Enter your email address:

    Delivered by FeedBurner

Decision Support Systems


I. Introduction

Information technology-enabled decision-making and decision support is also visible in the non-commercial arena. CommunityViz (www.communityviz.com) enables users to create community development models using scenarios, forecasting and 3-D visualization. Stopglobalwarming.org (www.stopglobalwarming.org) is representative of sites that allow users to learn of strategies to reduce energy usage and collaborate with others to promote conservation. Emily’s List (www.emilyslist.org), which allows users to support female political candidates, is one of many sites promoting political engagement. Federal, state and local governments use the Internet to make large volumes of data publicly available online, and facilitate routine activities such as applying for permits, or seeking out government officials (see O’Looney 2003 for a survey of decision support-oriented e-government applications).



However, there are a number of limits to IT-enabled decision support. IT-enabled services for search, service provision, recommendations and advice are less common for vulnerable or technologically-inexperienced populations: low-income persons, the elderly, the homeless, persons in the criminal justice system, those with mental or physical disabilities, as well as persons who work with or advocate for vulnerable populations. There are few applications that enable individuals to leverage the expertise and diversity of on-line social networks to identify, formulate and solve problems. In addition, there are few applications that enable public organizations to systematically organize best practices, collect data from diverse sources, collaborate with stakeholder groups and manage a variety of analytic models in order to pursue system redesign.

The purpose of this paper is to develop a framework for IT-assisted decision support for public choice and public action. In addition, this paper provides three case studies of research initiatives that represent contributions to this domain: counseling support for low-income housing mobility, policy support for senior services provision, and leveraging social networks to reduce energy consumption.

There are three ways that IT-assisted decision support applications might improve the quality and/or extent of public policy participation and personal action. First, these applications may improve the quality of high-stakes decisions affecting tangible, long-term outcomes for individuals and families. Examples of these decisions include choice of housing and neighborhood, career options and job training/retraining, behavioral change, drugs, services or providers to improve health quality, and family support services to assist households in crisis. In all of these cases, an established social service infrastructure exists to support individual choices. However, high-quality, easily accessible and personalized IT-enabled assistance provided in such a way as to improve the efficiency and effectiveness of human providers are not generally widely available (see e.g. the widely-criticized “Medicare Prescription Drug Plan Finder”, http://www.medicare.gov/MPDPF/Home.asp but also Vitolo and Vance 2002 for a successful example of a DSS to assist dislocated defense workers find new careers). Second, these applications may improve the quality of high-stakes decisions affecting outcomes for entire stakeholder groups, communities and regions. Examples of these decisions include provision of senior services, urban and regional planning, public safety service provision, and health care (see e.g. Timmermans 1997 for a collection of essays on spatial DSS for urban planning). Third, these applications can improve the quality of routine or repeated decisions that can improve the quality of everyday life. Examples of these include choice of transit route, changes in energy or food consumption and expanding or diversifying social networks for work, recreation and learning.

Developing decision support applications for public policy and personal action will require research to clarify a number of important issues. First, additional research will be required to understand the decision-making capabilities and preferences of typical application users, especially members of vulnerable or under-represented groups. Though the research literature on behavioral decision making is extensive (see e.g. Parker and Fischoff 2005, Scott and Bruce 1995), there is less understanding of the role that race, ethnicity, class and other personal characteristics may play in policy-related decision-making. Second, research will be required to understand how information should be presented to users to clarify decision alternatives. There is an extensive literature on the graphical presentation of quantitative information (see e.g. Tufte 2001, Bederson and Schneiderman 2003), but less understanding of presentation characteristics (type – maps, tables, charts; medium – paper, computer, video, verbal; style – prescriptive/suggestive, descriptive) and quantity of alternatives that best enable inexperienced users to frame and solve policy-related decision problems using information technology. There is also a need for research to understand how complex decision problems can be conveyed meaningfully on hardware other than personal computers, e.g. cellphones. Third, decisions may be performed using a wide variety of prescriptive methods (Eiselt and Sandblom 2004): compensatory versus non-compensatory methods; utility function-based compensatory methods versus non-utility function-based compensatory methods, and so on. However, there is no consensus regarding the decision methods that add the greatest value for inexperienced users solving policy-related decision problems.

The case studies in this paper provide limited and suggestive evidence to answer the questions listed above. Experience with housing mobility counseling support systems provides evidence that low-income families can make productive, IT-assisted decisions, but serving organizations face high barriers to empowering clients and re-aligning services to meet the needs of diverse clients. Policy systems for senior services provision have demonstrated that the significant potential for multiple models, data and expertise to support system redesign is counterbalanced by difficulties in sustaining the application in a resource-constrained environment. Preliminary research on use of social networks to reduce energy consumption indicates that changing intentions and the information base for making decisions is likely to be much easier than changing and sustaining actual energy consumption behaviors, but that online networks offer the potential for sustainable, long-term changes in daily behavior across large populations.

The remainder of the paper is organized as follows. Section II presents principles of DSS for public policy change and personal action. Section III contains a case study of housing counseling support systems. Section IV surveys research on policy support systems designed to improve senior services provision. Section V introduces a new research direction: information technology to induce and support changes in energy consumption. Section VI concludes and identifies next steps.

II. Principles of Decision Support Systems for Public Policy Change

In this section we present principles that are asserted to be important in designing decision support systems for public policy action and individual choice.

Values

While personal values as a basis for decision-making are important in all domains, the importance of values that clarify decision choices in the non-commercial domain is especially great, since there is less consensus regarding individuals’ motivations for actions that don’t involve commercial transactions. Keeney’s “value-focused thinking” model (Keeney 1996) represents a useful framework for understanding individual and organizational motivations for decision-making and action. (Of course, there are many other methods for clarifying goals, alternatives and attributes in policy-related problem-solving; see e.g. Bardach’s (2004) eight-fold path for policy analysis.) Keeney asserts that clarifying the relationship between shorter-term, more easily-quantifiable “means” objectives and longer-term, more qualitative “ends” objectives can help identify decision alternatives, or strategic directions, that are more likely to be consistent with decision-maker preferences. He also asserts that clarifying the relationship between basic, or fundamental ends objectives, and more specific ends objectives that clarify the dimensions along which decisions can be made can help identify measurable attributes by which decision alternatives might be ranked.

We assert that DSS for public policy and public decisions should be designed on the basis of a rigorous values analysis, distinguishing between individual values, organizational values and societal values. For example, should a housing counseling support system have as a fundamental aim helping users choose among affordable housing units known to be available to a housing authority (risking selection bias in neighborhood quality), or instead help users to better navigate the private and public housing markets themselves (risking confrontation with barriers such as housing discrimination, or difficulties in identifying feasible housing alternatives)? Should a policy system for senior services be motivated by the desire for incremental changes in the existing system (where should a new senior center be located?) or fundamental change (should the senior center system be based on a traditional bricks-and-mortar model for site-based services, or something different?) A system for encouraging reductions in individual energy consumption use as motivation individuals’ desire to save money, or enjoy positive regard from peers, or take action to improve the environment? While there is some preliminary evidence from surveys and field research that can address these questions, the body of evidence is currently thin.

Evidence

Modern policy analysis relies on evidence for and against specific proposals, derived from studies of existing policies and analytical models representing policies not yet implemented (see e.g. Bardach 2004). However, it is difficult to incorporate evidence into public DSS, as it is often contradictory, hard to summarize effectively, or simply not known. Examples of sources of evidence include social services, public policy and engineering. In housing counseling, evidence might come in the form of data on fair housing complaints in various jurisdictions, or summaries of studies indicating observed housing and life outcomes for families making choices similar to those considered by the current user. In policy systems for senior services design, evidence might be white papers or links to research that summarize strategies for services provision, or provide specific guidance in applying analytical methods. In applications to reduce energy consumption, evidence might be estimates of reductions in carbon emissions, or reductions in out-of-pocket costs, associated with specific energy-saving choices. DSS for public action should enable users to identify and assess the evidence supporting specific action alternatives, or for a variety of problem types, quickly and easily.

Models

As indicated previously, decisions are made using a wide variety of models, some explicit, others not. Herbert Simon (1955, 1990) powerfully advocated for decision models that accommodate conceptual limitations of ordinary users. Also, extensive research on decision-making styles (Scott and Bruce 1995) shows a wide range of attitudes regarding individuals’ capability to make coherent decisions on the basis of well-articulated values and evidence. Finally, there is a substantial literature in management science on prescriptive decision models (Eiselt and Sandblom 2004).

However, there is no consensus on the type, and sophistication, of decision models that are most-appropriate for public policy decisions and individual action, especially for vulnerable and under-represented populations. We assert that these DSS should facilitate model-based decision-making. This would seem obvious, except that from different perspectives, models per se are often not seen as central to the goal of the decision support process. For example, surveys of housing counselors that led to the development of the housing counseling support system (Johnson 2005c) found that they saw their fundamental role as administrative (ensuring that clients followed the rules of assisted housing programs) rather than participatory (ensuring that clients make the housing relocation decisions that are best for them). Models of decision-making were thus irrelevant to their conception of assisted housing services provision. Also, current research on applications to reduce energy consumption is confronting the problem of representing the energy reduction choice: should it be a single recommendation that is accepted or rejected by the user, or multiple recommendations, accompanied by a decision model that can help users choose a most-preferred alternative?

Therefore, public DSS research should address both descriptive models to increase the capacity and skills for human decision-making, and prescriptive models to increase the technical proficiency in identifying and quantifying decision alternatives and choosing most-preferred courses of action.

Creativity and Collaboration

In order to add value to decision-making processes, DSS must actually be used. However, technical design of DSS, and design or redesign of administrative processes to facility DSS usage is challenging. Recent research on a decision support system for evidence-based medicine practice in an outpatient setting (Zheng et al. 2005) found that, despite an application designed according to expressed preferences of users and administrators, actual usage was lower than expected, usage styles ran counter to the application design, and social networks that might support usage relied heavily on perceived support from administrators, whether or not they actually used the system, and self-assessed computer knowledge negatively influenced system usage, while self-assessed enthusiasm for IT positively influenced system usage. We assert that DSS for public decision-making must facilitate creativity and collaboration in order to maximize usage. First, systems should reflect characteristics of individual users. Personalization of energy reduction recommendations to account for user sociodemographic characteristics is a key component of the system for energy reduction choices currently under development. Second, systems should allow users to test alternative assumptions and hypotheses. For example, the housing counseling support system allows users to define multiple queries in order to identify a candidate set of housing units and/or neighborhoods, and allows users to rank alternatives using two different decision models. Third, systems should allow users to collaborate within and across stakeholder groups. For example, proposed policy systems for senior services design allow multiple users to add to the knowledge base, and the proposed system to encourage energy usage reductions is based on the ability of social networks to influence usage behavior. Last, systems should provide rapid feedback to modify choices: based on decisions generated by support system, a user should be able to easily revise values, or quantitative criteria by which decision alternatives are generated, or criteria used to rank alternatives. While many existing general DSS software have these features (Creative Decisions Foundation 2006, Visual Decision, Inc. 2004), they are less common in the types of public policy-oriented systems considered in this paper. While the housing counseling support system prototype does not yet allow users to easily revert between decision outcomes and decision inputs, the proposed “footprints” system is intended to allow users to consider different energy reduction recommendations based on a range of personal usage characteristics and values.


I. Introduction

Information technology-enabled decision-making and decision support is also visible in the non-commercial arena. CommunityViz (www.communityviz.com) enables users to create community development models using scenarios, forecasting and 3-D visualization. Stopglobalwarming.org (www.stopglobalwarming.org) is representative of sites that allow users to learn of strategies to reduce energy usage and collaborate with others to promote conservation. Emily’s List (www.emilyslist.org), which allows users to support female political candidates, is one of many sites promoting political engagement. Federal, state and local governments use the Internet to make large volumes of data publicly available online, and facilitate routine activities such as applying for permits, or seeking out government officials (see O’Looney 2003 for a survey of decision support-oriented e-government applications).



However, there are a number of limits to IT-enabled decision support. IT-enabled services for search, service provision, recommendations and advice are less common for vulnerable or technologically-inexperienced populations: low-income persons, the elderly, the homeless, persons in the criminal justice system, those with mental or physical disabilities, as well as persons who work with or advocate for vulnerable populations. There are few applications that enable individuals to leverage the expertise and diversity of on-line social networks to identify, formulate and solve problems. In addition, there are few applications that enable public organizations to systematically organize best practices, collect data from diverse sources, collaborate with stakeholder groups and manage a variety of analytic models in order to pursue system redesign.

The purpose of this paper is to develop a framework for IT-assisted decision support for public choice and public action. In addition, this paper provides three case studies of research initiatives that represent contributions to this domain: counseling support for low-income housing mobility, policy support for senior services provision, and leveraging social networks to reduce energy consumption.

There are three ways that IT-assisted decision support applications might improve the quality and/or extent of public policy participation and personal action. First, these applications may improve the quality of high-stakes decisions affecting tangible, long-term outcomes for individuals and families. Examples of these decisions include choice of housing and neighborhood, career options and job training/retraining, behavioral change, drugs, services or providers to improve health quality, and family support services to assist households in crisis. In all of these cases, an established social service infrastructure exists to support individual choices. However, high-quality, easily accessible and personalized IT-enabled assistance provided in such a way as to improve the efficiency and effectiveness of human providers are not generally widely available (see e.g. the widely-criticized “Medicare Prescription Drug Plan Finder”, http://www.medicare.gov/MPDPF/Home.asp but also Vitolo and Vance 2002 for a successful example of a DSS to assist dislocated defense workers find new careers). Second, these applications may improve the quality of high-stakes decisions affecting outcomes for entire stakeholder groups, communities and regions. Examples of these decisions include provision of senior services, urban and regional planning, public safety service provision, and health care (see e.g. Timmermans 1997 for a collection of essays on spatial DSS for urban planning). Third, these applications can improve the quality of routine or repeated decisions that can improve the quality of everyday life. Examples of these include choice of transit route, changes in energy or food consumption and expanding or diversifying social networks for work, recreation and learning.

Developing decision support applications for public policy and personal action will require research to clarify a number of important issues. First, additional research will be required to understand the decision-making capabilities and preferences of typical application users, especially members of vulnerable or under-represented groups. Though the research literature on behavioral decision making is extensive (see e.g. Parker and Fischoff 2005, Scott and Bruce 1995), there is less understanding of the role that race, ethnicity, class and other personal characteristics may play in policy-related decision-making. Second, research will be required to understand how information should be presented to users to clarify decision alternatives. There is an extensive literature on the graphical presentation of quantitative information (see e.g. Tufte 2001, Bederson and Schneiderman 2003), but less understanding of presentation characteristics (type – maps, tables, charts; medium – paper, computer, video, verbal; style – prescriptive/suggestive, descriptive) and quantity of alternatives that best enable inexperienced users to frame and solve policy-related decision problems using information technology. There is also a need for research to understand how complex decision problems can be conveyed meaningfully on hardware other than personal computers, e.g. cellphones. Third, decisions may be performed using a wide variety of prescriptive methods (Eiselt and Sandblom 2004): compensatory versus non-compensatory methods; utility function-based compensatory methods versus non-utility function-based compensatory methods, and so on. However, there is no consensus regarding the decision methods that add the greatest value for inexperienced users solving policy-related decision problems.

The case studies in this paper provide limited and suggestive evidence to answer the questions listed above. Experience with housing mobility counseling support systems provides evidence that low-income families can make productive, IT-assisted decisions, but serving organizations face high barriers to empowering clients and re-aligning services to meet the needs of diverse clients. Policy systems for senior services provision have demonstrated that the significant potential for multiple models, data and expertise to support system redesign is counterbalanced by difficulties in sustaining the application in a resource-constrained environment. Preliminary research on use of social networks to reduce energy consumption indicates that changing intentions and the information base for making decisions is likely to be much easier than changing and sustaining actual energy consumption behaviors, but that online networks offer the potential for sustainable, long-term changes in daily behavior across large populations.

The remainder of the paper is organized as follows. Section II presents principles of DSS for public policy change and personal action. Section III contains a case study of housing counseling support systems. Section IV surveys research on policy support systems designed to improve senior services provision. Section V introduces a new research direction: information technology to induce and support changes in energy consumption. Section VI concludes and identifies next steps.

II. Principles of Decision Support Systems for Public Policy Change

In this section we present principles that are asserted to be important in designing decision support systems for public policy action and individual choice.

Values

While personal values as a basis for decision-making are important in all domains, the importance of values that clarify decision choices in the non-commercial domain is especially great, since there is less consensus regarding individuals’ motivations for actions that don’t involve commercial transactions. Keeney’s “value-focused thinking” model (Keeney 1996) represents a useful framework for understanding individual and organizational motivations for decision-making and action. (Of course, there are many other methods for clarifying goals, alternatives and attributes in policy-related problem-solving; see e.g. Bardach’s (2004) eight-fold path for policy analysis.) Keeney asserts that clarifying the relationship between shorter-term, more easily-quantifiable “means” objectives and longer-term, more qualitative “ends” objectives can help identify decision alternatives, or strategic directions, that are more likely to be consistent with decision-maker preferences. He also asserts that clarifying the relationship between basic, or fundamental ends objectives, and more specific ends objectives that clarify the dimensions along which decisions can be made can help identify measurable attributes by which decision alternatives might be ranked.

We assert that DSS for public policy and public decisions should be designed on the basis of a rigorous values analysis, distinguishing between individual values, organizational values and societal values. For example, should a housing counseling support system have as a fundamental aim helping users choose among affordable housing units known to be available to a housing authority (risking selection bias in neighborhood quality), or instead help users to better navigate the private and public housing markets themselves (risking confrontation with barriers such as housing discrimination, or difficulties in identifying feasible housing alternatives)? Should a policy system for senior services be motivated by the desire for incremental changes in the existing system (where should a new senior center be located?) or fundamental change (should the senior center system be based on a traditional bricks-and-mortar model for site-based services, or something different?) A system for encouraging reductions in individual energy consumption use as motivation individuals’ desire to save money, or enjoy positive regard from peers, or take action to improve the environment? While there is some preliminary evidence from surveys and field research that can address these questions, the body of evidence is currently thin.

Evidence

Modern policy analysis relies on evidence for and against specific proposals, derived from studies of existing policies and analytical models representing policies not yet implemented (see e.g. Bardach 2004). However, it is difficult to incorporate evidence into public DSS, as it is often contradictory, hard to summarize effectively, or simply not known. Examples of sources of evidence include social services, public policy and engineering. In housing counseling, evidence might come in the form of data on fair housing complaints in various jurisdictions, or summaries of studies indicating observed housing and life outcomes for families making choices similar to those considered by the current user. In policy systems for senior services design, evidence might be white papers or links to research that summarize strategies for services provision, or provide specific guidance in applying analytical methods. In applications to reduce energy consumption, evidence might be estimates of reductions in carbon emissions, or reductions in out-of-pocket costs, associated with specific energy-saving choices. DSS for public action should enable users to identify and assess the evidence supporting specific action alternatives, or for a variety of problem types, quickly and easily.

Models

As indicated previously, decisions are made using a wide variety of models, some explicit, others not. Herbert Simon (1955, 1990) powerfully advocated for decision models that accommodate conceptual limitations of ordinary users. Also, extensive research on decision-making styles (Scott and Bruce 1995) shows a wide range of attitudes regarding individuals’ capability to make coherent decisions on the basis of well-articulated values and evidence. Finally, there is a substantial literature in management science on prescriptive decision models (Eiselt and Sandblom 2004).

However, there is no consensus on the type, and sophistication, of decision models that are most-appropriate for public policy decisions and individual action, especially for vulnerable and under-represented populations. We assert that these DSS should facilitate model-based decision-making. This would seem obvious, except that from different perspectives, models per se are often not seen as central to the goal of the decision support process. For example, surveys of housing counselors that led to the development of the housing counseling support system (Johnson 2005c) found that they saw their fundamental role as administrative (ensuring that clients followed the rules of assisted housing programs) rather than participatory (ensuring that clients make the housing relocation decisions that are best for them). Models of decision-making were thus irrelevant to their conception of assisted housing services provision. Also, current research on applications to reduce energy consumption is confronting the problem of representing the energy reduction choice: should it be a single recommendation that is accepted or rejected by the user, or multiple recommendations, accompanied by a decision model that can help users choose a most-preferred alternative?

Therefore, public DSS research should address both descriptive models to increase the capacity and skills for human decision-making, and prescriptive models to increase the technical proficiency in identifying and quantifying decision alternatives and choosing most-preferred courses of action.

Creativity and Collaboration

In order to add value to decision-making processes, DSS must actually be used. However, technical design of DSS, and design or redesign of administrative processes to facility DSS usage is challenging. Recent research on a decision support system for evidence-based medicine practice in an outpatient setting (Zheng et al. 2005) found that, despite an application designed according to expressed preferences of users and administrators, actual usage was lower than expected, usage styles ran counter to the application design, and social networks that might support usage relied heavily on perceived support from administrators, whether or not they actually used the system, and self-assessed computer knowledge negatively influenced system usage, while self-assessed enthusiasm for IT positively influenced system usage. We assert that DSS for public decision-making must facilitate creativity and collaboration in order to maximize usage. First, systems should reflect characteristics of individual users. Personalization of energy reduction recommendations to account for user sociodemographic characteristics is a key component of the system for energy reduction choices currently under development. Second, systems should allow users to test alternative assumptions and hypotheses. For example, the housing counseling support system allows users to define multiple queries in order to identify a candidate set of housing units and/or neighborhoods, and allows users to rank alternatives using two different decision models. Third, systems should allow users to collaborate within and across stakeholder groups. For example, proposed policy systems for senior services design allow multiple users to add to the knowledge base, and the proposed system to encourage energy usage reductions is based on the ability of social networks to influence usage behavior. Last, systems should provide rapid feedback to modify choices: based on decisions generated by support system, a user should be able to easily revise values, or quantitative criteria by which decision alternatives are generated, or criteria used to rank alternatives. While many existing general DSS software have these features (Creative Decisions Foundation 2006, Visual Decision, Inc. 2004), they are less common in the types of public policy-oriented systems considered in this paper. While the housing counseling support system prototype does not yet allow users to easily revert between decision outcomes and decision inputs, the proposed “footprints” system is intended to allow users to consider different energy reduction recommendations based on a range of personal usage characteristics and values.


Recent Articles

Blog Archive