussions
centered around developing constraints that would reflect the Board's concern
for
equal access.
The
ability to generate
model structure,
to easily test and,
eventually, incorporate these structural changes was an important capability.
This need to consider equity issues
demand
for at least a
in
the public sector results
three dimensional model.
in
a technological
One must be
able to easily
accommodate program
activity, time and equity dirnensicns in the models.
suggests
automated spread sheet modeling languages that are
that
current
essentially two-dimensional
in
may
This
be inadequate for end-user system development
the public sector.
Previous discussions addressed the need to link the model-based DSS to the
transaction system of the organization.
This linkage results
becoming an important stakeholder with regard
and maintain elementary data.
must have the capability
mentioned,
to
It
the
DSS user
to procedures to define, collect
also suggests that third generation technology
conduct a wide range of data analysis. As previously
this type of analysis has
sector policy analysts.
in
The need
provide both modeling and data
is
become standard practice
to
for
most public
provide a single DSS that can adequately
management
capabilities.
-17-
DSS technology must place greater emphasis on
Finelly, third generation
This study emphasized the need for alternative
communication capabilities.
modes
presentation,
of
graphics capability
that the
in
i.e.,
DSS
is
However,
widely recognized.
communication needs
for alternative
The incorporation
graphical versus tabular.
for a public sector
modes of presentation.
this study
suggests
DSS extend beyond providing
Public sector analysts have significant
requirements for distribution of the results of their analysis.
This distribution
normally takes the form of reports, memos, and/or press releases.
si
of a
nificant benefits will be gained by linking the
DSS
into the
This suggests
automated office
environment., For example, data related to the model-based DSS should be easily
accessible by the word processing system within the office.
The growing research findings relating
prototyping for
DSS were
to the use of adaptive design
strongly supported by this implementation.
typing design strategy similar to those discussed by
Morton (1979) and others was used
effective,
it
a
need
As the DSS grew
tions.
difficult.
for
in
DSS. While
A proto-
Keen and Scott-
this strategy
proved
for
easy
complexity, meeting these expectations
The modular design of
this
system, which explicitly recognized
model management, proved crucial to meeting these high expecta-
The authors were able
contained high-level
impact on the
components.
(1930),
also created high expectations on the part of the user
modifications.
became
to design the
Keen
or
to evolve a
commands and
command
structure
matrix-generation language with self-
flexible
for
the
user
interface with
little
or
no
data or information management
Further, each component proved necessary to the success of the
implementation.
These experiences suggest that while prototyping
is
successful
as a general strategy, structured design concepts and associated design aids are
quite important.
18-
This study also provided insight into the process used by Board
validate the model and the DSS.
The
stages.
members
to
This process appeared to have three distinct
stage was the acceptance of the conceptual structure for
first
modeling a mental health system (the BSS model) and for utilizing a multicriteria
allocation model.
involving
system,
(1)
(2)
This stage involved fairly abstract debates at the Board level
comparison of the BSS Model with other models of a mental health
reviewing alternative processes for obtaining information about the
impact of allocation strategies, and
(3)
reviewing alternatives for conducting
sensitivity analyses.
The second stage involved a macro-operational
to the
model were varied and trends
outputs of the model
made
in
verification
in
which inputs
output were examined to determine
if
the
intuitive sense or could be logically accounted for.
This stage resulted in structural changes to the model and helped to established
the conient of several
Finally,
sampling.
the
management
third
stage
reports.
involved
through
validation
micro-operational
This consisted of individuals selectively examining input data, model
parameters and tracing outputs at very detailed
Evaluations were
levels.
based on personal experience or independent data sources.
made
For example, a Board
member might
ask to see the unit cost for a particular type of childrens' service
at a particular
agency because he/she had been a provider
At
the
this
point,
origins of these
implementation became linked
parameters
to the actual
to
in
the ability to trace the
day-to-day transactions data base.
inconsistencies were found or
new formulations developed,
base had to be used to provide
new
transaction
that environment.
input to the model.
On
If
the transaction data
several occasions, the
data base was used to clarify demand characteristics or system
demographics that were not explicitly incorporated
into
the model.
Had
this
19-
capability been lacking, concerns about data quality would have impeded
the
implementation process and the DSS would not have been effective. Thus, while
model-based
the
DSS
basically
on
operated
from
extracted
files
large
a
transact ion data base, the ability to easily interact with, the transaction data
base
5.2
played a major role
still
the implementation process.
in
Impact On the Community Mental Health System
It
important to note the impact of implementing this DSS on the total
is
mental health system.
support tool that
For
example,
in
is
As noted
by
Section
2, a
study
this
a
examining
the
decision
to
transfer
were not
and led to
in
a
the
community board was arrived
allocation
models developed
examination showed programmatic independence
facilities)
DSS should serve
as a learning
capable of addressing both strategic and operational issues.
retardation services to another
part,
in
(e.g.,
for
budget
mental
for
at and justified, in
both
areas.
This
no shared resources or
conclusion that community level goals for these two area
conflict.
Similarly, the system
was used
to illustrate the
impact of alternative cost
accounting approaches, to communicate the impact of federal fund-matching
requirements and to examine a wide range of operational-level
have evolved beyond providing direct support
example, extensive "what
contingency planning
if"
for the
in
uses
For
the context of
success cr failure of a proposed tax levy.
to
revamp data collection processes,
new controls over system-wide data flows and
new data requirements.
Its
for the allocations process.
analysis has been performed
The DSS also became a focal point
establish
issues.
As
to
to create or legitimize
a direct result of this system
implementation, a
completely new data collection format was created and a new collection process
20-
This effort
instiiutionalized.
new
established
acquiring data as well as provided a means
validation
lo train
procedures used
in
providers on the BSS model of
the mental health system.
The quantity
of cost and service-related data obtained from agencies
substantially increased (by nearly a factor of two) over previous years.
agencies
were asked,
reductions
,
i.e.,
for
the
time,
first
to
this
The
indicate preferences on budget
where and at what level they would place lower
As might be expected,
was
limits.
new information management
effort led to a
desire for greater control over data quality and expansion of the types Oi data
made
Prior to the implementation of the DSS, such cooperation and
available.
involvement
in
the collection and quality control of data were lacking.
large extent, the
DSS created a planning process
that justified the effort and
Since These data were also used for other
cost necessary to provide such data.
financial and policy analysis
To a
the
tasks,
DSS produced
a significant secondary
impact on Board functions.
the
Finally,
process established
well-defined
points
within
process where priorities were established and decisions made.
continue
to
have
a
fundamental
community can influence
first
time,
direct
impact
allocations.
In
on
the
the
budget
This had and will
mechanism by which the
essence, the Board established, for the
linkages between a conceptual
model of a Mental Health
system, how such a system should function (goals), and the allocation process
employed
6.
to
achieve these goals.
Conclusions
Generalizations cannot be
made from
external validity of these experiences
is
high
a single data point.
in
that the system
However, the
was successfully
-21-
hnplemented and continues
for
"what
if"
to be used both for allocations decisions as well as
planning questions.
Further, these experiences appear generally
consistent with the growing body of research on DSS, and thus merit an attempt
io
draw conclusions.
First, process, i.e., the
is
critical
in
way a system
public sector decision
conclusion of this work
is
or organization arrives at a decision,
making.
Perhaps the most fundamental
management
the need for the
scientist to provide a
process-support aid rather than a model that provides an answer,
i.e.,
a product-
Thus, the importance of providing a range of learning and decision
oriented aid.
aids within an integrated, yet adaptive system
is
stressed.
Secondly, the emerging theory of DSS addresses a blend of design strategy,
system characteristics and required technological building blocks.
supports most current thinking with regaid to these areas.
design strategy proved effective both
as
well as providing a
mechanism
in
Prototyping as a
terms of defining user information needs
and analyst learning.
to support user
system characteristics of adaptability,
This work
flexibility,
The
modularity, simplified man-
machine interface, and alternative modes of presentation proved necessary to
successful
implementation.
Thus,
we
empirical
find
support
for
the
basic
principles of DSS.
Third,
model selection and formulation need careful attention. The model
structure must match the problem structure, but
maker understanding. Institutionalization of
facilitated
model.
when the model serves both
Care must also be
taken
to
a
DSS
it
must also support decision
that uses a complex model
is
as an analytic tool and as a conceptual
circumvent model usage
traps.
In
a
prototyping/evolutionary environment, the analyst must closely scrutinize model
modifications
in
order
to
avoid the temptation of solution manipulation and
-22-
model integrity. Failure
insure
appearing
(av least to the
to
do so can invalidate the entire DSS while
still
untrained eye) to perform correctly.
Fourth, the distinction between model-oriented DSS and data-ori2nted OSS
and the notion of independence
for
DSS
users does not appear appropriate given
This system was conceived as a rnodel-oriented DSS and
these experiences.
experience demonstrated
that
capability
the
transaction data base was very important.
from the need
effective, the
to
We
to provide the
DSS
means
to
That
is,
From
better coordinate
Fifth, the
model
to
a
large
users cannot remove themselves
To be
access this elementary data
in
a
to be successful, there will be pressure,
requirements, to link the DSS to
organization.
the
link
yet,
speculate this will become a feature of most successful
model-oriented DSS generators.
fact
to
examine, verify, and communicate fundamental data.
DSS has
timely fashion.
in
And
efforts were directed toward the modeling aspect of the system.
initial
a
the basic
management perspective,
DSS and MIS design
this will result
in
of
the
a need to
efforts.
importance of addressing equity issues
public policy
processing
data
is
The notion of
stressed.
a major research issue.
This study suggests
equity
in
future
DSS technology must enable the
user to incorporate an equity dimension
as well as activity and time dimensions.
This indicates public sector applications
is,
DSS generator
require a
in itself,
that extends beyond the two dimensional framework
DSS generators. The
currently represented by financially-oriented
generators
demands
in
the public sector
for flexiblity
include
in
least
DSS
three-dimensional increases
and sophisticated forms of presentation.
Finally, the benefits of
suggests
must be at
fact that
the concept of
DSS are
difficult to assess a priori.
value analysis
for
As Keen (19S0)
DSS, the benefits of DSS will
such issues as support of organization change, support of individual
-23-
learning and improved
tions.
This work
management
indicated that significant systemwide impact occurred and
should be explicitly recognised
DSS
effort.
of the technological growth of the organiza-
in
the evaluation of the success or failure of the
The DSS affected fundamental areas such
as learning, organizational
development, data processing and decision process framing.
means
learning by providing the
a
systematic manner.
It
It
influenced user
to investigate the complexity of the
problem
in
affected organisational development by unfreezing
positions and attitudes concerning both the mission of the Eoard as well as the
structure of the allocation process.
felt
It
affected data processing by providing the
need and political support necessary to revamp data collection procedures
and to increase the quality and integrity of their data base.
the
means
to
Finally,
it
provided
frame the decision as one of tradeoffs between goais rather than
increases or decreases
in
specific budget line items.
This not only changed the
allocation decision process but helped to institutionalize a goal-o r iented piann in 6
process.
o
c
CO
CO
o
Z
JSZ
o
1
Li.
K
—
'.;.'_. !L ..*!.
.
'
i
—
!
.
I
i
Model Generation
System
m&mmzat *»tvj*:.<*<
>i
I
L071U
S
3
*
-
•
Ok
m~~,*
Transection
m
$v
-»,
cm
l
p«
/»-,
:.
q
-
V
;
r
s
.-.
rj
,-.;.
r. •.
-it-
r-.i*i«
A
F-\*J'ji !%,'%£
?
tUi
(
Liv-ni
ijw'3
V'-wSsv!
UiiUf!
V
In-f
^j> Sy stem
Control
information processing
Report generation
*f
Management
reports
_
Figure 2.
j
Decision Support System:
Framework
-24-
REFERF.NCES
Alavi, M. and J. C. Henderson,
Decision Support System,"
1323,
"An Evolutionary Strategy
M anagement
Science
Implementing a
for
Vol. 27, No. 11, pp. 1309-
,
November, 19SI.
Alter, Steven, L., Decision Support Systems:
Current Practice and Continuing
Challenges Addison- Wesley, Reading, MA, 1980.
,
Downey,
E.
Brill,
"The
Use
Optimization
of
Management Science
Planning,"
Cohon,
Jr.,
,
Models
in
Sector
Public
Vol. 25, No. 5, (May) 1979, pp. 413-422.
and D. Marks, "Multiobjective Screening Models and Water Resources
3.
Investments," Water Resources Research Vol.
,
9, p.
826, 1973.
Delbeco, A. and A. Van de Ven, "A Group Process Model for Problem Identifica-
and Program Planning," Jou rnal o f A pplied Behavioral Science
tion
p.
,
Vol. 7,
466, 1971.
Andre
Delbecq,
L.,
Andrew H. Van de Ven, and David
for
Program Planning:
Techniques
A Guide
Process , Scott, Foresman and Co., Glenview,
Gustafson,
D.
H.,
R.
M.
Shukla,
A.
Comparative Study of Differences
Gustafson,
G roup
Nominal Group and Delphi
1975.
Illinois,
Delbecq
L.
in
to
H.
and
G.
W.
"A
Walster,
Subjective Likelihood Estimates
Made
by Individuals, Interacting Group, Delphi Groups and Nominal Groups,"
Organizational Behavior and
Human Performance
,
Vol. 9 (1973), pp.
280-
291.
Hammond,
K. R.,
"Toward Increasing Competence
Formation," Judgment and Decision
Hammond,
Harrald,
1.,
J.
Eds.,
AAAS
Leotta,
W.
in
of Thought
in
Public Policy
Public Policy Formation
,
Kenneth R.
Selected Symposium, 1980.
A.
Wallace, and
R.
E.
Limitations of Goal Programming as Observed
Wendell, "A
in
Note on the
Resource Allocation
for
-25--
Marine Environmental Protection," Na val Research Logistics Quarte rly,
Vol. 25, No.
Henderson,
3.
li,
pp. 733-739, 1978.
C. and D. A. Schilling. "Design and Implementation of Decision
Support System:-,
in
the Public Sector," Ohio Slate University, Colic-ge of
WPS
Administrative Science Working Paper Series,
Henderson, John
C,
81-77, 1981.
Michael A. Showalter and Leroy H. Krajewski,
Integrated Approach
Manpower Planning
for
in
Jr.,
"An
Public Sector," Working
Paper Series. Ohio State University, 1978.
Translating Useful Models into
Keen, Peter G. W., "Decision Support System:
Usable Technologies," Sloan Manage ment Revie w, Spring. 1980.
Keen, Peter G. W. and Gambino, T,
J.,
"Building
A Decision Support System: The
Mythical Man-Month Revisited," To appear
Systems,
J. F.
Bennett
P-uiidirg
Decision Support
Addison-Wesley, 1982.
(ed.)
Keen, Peter G. W. and Scott Morton, Michael,
Organizational Perspective
in
S..
An
Decision Support Systems:
Addison- Wesley Publishing Co., Reading MA,
,
1979.
Lamm,
R. A., "The Environment and Public Policy," Judgment and Decision
Public
Policy
Formation
,
edited
by
K.
Hammond, AAAS
R.
in
Selected
Symposium, 1980.
Major, D. and Lenton, R., "Multiobjective Multimodel Riverbasin Planning:
MIT-Agentina Project," Prentice-Hall, Englewood
Ness,
D.,
"Interactive
Systems:
Theories
of
Cliffs,
Design,"
Conference Interactive Information and DSS
,
New
Joint
The
Jersey, 1978.
Wharton/ONR
Dept. of Decision Sciences,
Wharton School, University of Pennsylvania, November, 1975.
Pressman,
J.
and A.
Wildavsky, Implementation:
Washington Are Dashed
in
How
Great Expections
Oakland University of California Press, 1973.
,
in
-26-
ReVelle, C.
EMS
Rockart,
S. et
ah, "r-acility Location Analysis:
A Review
of Context-Free and
Models," Health Se rvice:, Research, Spring, [977.
J.
and
F.
L. S. Fiannery,
"The Management of End-User Computing,"
Proceedings of Th e Second International Conference on Information Sys-
tems December 19S1,
pp. 351-363.
,
Saladin,
A
B.,
Methodology
for
the
Allocation
of
Police
Patrol
Vehicles
.
Unpublished Doctoral Dissertation, Ohio State University, 1930.
Savas, E.
S.,
"On Equity
in
Providing Public Service," Management Scien ce, Vol.
2k, No. 8 (April) 1978, pp. 800-808.
Schilling,
D.
A.
et
Alternatives
al.,
in a
"Design
and Analysis of Location and
Relocation
Fire Protection System," Proceedings of IEEE Modeling
and Simulation Conference, Pittsburgh, May, 1980.
Schilling D. A., ReVelie, C. S.. and Cohon, 3. L.,
"An Approach
to the
Display
and Analysis of Multiobjective Problems," Socio and Economic Planning
Science
,
in
press, 1982.
Sprague, R.H., and Carlson, E.D.,
Building Effective Decision Support Systems ,
Prentice Hall, Englewood Cliffs, N.3., 1982.
~
[*953
036
3
TDflO DDM
MflE
EflM
OEC2
1989
FEB 10
i
«•
Lib-26-67
fzA£.dx£i££ &h.
BASEMENT
~&A1£^ <f^/efe.