You are on page 1of 21

Transpn Res.-C, Vol. 5, No. 6, pp.

349±369, 1997
# 1998 Elsevier Science Ltd. All rights reserved
Pergamon Printed in Great Britain
0968-090x/98 $19.00+0.00

PII: S0968-090X(98)00002-3

GROUP-BASED GEOGRAPHIC INFORMATION SYSTEMS FOR


TRANSPORTATION IMPROVEMENT SITE SELECTION

TIMOTHY L. NYERGES,* ROBB MONTEJANO, CAROLINE OSHIRO and


MATTHEW DADSWELL
Department of Geography, University of Washington, Seattle, Washington 98195, U.S.A.

(Received 21 January 1997; in revised form 17 December 1997)

AbstractÐTransportation improvement site selection exempli®es transportation decision making that is col-
laborative in nature and geographically based. Such decision-making is part of a broad societal trend toward
shared and participatory discussions about public investment. Perspectives from three di€erent transporta-
tion decision contexts in the Puget Sound Region of Washington State, a regional council, a county govern-
ment and a public±private Coalition group, are combined with a literature review to develop a decision task
model that summarizes the need for information technology support during transportation improvement site
selection. The task model guides the development of a decision support system requirement speci®cation that
outlines integrated information capabilities provided by geographic information system (GIS) and group
support system (GSS) technologies. Together, GIS and GSS capabilities contribute to evolving group-based
GIS. The kinds of capabilities a group-based GIS could o€er in addressing the needs are identi®ed. A report
on the use of a prototype, group-based GIS called Spatial Group Choice highlights the possibilities in an
inter-organizational coalition decision context. The conclusions discuss needs for future technology develop-
ments and social±behavioral science studies on these developments. # 1998 Elsevier Science Ltd. All rights
reserved

Keywords: geographic information system, GIS, transportation site selection, groupware, group decision-
making, transportation improvement projects, collaboration, stakeholder.

1. INTRODUCTION

Preparation of a transportation improvement program (TIP) is a rather complex group activity.


The activity can involve elected ocials, planners, analysts, and concerned citizens, and take
months to complete, every two years. A TIP consists of a set of transportation projects proposed
for development on a `roll-over' basis, i.e. projects get added to a six-year programming budget to
be designed and then constructed, with those constructed getting `rolled-o€'. Cities, Counties,
Metropolitan Planning Organizations, and State Departments of Transportation are tasked with
sorting through the myriad of project proposals to determine the best ones for funding. The
activity is monitored constantly, as it should be consistent with a region's long-term transportation
plan. Because of the complexity of the group-based TIP decision problem, and the current devel-
opments in geographic information systems (GIS) for transportation (GIS-T) (Dueker and Vrana,
1992; McCormack and Nyerges, 1997; Nyerges, 1995a; Nyerges and Dueker, 1988; Vonderohe et
al., 1993), it is assumed that the TIP process can be enhanced through the use of group-based GIS-
T capabilities (Nyerges, 1995b).
The purpose of this paper is to examine and report on the potential for geographic information
technology to support group-oriented transportation improvement site selection as an example of
transportation decision-making, and suggest directions for technology development. It is the
intent of this paper to establish what enhancements are possible now in the form of group-based
GIS, and what capabilities might have to wait for future technology development and integration.
This purpose and intent are not directed at how GIS-T can provide `indirect' support for the
transportation decision-making process, as that kind of support has been provided for years in the
form of computer-based analysis and maps. Rather the purpose and intent is directed at how GIS-
T might support the process in a `direct' manner, helping decision-makers to reduce the `cognitive

*Author for correspondence. Fax: 206-543-3313; e-mail: nyerges@u.washington.edu

349
350 T. L. Nyerges et al.

complexity and e€ort' of their information handling by placing GIS-T on the desktop for
exploratory decision-making and to support meeting environments where information technology
can provide improvements in meeting process and outcomes (DeSanctis and Gallupe, 1987;
Nyerges, 1995c).
In the 1970s, when it was recognized that management information systems did not address
decision-making contexts very well, decision support system (DSS) technology was proposed
(Sprague and Watson, 1986). The original emphasis was on enhancing the accuracy of decision
outcomes, but in the 1980s emphasis turned to enhancing decision process as well (Benbasat and
Nault, 1990). During that time in the 1980s several researchers realized that group interaction was
very much a part of the decision process, hence group support systems (GSS) were devised, and a
special type called group decision support systems (GDSS) were developed (DeSanctis and
Gallupe, 1987).
In a similar manner, when GIS were recognized as not supporting spatial decision-making in a
direct way, spatial decision support systems (SDSS) became the topic of much interest for devel-
opment and study (Densham, 1991). It has been argued that GIS is the core technology in SDSS,
but specialized techniques are needed to support decision-making (Densham, 1991). Currently,
GIS and their o€shoot in SDSS are not designed to support groups, but many applications involve
groups. Therefore, interest in designing and developing GIS and SDSS for groups (SDSS-G) is
growing. Armstrong (1993) o€ers several suggestions for extending the development of GIS to
support locational problem solving oriented to groups, and Reitsma (1996) provides insight into
the structure and support of decision-making using a group-based SDSS for water resource man-
agement. In the transportation literature Kanafani et al. (1994) propose a system design for intel-
ligent transportation systems (ITS), as part of a re-examination of urban transportation planning
in a group context. Vlahos et al. (1994) elucidate the role of teamwork in an ITS planning meth-
odology. Research about TIP decision-making is part of a broader research agenda on investiga-
tion of software techniques used in collaborative spatial decision-making (Densham et al., 1995;
Jankowski et al., 1997; Nyerges, 1995c; Nyerges and Jankowski, 1997).
This paper views the TIP decision problem as a site evaluation and selection problem that is
socially and politically grounded and can be addressed with the use of multicriteria decision
(MCD) models integrated with GIS and GSS technology. MCD models permit individuals or
groups of users to evaluate several criteria describing several alternative choices (Carver, 1991;
Jankowski, 1995; Jankowski et al., 1997). Developments of MCD models have grown out of the
decision sciences and operations research literature (Hwang and Lin, 1987), but that literature has
rarely addressed geographical problems.
In the decision sciences literature concerned with MCD models, it has been reported that despite
the original interest in using a DSS for obtaining more accurate decisions, it has been found that
changes in cognitive workload (e€ort) may be the real bene®t of introducing decision aids (Todd
and Benbasat, 1994a,b). Consequently, gains from decision support technology come from
reductions in cognitive e€ort, enhancements of decision accuracy, and/or reductions in process
loss. Reducing the complexity of issues through appropriate management, analysis and visualiza-
tion of information might lead to better decisions, because the task of making the decisions is
simpli®ed. Having access to more comprehensive information sources, as well as analytical soft-
ware, at decision-making time can lead to more accurate decisions. In this research, interest is on
reduction in cognitive e€ort, which in turn may lead to more accurate assessments of projects
being proposed.
Although the GIS/SDSS literature is full of case studies of how GIS applications can be used
indirectly to support decision-making, few empirical studies on how individuals or groups directly
use GIS or SDSS have been performed (Davies and Medyckyj-Scott, 1995, 1996; Nyerges, 1993),
and even fewer yet focusing on GIS or SDSS use in a decision-making context (Nyerges, 1998). In
uncontrolled empirical studies generalizing on decision processes, Knapp (1995) and Armstrong
and Densham (1995) show how task analysis can be used to describe user procedural ¯ow in
decision processes. The former in a study of a scienti®c visualization task concerning earth science
and the latter in a facility siting task. In an ethnographic study of GIS specialists at work, Davies
and Medyckyj-Scott (1996) make use of codings from videotape data to ®nd a relationship
between errors committed by users and the conceptual model they hold of a problem. In a con-
trolled empirical study, Crossland et al. (1995) report on individual decision outcomes with GIS
Transportation improvement site selection 351

use for retail site selection, noting that GIS is de®nitely an important enhancement in the outcome.
Also in a controlled study, Reitsma et al. (1996) report on small-group use of a simulation model
in a SDSS for water resource management decision-making context, but describe mixed results
with their three-person groups due to the technical complexity of the model. In regards to group
size, it is also useful to point out that Vogel (1993) reviewed several experimental studies and
concluded that groups of ®ve or more have a better chance of showing improvements in group
process, whereas groups of three or four might show only negligible bene®ts from the introduction
of GSS technology.
At this time we can not tell if the same results concerning cognitive e€ort, decision accuracy, or
process loss will apply to spatial decision contexts, since few studies have been conducted. Ober-
meyer and Pinto (1994) hypothesize that near future use of GIS among large numbers of groups
focusing on controversial topics will most likely increase the con¯ict between groups with di€erent
interests, e.g. in land use decision-making, rather than reduce con¯ict. This is likely to occur
because participating groups often tend to see GIS as presenting information as if it were facts,
which might tend to increase cognitive con¯ict since fact interpretation is based on interests.
However, Obermeyer and Pinto (1994) and others (Brill et al., 1990) see a potential in GIS in the
longer term to decrease interest con¯ict during decision-making through the generation of multi-
ple, alternative decision scenarios. Extending the arguments made by the above authors, we sug-
gest that when basic GIS decision aiding techniques, in the form of data management for large
volumes of data, computer mapping for visualization, and spatial analysis for robust analytic
work, are combined with decision aiding techniques such as MCD models, a likely result is not
only a reduction in cognitive e€ort concerning complex TIP decisions, but also an increase in the
`discussion eciency' concerning con¯icts of interest. Development of advanced systems and tests
of their usability should precede experimental studies, hence the need for the research reported in
this paper.
The basic plan of this paper is as follows. First, we scope the need for geographic information
technology by describing TIP site selection decision processes using three examples, one from a
regional council perspective, a second from a county government perspective, and a third from a
private±public coalition. Those perspectives and material from literature review are used to devise
a task model that summarizes the relationship between technology and the decision processes.
Phases of a TIP decision process are outlined within the task model, providing a basis for assessing
the potential needs for group-based geographic information technology. In the third section, the
system capabilities for group-based GIS technology are speci®ed to address those potential needs
at each phase of the decision process. The capabilities are discussed in terms of what current and
near-term information technology solutions might be available to implement them. Examples of
software use come from a demonstration project that took the form of a software usability review
of a group-based GIS called Spatial Group Choice as applied to the selection of transportation
improvement sites that enhance freight mobility. Finally, in the last section we consider the longer-
term issues concerning a comprehensive approach that addresses needs for transportation
improvement site selection, the information technology to support it, and implications for social±
behavioral studies.

2. TIP PROJECT DECISION-MAKING AS SITE SELECTION

In the following discussion it is assumed that there is, and will continue to be, a general societal
need to undertake transportation improvement project site selection in the U.S., as well as many
other countries around the world. In the U. S., whether funding is expanded, stays the same, or is
more constrained because of competing societal resource allocations, the pervasive need to
undertake transportation site selection will continue across federal, state, regional and local
jurisdictions. With enhanced funding more projects are usually proposed, hence more alternatives
are likely to be considered. Within the current funding levels for transportation in the U.S. the
information complexity, hence cognitive load on decision-makers, including elected ocials,
planner analysts, and citizens continues to be quite high. With fewer funds, the scrutiny becomes
more intense because transportation needs remain high due to the diversity of needs, but fewer
funds are available to satisfy those needs. In the following subsections, two sources of literature
are used to characterize the need for group-based GIS in TIP decision contexts in order to develop
352 T. L. Nyerges et al.

a task model that summarizes the need. The ®rst is material from organizations within the Puget
Sound Region that describes TIP decision processes and criteria. The second is research literature
on decision-making, with special emphasis on preference/choice decisions and site selection.

2.1. An overview of TIP site selection decision-making


Although TIP decision processes among transportation organizations are similar, there are a
number of variations in process and criteria used to identify, evaluate and select transportation
improvement projects for funding. Three examples are treated below to give a sense of the simi-
larities and di€erences, with the ®rst treated in more detail than the other two.
The Puget Sound Regional Council (PSRC) is responsible for coordinating transportation
improvements that have a regional impact within a four county area consisting of King, Kitsap,
Pierce and Snohomish Counties (1996). In 1995, during preparation of the 1996±1998 TIP, pro-
jects were proposed for regional or local funding consideration by regional organizations, counties
and cities. The counties played a larger role than in previous review processes such that counties
helped determine which of the projects moved forward. This was possible because each of the
counties had certain set-aside money to be spent within the county based on a combination of
population-road miles.
At the regional level of decision-making about transportation projects, much of the direction in
1995 came from the 1991 U. S. Federal Intermodal and Surface Transportation Eciency Act
(ISTEA) which forms the basis of the federal funding guidelines (`Intermodal Surface Transpor-
tation Eciency Act of 1991'). ISTEA was passed with the intent to broaden and strengthen the
ability of urbanized areas to link their comprehensive planning programs to funding decisions on
transportation projects. ISTEA promotes the development of transportation systems embracing
various modes of transportation in a manner which eciently maximize the mobility of people and
goods within and through urbanized areas and minimize transportation-related fuel consumption
and air pollution. Metropolitan planning organizations, in cooperation with the respective States,
are mandated to develop transportation plans and programs for urbanized areas of the States. The
process for developing such plans and programs provides for consideration of all modes of
transportation and is continuing, cooperative, and comprehensive to the degree appropriate,
based on the complexity of the transportation problems.
To address the ISTEA mandate, the PSRC has developed a Metropolitan Transportation Plan
(MTP) (Puget Sound Regional Council, 1995a), and undertakes a TIP to select transportation
projects in line with certain ISTEA factors and the MTP (Puget Sound Regional Council, 1995b).
In developing the regional TIP to carry out that mandate, ISTEA requires that each regional/
metropolitan planning organization consider multiple factors as stated in the ISTEA to ensure
that projects follow funding guidelines.
To avoid unnecessary work on the part of proposers from regional organizations, counties and
cities, the PSRC implemented the ISTEA requirements in terms of a two-stage process. The ®rst
stage was a screening process, to cull the most signi®cant projects from the list. Proposers prepared
a 2±3 page document for each improvement project, in which they were asked to address six
`policy criteria':

1. improve mobility within the VISION 2020/MTP types of urban centers or along corridors
connecting centers
2. support adopted regional economic strategies
3. improve system performance, eciencies, and e€ectiveness
4. reduce reliance on single occupant vehicles
5. improve existing or provide new intermodal access to ports, airports or centers, and
6. improve air quality.

For the most part the criteria attempt to establish the `regional impact' of the project. Of the 109
projects proposed and requesting $262 million, 60 made it through that screening process with a
total request of $185 million.
The second stage process was meant to further clarify the regional implications of the project
with regard to policy concerns established by ISTEA and the MTP, by providing more technical
detail for each project. The additional ten criteria used in the second stage were:
Transportation improvement site selection 353

1. maintenance and preservation


2. trac congestion
3. mobility/connectivity
4. safety and security
5. eciency and reliability
6. accessibility
7. economic bene®t
8. air quality/energy savings
9. other environmental bene®ts, and
10.cost assessment.

Of the 60 projects considered from the ®rst stage, 18 projects were funded totaling $43 million.
It is clearly evident from these lists, that establishing and using criteria for decision-making was a
major part of the decision-making e€ort.
Transportation decision-making in King County, Washington is part of the County's capital
improvements program (CIP) decision-making process. The CIP is a six-year program with
updates every two years, including updates for transportation projects (King County, 1996). The
current CIP covers the period 1996±2001. The King County Oce of Transportation Planning
uses a three-step process for transportation project prioritization:

1. needs identi®cation and screening


2. technical evaluation and ranking, and
3. evaluation of non-quanti®ables (King County, 1993).

With the advent of ISTEA, and the recent update of the PSRC process, King County expects to
continue in general with their three-step process, but have tuned it to incorporate the ISTEA
factors.
In a third example, the Duwamish Coalition is a group of ocials from state, county, city, pri-
vate, and community interest organizations that have come together to foster the development of
high-wage skilled jobs while considering gender and minority issues, preserve and reclaim indus-
trial land, and foster water quality enhancement and habitat development in the Duwamish Cor-
ridor. The Duwamish Corridor, approx. 8500 acres in size, is the core industrial manufacturing
area of the Metropolitan Seattle area extending from the southern edge of downtown Seattle to
just north of the Seattle-Tacoma airport, including portions of the cities of Seattle and Tukwila,
portions of unincorporated King County, and approx. 13 miles of the Duwamish Waterway
(Duwamish Coalition, 1995).
The Coalition formed ®ve subcommittees to participate in the coalition activity: Job Creation,
Preserve and Reclaim Industrial Land, Regulatory Issues, Infrastructure Investment, Water
Quality and Habitat Enhancement. The goal of the Infrastructure Investments Subcommittee is to
`identify needs, set priorities, and solve speci®c problems relating to infrastructure that serves the
corridor' (Duwamish Coalition, 1996a). Like the Coalition as a whole, the Infrastructure Sub-
committee is comprised of representatives from local businesses, government agencies, the Port of
Seattle, and residential communities. These groups have a vested interest in the outcome of the
infrastructure decision-making process, and hence can be considered stakeholder groups. The
interests of the stakeholder groups are varied, and often at odds with each other. As part of the
Infrastructure Subcommittee's e€orts, they released two reports, an Infrastructure Assessment of
the Duwamish Corridor followed by a Freight Mobility Report. The latter report includes 24
infrastructure improvement sites considered for enhancing freight movement in the corridor.
The above three transportation improvement project decision contexts lay out many of the basic
issues facing regional and local transportation decision-makers. In the next section we draw the
common elements together, and add other issues to develop a task model as a framework for
understanding how group-based GIS might be useful for transportation improvement site selection.

2.2. A task model summarizing transportation improvement site selection decision-making


Since most transportation projects are localized in nature, given the funding and development
constraints, they can be likened to sites. Although laws and transportation policies set out
354 T. L. Nyerges et al.

categories for types of projects in a non-spatial manner, all projects eventually manifest location-
based concerns. Consequently, the decision problem becomes one of site selection, based on cri-
teria evaluation. It is important to keep in mind that criteria cover a variety of interests and per-
spectives relevant to all projects, and thus location is only one of many criteria. However, a
location-based perspective provides a natural means of evaluating the in¯uences that projects have
on surrounding areas, and the in¯uences that surrounding areas have on improvements.
Using the description of the TIP processes from the previous section, and a review of the lit-
erature on choice-based decisions (Nyerges and Jankowski, 1994, 1997; Nyerges, 1998), it is pos-
sible to develop a task model for summarizing the improvement project decision process (Fig. 1).
This task model summarizes the process-based needs for information technology support, and
allows us to establish more easily a requirement speci®cation for group-based GIS to meet those
needs. The task model depicts transportation improvement site selection as a process of social±
political interaction among participants in a group (Box 4 in Fig. 1).
As with any decision process, it is important to avoid oversimpli®cation in transportation deci-
sion-making. However, it is useful to abstract information about the process and simplify it in
order to be clear about this topic. The decision process is presented here as a normative one based
on a synthesis of the material presented in the previous section of this paper and material from the
group process and group decision-making literature (DeSanctis and Gallupe, 1987; Gray, 1989;
Hwang and Lin, 1987; McGrath, 1984; Simon, 1976; Simon and Associates, 1992). The normative
process for transportation improvement site selection can be described in terms of three general
steps: an initial problem screening where criteria are identi®ed; the use of criteria during project
evaluation, and the selection of alternatives through socio-politically charged conversations,
remembering that such a process of social interaction occurs among many people (Table 1). This
three step process identi®ed here is much like that presented by Simon (1976) as a general process
of ``intelligence, design, and choice'', respectively, with the updated recognition that each of the
steps is in¯uenced by multiple values when placed in a broader stakeholder group setting (Simon
and Associates, 1992). Bhargava et al. (1994) recognize that each of the three steps of intelligence,
design and choice, are themselves each composed of a sequence of ``intelligence, design and
choice'' subprocesses. This latter suggestion seems to apply to the decision processes reviewed in
our site selection research.
As noted above each of the three steps has several phases. The ®rst step, an initial screening of
transportation projects, has three phases. In Phase 1, a group or groups as representatives of
organizations establish basic criteria for preliminary project screening based on published man-
dates. For example, the Duwamish Coalition used committee members' assessment of the projects
to establish a degree of freight mobility enhancement with related impacts on economic vitality,
safety protection, and environmental protection. In Phase 2, a group identi®es a comprehensive
list of projects, collected from various sources, as suggestions for capital improvement. For

Fig. 1. A task model for summarizing the in¯uence of group-based GIS technology in support of TIP project selection.
Transportation improvement site selection 355

Table 1. Phases in a decision process for TIP improvement site selection

Initial screening of projects


Phase 1: Establish basic criteria for preliminary project screening based on mandates
Phase 2: Identify a comprehensive list of projects, collected from various sources as suggestions for capital improvement
Phase 3: Screen projects according to the basic criteria established in Phase 1
Re®ned evaluation
Phase 4: Re®ne criteria to evaluate the projects passing initial screening process
Phase 5: Describe aspects of the projects in more detail, using the criteria and weights from Phase 4, as well as any other
concerns that need to be made known
Selection of projects
Phase 6: Consider unusual circumstances in the evaluation of the projects
Phase 7: Compare the project alternatives against each other to determine those most worthy of consideration
Phase 8: Negotiate for the most important projects

example, the Duwamish Coalition makes use of the list of infrastructure projects identi®ed by the
member groups of the committee, many of which come from the PSRC and King County TIP
lists, although not necessarily. In Phase 3, a group or groups as a committee screen projects
according to the basic criteria established in Phase 1. All projects must meet a minimum quali®-
cation to be considered. Projects are then grouped according to type of project. For example, the
Duwamish Coalition would like to see projects related to freight mobility address their mission
statement.
After projects have passed an initial screening they enter a step 2 of re®ned evaluation process,
consisting of two phases. In Phase 4, criteria are re®ned, or added so that the projects that passed
the initial screening process may be evaluated. Organizations assign weights to the criteria to
indicate the signi®cance of some criteria in comparison to others, and thereby indicate stakeholder
priorities. They can cluster the criteria to determine the commonality in the criteria. For example,
PSRC has synthesized the more technical of the ISTEA factors into a second set of criteria that it
uses to evaluate projects in more detail, with special emphasis placed on the potential regional
impacts associated with the project. However, the Duwamish Coalition had a roundtable discus-
sion of what was thought to be the most important characteristics for projects, but did not speci-
®cally enumerate a list. In Phase 5, groups describe the spatial, attribute and temporal aspects of
the projects in more detail, using the criteria and weights established in Phase 4, as well as any
other concerns that need to be made known. For example, the PSRC requests a detailed descrip-
tion of the projects that passed the initial screening test. King County uses a data-based approach
to compute a score for each of the projects so they can be rated on a quantitative basis. The
Coalition created a list of projects without detailing all of the criteria used but did identify various
important criteria to evaluate each project.
The third step of the normative process, selection of alternatives, proceeds through three phases.
In Phase 6, organizations consider unusual circumstances in the evaluation of the projects. For
example, PSRC might identify a special opportunity that suggests some projects are to be called
out as highly desirable. King County recognizes that all criteria can not be reduced to numbers,
and that other factors need be considered such as special budgetary considerations or emergency
need. In this context, the Duwamish Coalition's interest in freight mobility actually represents a
special consideration for the PSRC and King County. In Phase 7, a group or groups as a com-
mittee compare the project alternatives against each other to determine those most worthy of
consideration. The Coalition created a list of the projects thought to bene®t freight mobility, but
did not enumerate all characteristics as the task was more dicult than they had sta€ time to
undertake. In Phase 8, groups negotiate the most important projects. All of the criteria employed
and categories identi®ed in the preceding phases are used for establishing an idea of the high
priority projects. The projects are considered in terms of their contribution to overall impact and
cost. The results of the PSRC process are published in the 1996±1998 TIP Report (Puget Sound
Regional Council, 1996). The results of the King County process are published in the 1996±2001
CIP Report (King County, 1996). The results of the Duwamish Coalition process are published in
the Freight Mobility Report (Duwamish Coalition, 1996b).
It is to be recognized that the internal dynamics of the process described above (as situated in
Box 4 of Fig. 1) occur over a long period of time. Poole (1985) helps us understand that such
356 T. L. Nyerges et al.

processes may in fact evolve in one or more of four di€erent styles: assumed consensus, political,
rational, and reactive. If information is not adequately treated in any given phase, it is likely that
the group will return to that phase at some point in time to deal with it more adequately as moti-
vated by one or more decision-makers in the group.
The internal dynamics of the decision process described above are in¯uenced by social±political
concerns. It is important to recognize that the character of the organizational environment in
terms of which organizations are participating and the organizational guidelines they follow,
in¯uence the structure of the interaction (Box 2 in Fig. 1). Clearly, the sets of organizational
guidelines that are adopted have a signi®cant in¯uence on the overall ¯ow of a decision process.
In addition, the group personality, in terms of `what stakeholder groups' and `what representa-
tives from what groups', in¯uence the nature of the interaction (Box 3 in Fig. 1). The values
underlying the perspectives of various groups involved in the process are a major in¯uence in set-
ting weights on the criteria, hence in¯uence the prioritization of sites in a major way. Together,
these four major constructs (Boxes 1±4) in¯uence the decision outcomes (Box 5) and the socio-
political outcomes (Box 6) of the decision-making processes. Decision outcomes, in terms of
which projects are chosen as part of the process, can be assessed in terms of eciency, e€ec-
tiveness and equity. Socio-political outcomes are the relationships among organizations that
(re)develop as a result of the TIP process, and likely enhance or hinder the next round of
decision-making.
An advantage of setting up a task model for a decision-making process is that it is possible to
provide details for the process in an incremental manner and develop information technologies
that facilitate the workings of such a process in an evolutionary manner. The purpose of this
model in the context of this paper is to establish which phases of the decision process might make
use of decision aid capabilities under current technology, and which phases might be better served
with future capabilities. To address this issue it is important to know more about the types of
decision aid techniques in a group-based GIS that might support TIP decision-making. Next we
turn to the information technology capabilities that might support the transportation decision-
making process as articulated above.

3. GROUP-BASED GIS REQUIREMENTS FOR TIP SITE SELECTION

System requirements are often expressed in terms of the basic capabilities that are needed to
address information needs, and in this case decision process needs. Most would agree that the
three core software technologies in a GISÐdatabase management for handling large volumes of
data, computer graphics for creating visual representations, and spatial analysis for deriving rela-
tionships among dataÐprovide the core technology for decision aiding techniques in a SDSS.
However, additional decision aiding techniques used for support of decision-making contexts dif-
ferentiate SDSS from GIS (Densham, 1991). Two additions are analytical process modeling and
choice (preference) modeling. The former is useful for transportation demand modeling as part of
transportation planning, whereas the latter is of central interest for a site selection task such as TIP
decision-making, particularly in the context of MCD models. In addition to SDSS capabilities, it is
recognized that capabilities from GSS can provide part of the core capabilities needed in a GIS/
SDSS for groups (Armstrong, 1993). As such, capabilities from SDSS plus GSS can provide a set
of core capabilities to support TIP decision-making, whether the system is called group-based
GIS-T, SDSS-G, collaborative GIS, or collaborative planning systems.

3.1. Potential decision aid techniques for a group-based GIS-T


A set of potential decision aid techniques (Box 1 in Fig. 1) that might apply to group-based GIS-
T appears in Table 2. The techniques have been synthesized from a combination of literature in
DSS (Benbasat and Nault, 1990; Zachary, 1988), SDSS (Densham, 1991), GSS (DeSanctis and
Gallupe, 1987; Coleman and Khanna, 1995), and the as yet ¯edgling literature on collaborative
GIS (Armstrong, 1993; Jankowski et al., 1997; Nyerges, 1995b; Shi€er, 1995).
The techniques in Table 2 are listed in order of most basic to most sophisticated. Level 1 tech-
niques are likely to be used the most, and therefore, appear in software packages as basic support
as they satisfy a basic cognitive need for information manipulation. Whereas, techniques in Level 2
are specialized enough to be needed only by some groups who have a requirement for data analysis.
Transportation improvement site selection 357

Table 2. Decision aiding techniques for a group-based GIS-T. (adapted from material in DeSanctis and Gallupe 1987)

Level 1 : Basic information handling support


(a) Information management : storage, retrieval and organization of transportation data and information
(e.g. distributed database management system support)
(b) Representation aids : manipulation (analysis) and expression (visualization) techniques for a speci®c part of a
transportation problem (e.g. shared displays of charts, tables, maps, diagrams, matrix and/or other representational
formats)
(c) Group collaboration support : techniques for idea generation, collection, and compilation; includes anonymous
input of ideas, pooling and display of textual ideas, and search facilities to identify possible common ideas, (e.g. data
and voice transmission, electronic voting, electronic white boards, computer conferencing, and large-screen displays)
Level 2 : Decision analysis support
(d) Process models : computational models that predict the behavior of real-world processes, (e.g. simulation models
of trac congestion, or air pollution contribution based on trac volume)
(e) Choice models : integration of individual criteria across aspects or alternative choices, (e.g. multicriteria
decision models using multiattribute and multialternatives for systematically weighted rankings or preferences)
(f) Structured group process techniques : methods for facilitating and structuring TIP decision-making, (e.g.,
automated Delphi, nominal group techniques, electronic brainstorming, and technology of participation)
Level 3 : Support for group process consistency
(g) Judgment re®nement/ampli®cation techniques : quanti®cation of heuristic judgment processes (e.g. Bayesian
analysis, social judgment analysis for tracking each members judgments for feedback to the individual or group,
and sensitivity/trade-o€ analysis for comparing transportation project alternatives)
(h) Analytical reasoning methods : perform problem speci®c reasoning based on a representation of the
transportation decision problem, (e.g. using mathematical programming or expert systems guided by automatic
mediation, parliamentary procedure, or Robert's Rules of Order, identifying patterns in reasoning process)

Level 3 techniques are the most sophisticated, and are rather complicated when it comes to
development. Overall, the set of techniques has been developed as stand-alone capabilities in the
past, and are now being brought together in a synergistic manner for enhanced user bene®t.

3.2. Required techniques for the TIP decision processes


Identifying the appropriate techniques for TIP decision-making requires us to relate the poten-
tial techniques (Box 1) to each of the phases in the decision-making process (Box 4), with con-
straints provided by the nature of transportation decision tasks and organizations (Box 2), plus the
personalities of transportation decision-making groups (Box 3). The potential techniques applic-
able to the decision process become the design considerations for a group-based GIS for TIP
decision-making. In the spirit of the framework devised by DeSanctis and Gallupe (1987) for
GDSS, the decision aid techniques of Table 2 are cross-referenced to the phases listed in Table 1,
the result being the decision support requirement speci®cation outlined in Table 3.
Information management and representations appear in almost every phase, as these techniques
are fundamental to reducing the cognitive e€ort of decision-makers to maintain an inventory of all
projects. Multicriteria models are central to the selection and re®nement processes, as a basic type
of analysis integral to the TIP decision process. The reasoning methods are a bit beyond the cur-
rent expectations of decision-makers for socio-psychological reasons. The only decision aid tech-
nique that was not identi®ed for use was a process model, i.e. simulation models for detailed
analytical predictions. However, animated sequences of displays as in multimedia systems may
become a useful information display technique. Therefore, visual simulation could be quite useful,
although it may not be necessary.
It is important to point out that a conventional meeting venue, i.e. one characterized by face-
to-face interaction in a same-time and same-place decision activity (Jarke, 1986), is assumed as
the context for the above-suggested use of decision aid techniques. However, relaxing those
assumptions about time and place could have a major in¯uence on the usefulness of the techni-
ques. Implementation of such techniques in a distributed-meeting, i.e. di€erent time and di€erent
place setting, increases the challenge for system development. A reduction in face-to-face interac-
tion is likely to require more support from software to assist in managing the distributed
interaction. With a face-to-face meeting in mind, but without eliminating the possible use in a
distributed meeting setting, we turn to describe the current and near-term availability of
capabilities.
358 T. L. Nyerges et al.

Table 3. A requirement speci®cation for group-based GIS-T techniques

Initial screening of projects


Phase 1: Establish basic criteria for preliminary project screening
Level 1: Basic information handling support
(a) Information control for maintaining list of issues to be consider as part of review criteria
(b) Representation of di€erent issues under consideration for project review
(c) Collect ideas and synthesize a category of what is important for basic criteria
Level 2: Decision analysis support
(f) Getting the group to interact for positive contribution to the overall goal of establishing the most
important criteria to use on project review
Level 3: Group process consistency
(g) Detail the description about each criterion to be used for project review, and determine how close this
understanding is to others in the group
(h) Examine the consistency of establishing the criteria
Phase 2: Identify a comprehensive list of projects, collected from various sources as suggestions for capital improvement
Level 1: Basic information handling support
(a) Retrieve project suggestions from various sources
(b) Depict a representation of what documents came from where organizationally, and what projects apply to
where geographically
(c) Collect ranked ideas to see which ones are the most important from a group perspective
Level 2: Analysis support (not needed)
Level 3: Group process consistency (not needed)
Phase 3: Screen projects according to the basic criteria established in Phase 1
Level 1: Basic information handling support
(a) Retrieve project information
(b) Display maps of projects
(c) Share ideas about the di€erent projects
Level 2: Decision analysis support
(e) A MCD model can be used to eliminate projects, or compare them if some in the group are unsure how one
compares against each other
Level 3: Group process consistency (not needed)

Re®ned evaluation
Phase 4: Re®ne the criteria to evaluate the projects that passed the initial screening process
Level 1: Basic information handling support
(a) Use information management techniques to develop a database of criteria and weights
(b) Graphics to depict weights are useful
(c) Collect the raw data to create the weights
Level 2: Decision analysis support
(f) Group decision techniques might be employed to come to agreement on weights
Level 3: Group process consistency
(g) Sensitivity analysis might be useful to identify inter-relationships between criteria
Phase 5: Describe the spatial, attribute and temporal aspects of the projects in more detail, using the criteria and weights
established in Phase 4, as well as any other concerns that need to be made known
Level 1: Basic information handling support
(a) Data base of values retrieved for each of the projects
(b) Show the projects on a map
(c) Share the maps with each other and comment on the data quality being depicted
Level 2: Decision analysis support
(e) Load the MCD models with scores
Level 3: Group process consistency (not needed)

Selection of projects
Phase 6: Consider unusual circumstances in the evaluation of the projects.
Level 1: Basic information handling support
(a) Retrieve and review the circumstances of projects
(b) Highlight projects with special circumstances on a map
(c) Identify the special circumstances with each other, and vote on the signi®cance
Level 2: Decision analysis support
(f) Sort through a discussion of unusual circumstances
Level 3: Group process consistency
(g) If a project did not get identi®ed as having a special circumstance, would it survive?
Phase 7: Compare the project alternatives against each other to determine those most worthy of consideration
Level 1: Basic information handling support
(continued)
Transportation improvement site selection 359

Table 3Ðcontd

(a) Retrieve all of the data


(b) Depict all of the projects on a map; if some where removed which ones?
(c) Share the map as a white board to be annotated; vote on the top choices
Level 2: Decision analysis support
(e) Make use of the MCD modeling
Level 3: Group process consistency
(g) If some projects were down-graded in terms of criteria, what would happen to the map display?
Phase 8: Negotiate for the most important projects
Level 1: Basic information handling support
(a) Retrieve the projects at the top of the list
(b) Depict all of the projects on a map
(c) Express concerns about the list of projects
Level 2: Decision analysis support
(e) What would be a negotiation model? negotiate criteria? or funding?
(f) The group may need a way of dealing with the negotiation
Level 3: Group process consistency
(g) Change some of the constraint criteria to see what happens with the projects

3.3. Current and near-term implementations of decision aid techniques


Group support system (GSS) information technology is relatively new with initial commercial
products introduced in 1989. Coleman and Khanna (1995) provide a comprehensive overview of
the GSS technology developments, including a list of product vendors, mostly for business appli-
cations. Unfortunately, none of the developments include GIS. The implementations of software
for GIS are much newer, but are expected to grow rapidly since the geography of places is always
being discussed in multiple local government contexts. This section provides some indication of
the status of the software and examples of where developments are heading. Implementation
status for the techniques listed in Table 3 (based on the order in Table 2) are discussed in turn.
However, process (simulation) models will not be discussed because they were not identi®ed as a
need in Table 3; these models pertain more directly to transportation analysis than to decision-
making. The following discussion draws from implementations of GIS, DSS, MCD models, and
GSS software currently on the market. It is important to make note that implementations of
software depend on implementations of local area network and wide area network data commu-
nications. Discussion of the hardware requirements are not included in this paper, but would have
to be considered when software design moves into implementation.

3.3.1. Information management techniques. Data and information management is one of the
three core geographic information technologies because of the large amounts of data being processed.
Information management for groups can be supported by virtual integration strategies, which rely on
network retrieval of copies of data in a transparent fashion. In addition, distributed data management
systems, e.g. Oracle and Ingres, are two major competitors in the market that support attribute data
retrieval and storage across local and wide-area computer networks. Although all GIS have basic
data management capabilities, few GIS packages have distributed data management capabilities
built into the spatial data manager component of the system. An exception is the recently
announced extensions to Oracle for managing spatial data that are used by Environmental Sys-
tems Research Institute in their Spatial Data Engine product (ESRI, 1996). Distributed data
management is a basic requirement for group support, but decision-makers would seldom use this
technology directly, because summary information is provided only with customization.
Group-based GIS has a requirement not only for access to shared data, but applications as well.
This has not been introduced as yet into commercial software, and may appear ®rst in the micro-
computer market through new operating systems such as Microsoft Windows NT, or IBM OS/2
through object linking. This is heralded as a breakthrough in applications processing, as we move
from applications-centered data processing to document-centered data processing. The latter
assumes integrated applications support for data management, and can only help with integration
of visual representation and analysis techniques, the other two core technologies in GIS.
360 T. L. Nyerges et al.

3.3.2. Representation techniques. Computer mapping display technology has been a mainstay in
GIS as one of the three core technologies. Small platform (increased performance) GIS packages
are at the stage where GIS can be on every desktop. All the major vendors are claiming at least
25,000 copies being distributed. User interfaces, however, are still a stumbling block because the
representations have not been developed by task, i.e. market sector, to create representations
suitable for speci®c work activities. This is beginning to change, and will continue to get better, as
third party vendors develop packages customized for speci®c applications. In addition, multimedia
support is appearing in GIS packages as indicated by photo and sound manipulation capabilities
in small platform packages. Charts, diagrams and tables may be linked to those representations to
enhance information presentations. For example, students in a GIS workshop class at the Uni-
versity of Washington made use of the Spatial Group Choice software for a freight mobility
enhancement demonstration decision project (Montejano et al., 1996) based on the documentation
from the Infrastructure Subcommittee of the Duwamish Coalition (Duwamish Coalition, 1996b).
The student freight mobility group used text and picture hotlinks to provide additional description
on a multimedia basemap (Fig. 2). [A detailed description of the Spatial Group Choice capa-
bilities, operation and system architecture are provided by Jankowski et al. (1997)].
A technical challenge for graphical representation involves evolving the technology into dis-
tributed display capabilities, following on the heels of distributed data management, and leading
the requirement for collaborative document support. Spatial Group Choice implements shared
displays in a local area network using map image transfers, but such a solution is not as bene®cial
as in a direct `shared whiteboard' solution. Such a capability can be implemented through use of
posting of maps and diagrams to shared whiteboards as in Ventana Corporations Group Systems
for Windows (Ventana, 1994). This capability can support group notations on maps, in part pro-
viding the functions to support a `blackboard-type meeting' environment where notations are left
in a common space on a data server for others to comment on.
However, it is necessary to have shared displays where decision-makers can change the character
of the display, as for example when a sensitivity analysis is performed. During sensitivity analysis a
criterion weight can be changed relative to other weights, e.g. the importance of trac safety in
relation to environmental concerns, to determine what in¯uence such a change might have on the
rankings of alternatives. Such a capability for criterion depiction was implemented in the Spatial
Group Choice software.
Since the objective in site selection decision-making is to prioritize sites, one major need is the
ability to display prioritized sites in map form. The student freight mobility group created grad-
uated circle maps to depict site rankings according to the criteria priorities of several of the coali-
tion stakeholder groups. The site rankings of the Port of Seattle stakeholder group are depicted in
Fig. 3, whereby larger circles indicate higher ranks and smaller circles indicate lower ranks.

3.3.3. Group collaboration techniques. Group collaboration techniques are at the heart of group
support systems (GSS). These techniques support basic communication. Electronic mail is a crude
form of such techniques through which one can broadcast a message to all members of a group. In
the student freight mobility project, several coalition stakeholders were interviewed and the results
of the interviews were discussed by the students in both face-to-face and distributed settings, with
the distributed setting making use of e-mail. The result was a set of stakeholder priority objectives
(Table 4). Those priorities were then synthesized into a set of criteria for prioritizing infrastructure
improvement project sites to enhance freight mobility (Table 5). However, a major diculty with
using e-mail for such discussion activity is that there is very little support to manage the history of
deliberations. Functions for idea generation, collection, and compilation of ideas, using anon-
ymous or named input of ideas, pooling and display of textual ideas, and search facilities to
identify possible common ideas are important for organizing group thoughts in a systematic way.
Several GSS packages exist on the market that include these capabilities. Many of them are
reviewed in Bostrom et al. (1992). These capabilities make use of hardware technology that
includes using data and voice transmission, electronic voting, electronic white boards, computer
conferencing, and large-screen displays.
Implementation of such capabilities with GIS depends to a large degree on the operating system
environment, and particularly the user interface support tools. Integration can range from loose to
tight integration (Nyerges, 1992). One way to implement such capabilities in a GIS environment is
Transportation improvement site selection 361

Fig. 2. Priority action sites for improvement of freight mobility in the Duwamish Corridor.

to do it along side GIS, relying on a similar operating system and user interface such as Microsoft
Windows. This type of strategy was used in two GIS courses at the University of Washington to
implement a collaborative learning and critical thinking support capability using GroupSystems
for Windows (Ventana, 1995) and ArcView 2.1 (ESRI, 1995), but was thought to be too cumber-
some to be used by the student freight mobility group for their GIS project, which was actually in
the third course in the GIS sequence (Nyerges and Chrisman, 1994). Another approach is to rely
on the GSS environment to provide hooks to GIS functions; these functions being invoked from a
menu with the GSS user interface (Faber et al., 1996). However, such approaches limit the use-
fulness of both packages, as low-level data passing does not exist to provide a document-centered
approach to information use. For this reason, development has been performed to make use of the
362 T. L. Nyerges et al.

Fig. 3. Student's interpretation of Port of Seattle ranking of improvement project sites.

Table 4. Improvement project priorities sampled from stakeholder interviews

Stakeholder title Stakeholder organization Overview of priorities

Planner Port of Seattle Freight mobility in the northern Duwamish


Corridor; emphasis on rail
Community ocial Georgetown Crime Prevention and Reducing stress on residential communities
Community Center (Georgetown speci®cally)
Manager King County, Infrastructure to support new stadium
Government Relations (northern part of Corridor)
Business owner Manufacturing Freight mobility for business owners
Manager City of Seattle, Engineering Intermodal access in the northern corridor
Transportation planner King County, Transportation Projects in unincorporated King County
Transportation improvement site selection 363

Table 5. Criteria for prioritizing infrastructure improvement projects

Criterion Description/example Measurement rating

Vehicular congestion Current amount of congestion Time rating 1±5


1: Least congestion
5: Most congestion
Potential impediments Future impediments to freight movements, Time rating 1±5
e.g. 16th Avenue bridge mechanism frequently
breaks down and will need to be replaced soon
1: Least impediment
5: Most impediment
Level of service Current ¯ow of trac Level rating 1±5
1: Least volume
5: Greatest volume
Fundability The cost of the project, and likelihood of Potential rating 1±5
acquiring necessary funds
1: Least potential
5: Most potential
Residential stress Potential for reducing the volume of freight Reduction rating 1±5
reduction movement through residential areas, e.g. Spokane
Street corridor is high priority for Georgetown residents
because it would result in a decrease in freight movements
through the residential areas
1: Lowest reduction
5: Highest reduction
Delay reduction Potential for reduction in the delay of trac Reduction rating 1±5
1: Lowest reduction
5: Highest reduction
Capacity increase Potential for an increase in the volume of freight Capacity rating 1±5
1: Least increase
5: Highest increase

dynamic data exchange functions within Microsoft Windows for Workgroups to provide a direct
connection for pooling votes from decision-maker workstations. Such an implementation was used
to develop a loose coupling linkage between the ArcView 2.1 GIS and multicriteria decision
modeling software in Spatial Group Choice to transfer data between the two modules. Although a
tight coupling of data transfer functionality is desired, it takes a total re-engineering of a software
platform to get modules to work seamlessly under one interface.
Throughout criteria development, weighting, and site prioritization a group can share their
decision-making concerns with each other in an attempt to move toward a consensus position.
Sharing the results of the criteria selection, weighting, and aggregation steps can be performed in
the `Consensus' module of Spatial Group Choice. Individual selections or results for any stage can
be submitted to a facilitator using the voting process implemented in the Consensus module. The
consensus voting operation uses a modi®ed Borda technique to combine the ranked alternatives
(Black, 1958; Hwang and Lin, 1987), although a simple majority vote capability also exists in the
software. The Borda technique takes into consideration the ranking of the alternatives to establish
an overall set of criteria selections, weights, or alternative site rank scores. For this example, a
consensus rank table and its corresponding consensus rank map are depicted in Figs 4 and 5,
respectively. Higher priority sites are depicted with larger circles, and lower priority with smaller
circles. Furthermore, the variance among the combined ranks is depicted using a shading scheme.
A lighter shade indicates a higher degree of consensus, and a darker shade indicates lower degree
of consensus. Low variance is indicative of situations where most decision-makers ranked a single
site as high, medium or low in the same way. High variance is indicative of one decision maker
ranking a site higher in a ranked-list relative to other decision-makers who might have ranked the
same site lower, or in the middle of the list.

3.3.4. Choice model techniques. Choice models provide assistance in comparing numerous alter-
natives and criteria against each other to select the best alternatives. Multicriteria decision models
are among the most popular of these, having been researched for at least 25 years, and imple-
mented as products for at least the past ten years (Ozernoy, 1991). There are many kinds of mul-
ticriteria decision models, each having advantages and disadvantages. Ozernoy (1991) presents a
detailed discussion of using MCD models to select the best MCD methods.
364 T. L. Nyerges et al.

Fig. 4. Consensus scores and variances of site rankings among the ®ve stakeholder groups: Business Community, City of
Seattle Planning, Neighborhood Council, King County Transportation, and Port of Seattle.

Integration of multicriteria decision models into GDSS has been an active area of research for
over ten years (Jelassi et al., 1985), but few have found their way into commercial GDSS. The
MeetingWorks software (Lewis, 1994) is one that does have a MCD model. However, because
there are so many types of decision analyses techniques (hence models), an exhaustive study of
which ones to include has probably prohibited most companies from investing in this development
path. Team Expert Choice (Expert Choice Inc., 1995) implements the analytical hierarchy process
approach to MCD models in a group context, but does not provide a distributed group commu-
nication capability.
Integration of multicriteria decision models with GIS has been an active area of research in the
past few years (Janssen and Rietveld, 1990; Carver, 1991; Jankowski, 1995), and is now becoming
a development activity. The IDRISI package now has several MCD models integrated in the
packageÐhowever the human-computer interface is still designed for individual use (Eastman et
al., 1993). Jankowski (1995) has undertaken a thorough study of the advantages and dis-
advantages of various aggregation methods, creating a framework to understand their di€erences.
From this basis of understanding, two aggregation methods were implemented as part of Spatial
Group Choice (Jankowski et al., 1997). Those two methods are weighted summation and rank
order. The weighted summation technique is based on a linear combination of criterion scores and
weights (Voogd, 1983). An evaluation score is calculated for each alternative site by multiplying
each criterion score by the corresponding criterion weight and adding the products. The sum of
the products calculated for each alternative represents the evaluation score for that alternative.
The rank order technique uses weighted summation with a linear combination of weights and
scores, but with rank ordering as the basis of computing the evaluation scores. Instead of using the
ratio scale properties of the criterion scores, each alternative's criterion scores are based on the
alternative's position in the ordered list of all of the scores for a criterion across all alternatives.
Since the scores are normalized for each criterion, an interval level of measurement exists. The
ranks (normalized scores) are then summed for each alternative, resulting in a rank ordered score
between 1 and 100. Students in the freight mobility project chose to use the weighted summation
technique to compute site ranking scores (Fig. 6).

3.3.5. Structured group process techniques. Agendas help structure group interaction for meet-
ings; but agenda topics do not help structure conversation about those topics. The structuring is
called `group process'. GSS have been developed to help reduce what is called group process loss,
i.e. loss of productivity due to `wandering' social interaction. For several years techniques of var-
ious kinds have been available to help facilitate group interaction. Some of the better known ones
Transportation improvement site selection 365

Fig. 5. Consensus site rankings using graduated circle symbols, with agreement variance depicted as shading in circles.

that can be used for structuring computer-assisted meetings are: automated Delphi, nominal group
techniques, and electronic brainstorming (Hwang and Lin, 1987). These techniques, together with
the techniques for collaborative communication support, can be used to organize meetings for
structured creativity. Such techniques are often useful at the beginning of group processes where
information creativity and planning tasks are more common, than at the end of group processes.
Many of the GSS described by Bostrom et al. (1992) implement such capabilities, with a meeting
facilitator used to encourage group process direction. No GIS to date make explicit use of struc-
tured group process techniques, but such techniques are available through a loose coupling to a GIS.
366 T. L. Nyerges et al.

Fig. 6. Criteria weighting and weighted summation results from the business community stakeholder group perspective.

3.3.6. Judgment re®nement/ampli®cation techniques. Judgment re®nement and ampli®cation


techniques are specialized techniques for detailing the character of choices made in relation to the
overall pattern of choices. One of the most signi®cant developments has been the addition of sen-
sitivity analysis to MCD packages. Expert Choice (Expert Choice Inc., 1994) is one such package
that has had this, and now implemented for group support as Team Expert Choice (Expert Choice
Inc., 1995). Spatial Group Choice links sensitivity analysis techniques with map displays to
visualize the spatial manifestations of alternative rankings of sites. Although software packages
that implement Bayesian decision techniques for judgment re®nement do exist, they have not
found there way into GSS or GIS packages.

3.3.7. Analysis and reasoning techniques. Analysis and reasoning techniques are still research
topics for the most part. Expert systems and mathematical programming packages are examples of
such techniques that have been integrated with GSS on a limited basis (Jarke, 1986), but not with
commercial packages. Expert systems and mathematical programming packages have been inte-
grated successfully with GIS for individual users, but group-based GIS integration is only begin-
ning to be proposed (Kanafani et al., 1994). More investigation must be done to determine how
useful they might be for decision environments, which tend to shy away from complex technology.

4. CONCLUSIONS AND PROSPECTS FOR THE FUTURE

In the U. S., as a result of ISTEA legislation, TIP decision processes are being `revealed' more
completely so citizens and organizations have a better idea as to what decisions are being made
and who is making them. The reason for establishing criteria is to make the process more under-
standable and justi®able. Capital resources are as scarce as ever. Decision processes must treat
multiple perspectives, particularly in regards to urban/regional growth issues as mandated in
Federal ISTEA and State growth management laws. Being able to facilitate, and therefore, reduce
the complexity, of the decision process is the goal of advanced information technology. Of course,
biases exist in the use of information technology depending on how it is put to use. However, being
able to reduce the complexity of the process by reducing the cognitive workload of TIP decision-
makers, is the goal of integrating capabilities from GSS, MCD models and GIS technologies.
Reducing the workload will hopefully lead to a more thorough treatment of information, exposing
initial assumptions more clearly, and subsequently, resulting in more participatory decisions
(Obermeyer and Pinto, 1994).
The design considerations as part of system requirements for group-based GIS-T documented in
this paper are based on a task model of the TIP improvement site selection process. This normative
Transportation improvement site selection 367

decision process is assumed to be a process of social±political interaction. In this way, technology


should be facilitating the social and political dynamics of the process, before it is marketed as a
solution to change the character of the process. The social science perspective therefore informs
the computer engineering design perspective.
To detail the framework, decision process issues were extracted from materials published by the
PSRC, King County Transportation Planning, and the Duwamish Coalition. A series of phases
are documented that describe a complex decision process involving many organizations. The
information activities in each of the phases has been documented, and is referred to as a general
statement of `user information need'.
To identify decision aid requirements, the decision phases and corresponding information needs
were examined in turn. Each phase was described in terms of the potential decision aid capabilities
that address the information needs of the phase. What has been created is a general statement of
requirements informed by related experience and research literature. The details of decision aid
requirements can only be established in particular decision contexts. Even if particular require-
ments were generated for one organization, that does not mean they necessarily would work for
all. The reason for developing a task model to begin with, is to explicitly incorporate social and
organizational design constraints (opportunities) that inform the technological design considera-
tions. The constraints vary from one situation to the next and make the technology what it is.
To avoid the `here it all is' syndrome with technology, it is recommended that an incremental
approach to system implementation be pursued. Several of the capabilities are more fundamental
than others. Level 1 functionality (basic information handling support) can be developed as an
extension of current GIS support. Thus, information retrieval making use of virtual integration as
a retrieval strategy can be easier than implementing distributed data management. Using repre-
sentational aids, such as maps on-line, can clarify simple questions like, `Where is that project?'.
Group collaboration techniques can provide the essentials for quick response opinions and rank-
ings from decision-makers. Such responses can be tabulated and displayed easily to the group for
any project and for all projects.
The ®ndings from the demonstration project indicate that there are considerable potential ben-
e®ts associated with the use of a group-based GIS-T in freight mobility decision-making. It is
however important to note that this type of software is only as good as the data upon which it is
based. Using a group-based GIS-T does not resolve con¯ict in itself, rather it helps discover some
possible areas in which some type of consensus might be achieved by helping to `visualize common
ground' among priorities. Political decisions that favor some groups over others are still likely to
be part of collaborative transportation decision-making, but the reasons for the biases can be
made more obvious using group-based GIS-T software. Undoubtedly the backgrounds and agen-
das of participants involved in the decision studies will in¯uence the result. If there are more
representatives from one stakeholder group than another, or if stakeholder groups are not at `the
table', then the consensus tally will be skewed.
The ®ndings in this research are only preliminary with regard to behavioral and social implica-
tions for transportation decision-making. Generalizations about the types of displays and the
types of decision models generally useful in decision-making must await formal studies like those
being conducted for habitat development (Nyerges and Jankowski, 1994). A theoretical frame-
work for undertaking such studies has been developed (Nyerges and Jankowski, 1997). Labora-
tory experiments, ®eld studies, and ®eld experiments could all be useful in developing
generalizations about relationships between decision-making variables, impacts of information
technology in realistic situations, and a bit of both, respectively (Nyerges, 1998). Better under-
standing of the behavioral and social issues in the context of information technology use will foster
better designs for the technology. A better understanding of design implications will encourage the
GIS industry to create collaborative information technology support for transportation improve-
ment project decision-making that works for people, rather than technology that requires people
to drastically change the way they work.

AcknowledgementsÐThis work was partially funded by grants from the National Science Foundation Divisions of
Undergraduate Education and Social and Behavioral Science under grant numbers DUE-9450934 and SBR-9411021,
respectively. Environmental Systems Research Institute provided the ArcView 2.1 software and Ventana Corporation
provided a copy of Group Systems for Windows for early tests of a GSS. The authors thank Piotr Jankowski of the
368 T. L. Nyerges et al.

University of Idaho for comments on an earlier draft of this paper. Special thanks to the other members of the Freight
Mobility Student group (Glenn Hwaung and Eric Rodriguez) in Geography 463 for allowing us to incorporate part of
their work in this article. Thank you to the anonymous reviewers for critical comment on the manuscript.

REFERENCES
Armstrong, M. (1993) Perspectives on the development of group decision support systems for locational problem-solving.
Geographical Systems 1, 69±81.
Armstrong, M. and Densham, P. J. (1995) A conceptual framework for improving human±computer interaction in loca-
tional decision-making, in Cognitive Aspects of Human±Computer Interaction for Geographic Information Systems,
Proceedings of the NATO ARW, T. Nyerges et al. eds. March, 1994 Mallorca, Spain, pp. 343±345. Kluwer Academic
Publishers, Dordrecht, Netherlands.
Benbasat, I. and Nault, B. R. (1990) An evaluation of empirical research in managerial support systems. Decision Support
Systems 6, 203±226.
Bhargava, H. K. Krishnan, R. and Whinston, A. B. (1994) On integrating collaboration and decision analysis techniques.
Journal of Organizational Computing 4(3), 297±316.
Black, D. (1958) The Theory of Committee and Elections. Cambridge University Press, Cambridge.
Bostrom, R. P., Watson, R. T. and Kinney, S. T. (1992) Computer Augmented Teamwork. Van Nostrand Reinhold, New York.
Brill, E. D., Flach, J., Hopkins, L. and Ranjithan, S. (1990) MGA: a decision support system for complex, incompletely
de®ned problems. IEEE Transactions on Systems, Man and Cybernetics 20, 745±757.
Carver, S. J. (1991) Integrating multi-criteria evaluation with geographical information systems. International Journal of
Geographical Information Systems 5(3), 31±339.
Coleman, D. and Khanna, R. (ed) (1995) Groupware: Technology and Applications. Prentice-Hall, Upper Saddle River, NJ.
Crossland, M. D., Wynne, B. E. and Perkins, W. C. (1995) Spatial decision support systems: an overview of the technology
and a test of eciency. Decision Support Systems 14, 219±235.
Davies, C. and Medyckyj-Scott, D. (1995) Feet on the ground: studying user-GIS interaction in the workplace. In Cognitive
aspects of human±computer interaction for geographic information systems, Proceedings of the NATO ARW, T. Nyer-
ges et al. eds, 21±25 March 1994 Mallorca, Spain, pp. 123±141. Kluwer Academic Publishers, Dordrecht, Netherlands.
Davies, C. and Medyckyj-Scott, D. (1996) GIS users observed. International Journal of Geographic Information Systems
10(4), 363±384.
Densham, P. J. (1991) Spatial decision support systems. In Geographical Information Systems: Principles and Applications,
eds, D. J. Maguire, M. F. Goodchild, D. W. Rhind, pp. 403±412, John Willey & Sons, New York.
Densham, P. J., Armstrong, M. P. and Kemp, K. (1995) Report from the Specialist Meeting on Collaborative Spatial
Decision Making, Initiative 17, National Center for Geographic Information and Analysis, UC Santa Barbara, 17±21
September 1995.
DeSanctis, G. and Gallupe, R. B. (1987) A foundation for the study of group decision support systems. Management Sci-
ence 33, 589±609.
Dueker, K. and Vrana, R. (1992) Dynamic segmentation revisited: a milepoint linear data model. Journal of URISA 4(2),
94±105.
Duwamish Coalition (1995) The Duwamish Industrial Area: An Economic Pro®le. King County Oce of Management and
Budget, King County, WA.
Duwamish Coalition (1996a) Infrastructure Assessment of the Duwamish Corridor. King County Oce of Management and
Budget, King County, WA.
Duwamish Coalition (1996b) Freight Mobility Report. King County Oce of Management and Budget, King County, WA.
Eastman, J. R., Kyem, P. A. K., Toledano, J.and Jin, W. (1993) GIS and Decision-making Explorations in Geographic
Information Systems Technology. 4, (Geneva: UNITAR).
ESRI (1995) ArcView 2. Environmental Systems Research Institute, Inc. Redlands, CA.
ESRI (1996) Spatial data engine. Environmental Systems Research Institute, Inc. Redlands, CA.
Expert Choice Inc. (1994) Expert Choice Reference Manual. Expert Choice, Inc., Pittsburgh, PA.
Expert Choice Inc. (1995) Team Expert Choice Reference Manual. Expert Choice, Inc., Pittsburgh, PA.
Faber, B. G., Wallace, W. W. and Miller, R. M. P. (1996) Collaborative modeling for environmental decision making.
Proceedings, GIS'96 Symposium, March, Vancouver, British Columbia.
Gray, B. (1989) Collaborating: Finding Common Ground for Multiparty Problems. Jossey-Bass Publishers, San Francisco, CA.
Hwang, C. and Lin, M. (1987) Group Decision Making Under Multiple Criteria, Lecture Notes in Economics and Mathe-
matical Systems, Vol.281. Springer-Verlag, New York.
`Intermodal Surface Transportation Eciency Act of 1991' (PL102-240, 18 December 1991), 105 United States Statutes at
Large, 1914-2207.
Jankowski, P. (1994) author to supply details.
Jankowski, P. (1995) Integrating geographical information systems and multicriteria decision-making methods. Interna-
tional Journal of Geographical Information Systems 9(3), 251±273.
Jankowski, P., Nyerges, T., Smith, A., Moore, T. J. and Horvath, E (1997) Spatial group choice: a SDSS tool for colla-
borative spatial decision making. International Journal of Geographical Information Systems 11(6), 577±602.
Janssen, R. and Rietveld, P. (1990) Multicriteria analysis and geographical information systems: an application to agri-
cultural land use in the Netherlands. In Geographical Information Systems for Urban and Regional Planning, eds H. J.
Scholten and J. C. H. Stillwell, pp. 129-139. Kluwer, Dordrecht.
Jarke, M. (1986) Knowledge sharing and negotiation support in multiperson decision support systems. Decision Support
Systems 2, 93±102.
Jelassi, M. T., Jarke, M. and Stohr, E. A. (1985) Designing a generalized multiple criteria decision support systems. Journal
of Management Information Systems 1(4), 24±43..
Kanafani, A., Khattak, A. and Dahlgren, J. (1994) A Planning methodology for intelligent urban transportation systems.
Transportation Research - C 2(4), 197±215.
King County (1993) King County Transportation Plan Priority, Process Chapter, June 1993.
King County (1996) Adopted Capital Improvement Program Resource Book 96. King County Oce of Financial Man-
agement, King County Courthouse, Seattle, WA, 98104.
Transportation improvement site selection 369

Knapp, L. (1995) A task analysis approach to the visualization of geographic data, in Cognitive Aspects of Human±Computer
Interaction for Geographic Information Systems, Proceedings of the NATO ARW, T. Nyerges et al. eds. March 1994,
Mallorca, Spain, pp. 355±372. Kluwer Academic Publishers, Dordrecht, Netherlands.
Lewis, F. (1994) A Brief Introduction to Group Support Systems. Meeting Works Associates, Bellingham, WA. (also avail-
able from Enterprise Solutions, Seattle, WA).
McCormack, E. and Nyerges, T. (1997) What Does Transportation Modeling Need from a Geographic Information
System?, Journal of Transportation Planning 21, 5±23.
McGrath, J. E. (1984) Groups: Interaction and Performance. Prentice-Hall Inc, Englewood cli€s.
Montejano, R, Oshiro, C, Dadswell, M., Rodriguez, E. and Hwaung, G. (1996) Exploring the eciencies of a group deci-
sion-making GIS in the deliberation and selection of freight mobility project sites in the Duwamish Corridor. Unpub-
lished ®nal report, Freight Mobility Group, Geography 463, Department of Geography, University of Washington,
Seattle, WA.
Nyerges, T. L. (1992) Coupling GIS and spatial analytical models. Proceedings, Spatial Data Handling Conference, Char-
leston, SC, IGU Commission of Geographical Information Systems, pp. 534±523.
Nyerges, T. L. (1993) How do people use geographical information systems?. In Human Factors for Geographical Informa-
tion Systems, eds. D Medyckyj-Scott and H. Hearnshaw pp. 37±50. John Wiley & Sons, New York.
Nyerges, T. L. (1995a) GIS support for urban and regional transportation analysis. In The Geography of Urban Transpor-
tation, 2nd edn, ed. Susan Hanson. pp. 240±265 Guilford Press.
Nyerges, T. L. (1995b) Design considerations for group-based GIS: transportation improvement project decision-making as
an example. Proceedings, GIS-T '95, Reno, Nevada.
Nyerges, T. L. (1995c) Cognitive task performance using a spatial decision support system for groups. In Cognitive aspects
of HCI for GIS. Proceedings of the NATO ARW, eds. T. Nyerges, D. M. Mark, R. Laurini, and M. Egenhofer, pp. 21±
25 March 1994 Mallorca, Spain, March 21-25, Kluwer, Dordrecht, Netherlands.
Nyerges, T. L. (1998) Progress in spatial decision-making using GIS. Proceedings, 1996 International Young Scholars Sum-
mer Institute in Geographic Information, National Science Foundation, European Science Foundation, Berlin, Germany,
24 July±1 Aug, 1996 (in press).
Nyerges, T. L. and Chrisman, N. R. (1994) Facilitating collaborative learning and critical thinking in a geographic infor-
mation system laboratory for undergraduate education, funds awarded by National Science Foundation, Division of
Undergraduate Education, Instrumentation and Laboratory Improvement Program, DUE-9450934.
Nyerges, T. L. and Dueker, K. (1988) Geographic Information Systems in Transportation, invited summary paper included
in Proceedings, Eastern and Western Computer-Assisted Cartography Conferences in Transportation. Federal Highway
Administration, Planning Division, HPN-22, Washington, D. C..
Nyerges, T. L. and Jankowski, P. (1994) Collaborative spatial decision-making using geographic information system dis-
plays and multicriteria decision-making techniques, funds awarded by National Science Foundation, Division of Social
and Behavioral Science, Geography and Regional Science Program, and Decision, Risk and Management Science Pro-
gram SBR-9411021.
Nyerges, T. L. and Jankowski, P. (1997) Enhanced adaptive structuration theory: a theory of GIS-supported collaborative
decision-making. Geographical Systems 4(3), 225±257.
Obermeyer, N. and Pinto, J. (1994) Managing Geographic Information Systems. Guilford Press, New York.
Ozernoy, V. M. (1991) Choosing the best multiple criteria decision-making method. INFO 30(2), 159±171.
Poole, M. S. (1985) Tasks and interaction sequences: a theory of coherence in group decision-making interaction.In
Sequence and Pattern in Communicative Behaviour, eds. R. L. Street, Jr. and J. N. Cappella, pp. 206±224, Edward
Arnold, London.
Puget Sound Regional Council (1995a) Metropolitan Transportation Plan, Puget Sound Regional Council, 1011 Western
Avenue, Suite 500, Seattle, WA 98104-1035, May, 1995.
Puget Sound Regional Council (1995b) Call for Regional ISTEA Projects, Puget Sound Regional Council, 1011 Western
Avenue, Suite 500, Seattle, WA 98104-1035, 7 February, 1995.
Puget Sound Regional Council (1996) Regional Transportation Improvement Program, 1011 Western Avenue, Suite 500,
Seattle, WA 98104-1035.
Reitsma, R. (1996) Structure and support of water-resources management and decision-making. Journal of Hydrology 177,
253±268.
Reitsma, R., Zigurs, I., Lewis, C., Wilson, V. and Sloane, A. (1996) Experiment with simulation models in water-resources
negotiation. Journal of Water Resources Planning and Management 122(1), 64±70.
Shi€er, M. (1995) Geographic interaction in the city planning context: beyond multimedia prototype. In Cognitive aspects of
human±computer interaction for geographic information systems. T. Nyerges, D. M. Mark, R. Laurini, and M. Egenhofer.
Proceedings of the NATO ARW, pp. 21±25 March 1994, Mallorca, Spain, 295±310. Kluwer, Dordrecht, Netherlands.
Simon, H. Administrative Behavior, 3rd ed. MacMillan, New York.
Simon, H. and Associates (1992) Decision making and problem solving. In Decision-making: Alternatives to Rational Choice
Models, eds. M. Zey, pp. 32±53 Sage, Newbury Park, CA.
Sprague, R. and Watson, H. J. (eds) (1986) Decision Support Systems, Putting Theory into Practice. Prentice-Hall, Engle-
wood Cli€s, NJ.
Todd, P. and Benbasat, I. (1994) The in¯uence of decision aids on choice strategies: an experimental analysis of the role of
cognitive e€ort. Organizational Behavior and Human Decision Processes 60, 36±74.
Todd, P. and Benbasat, I. (1994) The in¯uence of decision aids on choice strategies under conditions of high cognitive load.
IEEE Transactions on Systems, Man and Cybernetics 24(4), 537±547.
Vlahos, N. J., Khattak, A., Manheim, M. L. and Kanafani, A. (1994) The role of teamwork in a planning methodology for
intelligent urban transportation systems. Transportation ResearchÐC 2(4), 217±229.
Ventana (1994) GroupSystems for Windows. Ventana Corporation, Tucson, AZ.
Vogel, D. R. (1993) Electronic meeting support. The Environmental Professional 15(2), 198±206.
Vonderohe, A. P., Travis, L., Smith R. L. and Tsai, V. (1993) Adaptation of Geographic Information Systems for Trans-
portation, Report 359. National Academy Press, Washington D.C.
Voogd, H. (1983) Multicriteria Evaluation for Urban and Regional Planning, London: Pion.
Zachary, W. (1988) Decision support systems: designing to extend the cognitive limits. In Handbook of Human±Computer
Interaction, eds. M. Helander, pp. 997±1030. Elsevier Science Publishers, Amsterdam.

You might also like