« شما نمی توانید این را یک علم بکنید! » - تجزیه و تحلیل سیستم های پشتیبانی تصمیم گیری در زمینه های سیاسی

This paper reports on problems and conflicts encountered when using decision support systems (DSS) in political contexts. Based on a literature study and two case studies we describe problems encountered in relation not only to the DSS itself, but also to the political decision process. The case studies have been carried out in two cities in Sweden that at different times but in similar situations have used DSS in order to reach a decision in complicated and contested matters. In both cases we have previously found that the method and IT tool used for decision analysis were appreciated by most participants, but the inherent rationality of the DSS was in conflict with how participants usually make decisions as well as with the political process. The assumption was that a strict and open method would make grounds for clear decisions, but the results of the decision process were none of the cases implemented. In one case the result of the decision analysis was that no clear decision was made. In the other case the lowest ranked alternative was implemented. Furthermore, in neither city the method was ever used again. We therefore ask: What are the challenges and limitations to using DSS in political contexts? Our study shows that challenges relate to selecting and using criteria; eliciting weights for criteria (high level of subjectivity); understanding all the amount of facts available in the system; time constraints; and lack of impact on the final decision. This study contributes to both research and practice by increasing the understanding of what challenges are experienced in DSS use, since the findings can be used as a framework of challenges that should be addressed, in design of systems as well as method for use. The study also contributes to understanding the role of politicians in decision-making and the consequences for the use of DSS. Further, the literature study showed that there are overall very few studies on the actual use of DSS in a political context, and we therefore conclude by encouraging more studies reporting actual use.

In the public sector – whether referring to local or regional authorities, national governments, or other public bodies – decisions are made every day. On a general level the decision-making process can be described as based on three activities (Simon, 1977): Intelligence (identifying the need for a decision), Design (starts when a decision need is identified), and Choice (emerges when a decision is ready to be made). Public sector decision-making is, however, often multi-faceted and complex. Cooper's often-cited decision-making model (Cooper, 2006), aiming at supporting public sector decision makers to make ethical and rational decisions, includes the following steps and activities. First there is the descriptive task where the decision maker (or the administrator preparing a proposal) must make sure that the description is objective. This is usually done by indentifying key stakeholders and integrating their views. The second step concerns ethical issues and calls for an identification of conflicting values and principles underlying the decision to be made. The third step is about identifying possible alternative courses of action in order to force the decision maker to openly consider every possible alternative (not rejecting any alternative at this stage). Thereafter the model stipulates that the decision maker use his/her “moral imagination” by thinking through what the consequence of each alternative would be. Finally, the decision maker needs to “find a fit” where the alternative chosen should be balanced between various moral and ethical principles. The process of public sector decision-making of course varies depending on country, administrative level and what kind of decision is to be made, but the basic principles are the same — i.e., that the decisions taken should be based on fairness, objectivity, thoroughness, and in compliance with the law. As an example, when decisions are being made in the European Union the decision makers are supposed to first make an impact assessment (potential economic, social and environmental consequences) before proposing a new initiative. The commission is also supposed to consult stakeholders such as non-governmental organizations, local authorities and representatives of industry and civil society and seek legal advice. Decisions in European public sectors are usually made by political representatives, but even so these representatives are supposed to invite citizens, businesses and organizations to participate in consultations about the issue at stake (EU, 2011). These prescriptive models of how decisions should be made are often contrasted by analysis of how decisions are made in practice. The core insight into decision-making by Charles Lindblom (1959) was that decision-making in practice usually proceeds by successive limited comparisons rather than by complete assessment of values, alternatives, costs and benefits. Further, deciding through successive limited comparison involves simultaneous analysis of facts and values. Instead of specifying objectives and then assessing what policies would fulfil these objectives, the decision maker reaches decisions by comparing specific policies and the extent to which these policies will result in the attainment of objectives. While many are prepared to accept the validity of Lindblom's incrementalism as a descriptive theory, the incrementalist mode of decision-making has nonetheless been criticized for insufficient analysis of alternative policies and conservatism (i.e., Dror, 1968). It has also been criticized from a participatory point of view, since access is limited to political coalitions already in power and stakeholders that have already established themselves as influential actors (Bäcklund, 2010 and Sager, 1994). In Europe, for instance, citizen participation in policy processes has been low and it is accompanied by a disconcerting decline of voter participation in European elections (Rayner, 2003). With stronger calls for a more deliberative democracy – partly due to EU laws and directives (Horlitz, 2007) – attempts have been made to include more actors in the decision-making process. It is believed to be of future strategic importance to facilitate that those citizens and citizen groups who want to participate in democratic processes can do so (Dawes, 2009) and this means that different alternatives will have to be discussed openly. Including citizens or citizens' groups in a decision-making process is a cumbersome process not only due to the many actors involved but also due to the need for clarity in argumentation – e.g., clarifying the motives for a standpoint, analyzing the consequences of the suggestions, balancing demands against costs – in order to allow for all actors to make informed choices. For informed choices to be made there is thus a need to create easy-to-use and transparent choice-making models for complex matters where many actors are involved. This is where decision support systems (DSS) become an attractive and viable option. 1.1. Decision support systems in public sector Decision support systems (DSS) are computerized information systems designed to help decision makers and stakeholders define and discuss different problems and come up with various solutions and paths to take. DSS typically take use of different criteria, show the interrelations among multiple criteria, and also enable a comparison of the results (Horlitz, 2007). The use of computers as support for decision-making is nothing new — the public sector has a long history of using computers to make correct calculations and reasonable assessments. However, the advances made in technology, accompanied by an accumulated knowledge in the fields of decision theory, cognitive science and information science, have made it possible to use the computers in “more expansive ‘advisory’ roles to the decision making” (Saunders-Newton & Scott, 2001, p. 47). The hopes have thus long been set for computerized systems that can support and improve the decision-making in the public sector. But whereas the roots of DSS stem from the private sector we need to be aware that decision makers in the public sector and private sector “operate in different decision-making contexts and employ different decision processes” (Dillon et al., 2010 and Papadakis and Barwise, 1998). The political context in public sector affects the logic and mode of thinking (Murray, 1975) as well as the level of public involvement (Posas & Fischer, 2008). As for the logic and mode of thinking, public politicians and administrators have less flexibility and autonomy in identifying the problem that requires a decision — this may be politically governed or regulated by law. Public sector goals are also often vague and emphasize notions of equity (e.g., “a school where everyone belongs”), which affects the way strategic decisions are made (Nutt, 2006). Decision-making in the public sector is also more often than in the private sector supposed to be based on consensus and the broadest social good (Murray, 1975) and the expectations and accountability claims will therefore often differ (Nutt, 2006). Moreover, organizational survival in the public sector is more likely to depend on public perceptions that management is responsible and that procedures are “rational” than on objective efficiency, which may be difficult to measure (Langley, 1989). As for the level of public involvement, this is often supposed to be higher in public decision-making (EU, 2011 and Posas and Fischer, 2008). For the public sector it is unfortunate that most research on DSS use concerns the private sector because there are differences that restrict generalizations from one sector to another (Papadakis & Barwise, 1998). Ranerup (2008), scanning the field of DSS, found that most DSS are targeted toward the private sector and that in the few instances where reports of public sector use were made even fewer were aimed at citizens as users. Against this backdrop, there is clearly a need for more research on instances where DSS have been used in the public sector. Two such instances can be found in Sweden where two Swedish cities, Nacka and Örebro, used a decision support system (DSS) to help various parties involved in city planning processes come to a consensus in complex and contested issues. 1.1.1. Two cases of DSS use in Sweden The traditional model of decision-making in Swedish municipalities is similar to what was previously described as the European model, where the politicians in majority are in control of decision-making. Even though the Committees in Sweden are appointed proportionally, they are usually governed by the majority coalition. Decisions are most often based on one single proposal, prepared by civil servants in dialogue with the Chair of the Committee. The City Council or the Committee will either reject or accept the proposal, but acceptance is common as the majority coalition is involved already in the preparation. If the proposal is rejected it is sent back to the civil servants who prepare a new proposal for the politicians to consider. However, with stronger calls for a more deliberative democracy, and in cases where the matters at stake are known to be complicated and contested, attempts have been made to include more actors in the decision-making process. We have researched two such contested cases in the two Swedish cities, Nacka and Örebro. The contested matter in Nacka was about whether services such as roads, water supply, and sewers should remain private or be run by the municipality. The conflict in the municipality related to the islands Älgö and Gåsö which are situated in the attractive Stockholm archipelago. On these islands expensive houses owned by well-off newcomers were neighboring older houses inhabited by residents since generations and the conflict was mainly between these two groups. They had opposite opinions on the road, water, and sewers issues. Citizen groups from each camp had lobbied politicians for years, and there were political parties to support both fractions. In 2003, after years of failing to reach a decision, the City Planning Office decided to try to solve the problem by inviting all political parties and the civil society to the decision-making process. In order to manage all various arguments and options it was decided to use a decision support system (Grönlund, 2005). In Örebro the complex matter concerned which measures to take to clean the Svartå River running through the city. For many years the river had been deemed unhealthy to swim in, and drinking water for the municipality tapped upstream needs to be filtered. The debate had been going on for several election periods but had become more urgent due to the European Water Framework Directive which stipulates that the river should have a good ecological status by 2015 (including threats of heavy fines if this has not been achieved by 2021). Depending on which measures are taken to clean the river different stakeholders are affected and politicians have been vulnerable to much lobbying, in particular from farmers on whose lands changes must be implemented. In the winter 2008/2009 the planning board of Örebro decided to include all political parties and many expert civil servants in the decision-making. In order to facilitate the decision process it was decided to use a decision support system. 1.1.2. The process of decision-making in the two cases The same DSS was used in the two cases, “DecideIT”, which was based on a multicriteria decision analysis (MCDA) approach. The particular method used in our cases was a specific operationalization of the MCDA approach developed by Danielson, Ekenberg, Grönlund, and Åström (2008). The MCDA approach was supposed to provide means to make the decision-making process more structured and provide decision-makers with a better understanding of the trade-offs involved in a decision, e.g., between economic, social and environmental objectives. Briefly, the MCDA method is about making the decision makers denote preferences (importance) of different criteria before evaluating different alternatives (Bollinger & Pictet, 2003). This is in contrast to the traditional municipal decision-making previously described where the politicians in power express themselves directly on one alternative. The main idea of the DSS was, however, to support, inspire and validate a decision — not to actually make it. DecideIT was designed to aid the decision-making process and the guiding steps of the overall decision process were to provide: a) a description of the alternative options available, b) a description of the criteria (perspectives) under which to view the alternatives, c) a description of the consequences of each alternative with regard to each criterion and d) a procedure that can evaluate and compare the alternatives (Danielson, Ekenberg, Grönlund, & Larsson, 2005). In the Örebro case civil servants and all political parties, including those in opposition, took part in the decision analysis process. In the Nacka case this was also the case, but with the extension that they opened up for civil society to participate. The Nacka project was highly visible on the web and citizen representatives also took part in the ranking of criteria. The process in both cases was as follows: 1.Participants identified the criteria that should be discussed during the evaluation of alternatives. This was done collectively by the politicians during workshops where the developers of the DecideIT decision tool also participated. Two researchers researching the DecideIT tool and method led the discussions. 2.Participants elicited weights for the criteria, i.e., decided how important each criterion was. This was done individually by each participating person. 3.Based on the result of the elicitation different strategies and alternative action paths were brought forward and analyzed. These strategies were prepared by consultancy firms in cooperation with city civil servants. 4.The different alternatives were evaluated using the criteria set. Civil servants with different expertise usually took a leading role in this. 5. The system automatically calculated scores for each alternative using the criteria, the weights (relative importance of the different criteria), and the civil servants' assessment of the alternative actions. A “best solution” was found. In relation to the decision-making process the method stops after the evaluation of alternatives. The involvement and responsibility for the citizens, civil servants and the opposition thus ends here. In both cases the governing politicians are the final decision makers. Table 1 and Table 2 below summarize the generated criteria, weights and alternatives that the decision support process yielded in each case. As the tables show, the Nacka process yielded a greater distribution of average weights (from 3 to 32) whereas the Örebro criteria were quite tight ranging from 11 to 17.1.1.3. Outcome of the process The result of the decision analysis was in neither of the cases (fully) used or implemented in the political agenda. In Nacka, no clear decision was made. Both main alternatives – one in support of local community ownership and one in favor of municipality ownership – were handed over as suggestions to the City Council (Nacka City Council, 2004). In 2006, however, politicians decided in favor of municipality ownership. This decision was contested, but finally in 2010 the decision became a law and the public development of roads, water supply, and sewers are now initiated. The decision was made using traditional political methods. There is no evidence that the MCDA project has speeded up this process. In Örebro the best alternative was not implemented. Instead, the second lowest ranked criterion, bathability, was used as the argument for action taken in the summer of 2009. The leading politician had made a public promise to swim in the river by 2009 – and so he did in a small section of the river that had been sealed off from the river flow by a barrier – i.e., a pool in the river – where water was cleaned by means of a filter. This was in sharp contrast to the MCDA process where the top ranked criteria concerned ecological sustainability and water quality. Investigations made showed that the only feasible way of meeting these criteria was to build buffer zones between the river and the adjacent farmland. Civil servants have up to date prepared two suggestions for how to implement these buffer zones, but since the farmers – who have strong political influence – objected both suggestions have been rejected. This, again, is in contrast to the criteria set, where “Impact on existing business” was the lowest ranked one in the decision analysis. A year later a new decision was taken to start investigating the potential of getting all water to the municipality via a pipeline from one of the biggest lakes in Sweden. This option, although discussed in parallel with the decision analysis, was never a criterion in the final analysis and ranking. In summary, also in Örebro the political process went on looking for a more permanent solution, but it does not appear that the MCDA exercise had any influence on that process. It did not supply new options or criteria, and it did not speed up the process of, eventually, arriving at a solution that meets the requirements from the EU. 1.2. Research objectives How come the results of carefully crafted and agreed criteria and cumbersome ranking procedures were not used in the final decision-making? Reasons given by the politicians in the cases include, “this is not how decisions are made” (Örebro) and that “it is always about values at the end of the day — you can't make this a science” (Nacka). Another observation is that the system (and method) has not been used again in neither city. It seems that something happened on the way: “In the beginning we all agreed to give it [the method] an honest try. But as time passed and we applied it in real life there were compromises and deviations. Gradually one lost respect and was back in the old usual procedures” (Politician, Nacka). Against this backdrop this research asks: What are the challenges and limitations to using DSS in political contexts? By “using” we refer both to the very process of using the method as well as in making use of the result in the final decision-making. There are of course positive outcomes from the two cases. In a follow up on the Nacka case (Grönlund, 2005) we found that participants felt the method provided them with better information, as well as increased openness and clarity (e.g., the role of politicians had been made clearer). The DSS method was also tested with 90 students at the upper secondary school where it was proved usable for mass-scale participation. The students found the method to be easy to use and that it gave every user equal influence (Riabacke, Åström, & Grönlund, 2011). Notwithstanding these positive outcomes of the DSS use this study focuses on the challenges and limitations encountered. We believe that for future DSS design the shortcomings are most important to investigate. Whereas most studies on decision support systems report on their success (most often with very little evidence) we take a critical standpoint by focusing on the problems encountered. In a similar way as the Saunders‐Newton and Scott study (2001) we consider the role and limitation of the DSS as an adviser to the decision makers — with the addition that we use two empirical cases to get a deeper understanding of these challenges. The paper is organized as follows: after this introduction to the subject area and the two cases the method of this research is discussed. Thereafter the results of this study are presented: first challenges addressed by previous research, then challenges addressed in the two case studies. These results are then compared and we conclude by discussing our findings in relation to decision-making literature in general and difficulties of adaptation and change in particular.

A good deal of time has been spent advocating more formal, systematic and rational approaches to decision-making. However, in spite of this normative emphasis on the use of rational analysis, surprisingly little is known about how decision support systems (DSS) are used in practice. This study set out to investigate what the challenges and limitations to using DSS in political contexts are — so what have we learned? Comparing the results from the literature study to the results of our two Swedish case studies (Table 4) we basically find support for all the challenges experienced, relating to complexities of the method itself as well as to issues of power. It is only the challenge of ‘comprehensibility’ that is not given full support in our case studies. What is mentioned in the Örebro case is an assumed challenge regarding this issue if the citizens were to be involved. There is no evidence that this would have posed a challenge for the citizens. In the Nacka case it is likewise a representative who said that the issue was too technical and complex for the citizens — and we do not know if this was actually the case.Our findings are thus consistent with the (small) set of previous research where actual real cases are reported, and we have in this way strengthened the validity of these challenges. Also, our findings support more general descriptions of how decisions are made in complex organizations charged with creating and implementing public policy. The concerns related to the complexity of the method evidently echoes the long standing critique of separating values and alternative policies first put forth by Lindblom (1959) in his seminal work on incrementalism. He stated (1) that evaluation and empirical analysis usually are intertwined; that is, decision makers usually choose among values and among policies at one and the same time and (2) that decision makers do not find general formulations of objectives very helpful but rather focus their attention on marginal or incremental values. What our results add to this picture is that decision makers, even when they adopt a DSS premised on rationality, tend to distrust the process of first clarifying his/her values and then choose among policies. When it comes to the question of power, many scholars have questioned the value of adopting rational models of decision-making on the grounds that they fly in the face of political realities (i.e., Pettigrew, 1973 and Pfeffer, 1992). Political processes add constraints on possible solutions and choice is often a function of the distribution of power, as well as political tactics by participants. Since political decisions are marked by widespread political behavior, some scholars have described rationality and political behavior as opposite ends of one continuum. As political constraints accumulate, rational decision-making decreases (Janis, 1989). However, our cases indicate, in line with the argument of Dean and Sharfman (1993), that procedural rationality and political behavior can be seen as separate dimensions of strategic decision-making processes. In both Örebro and Nacka procedurally rational processes were introduced by the adoption of multicriteria decision analysis, but the choices made were nonetheless highly political. The cases illustrate that if disagreements are really based on political interests, factual information and logical arguments are unlikely to provide resolution. But the cases also show that uncertainty and ambiguity about the method can undermine the analysis. Thus one should perhaps not speak of decision processes as rational or political, as if they must be one or the other. There is always an interaction between these two dimensions that need to be more fully understood (Langley, 1989). When decision processes do not influence the choices made, as in the case of Nacka and Örebro, one may find it irrational to engage in difficult, cumbersome, and time-consuming efforts using a DSS. Why should so many people put their efforts and time into such a process when their gain is very small or even non-existent? The answer given from our informants is clear: “they shouldn't!” If these methods cannot inform or supplement representative democracy they are – in the best case – just a waste of time. In the worst case they are used as a political façade to make people think they have an influence. Rayner's (2003) suggestion that under current arrangements it is impossible for participatory techniques “to escape political–cultural constraints that reduce complex moral and esthetic issues to scientific framings” (Rayner, 2003, p. 163) appears valid also in our studies. On the other hand, if we move back to the basic principles for the public sector decision makers we may have to look at it from another perspective. If decisions are to be taken based on fairness, objectivity and thoroughness — then the DSS method has more strengths than the traditional one. It is not fair not to consider and assess all options and alternatives fairly and openly (which is often the case in traditional decision-making). Objectivity cannot be gained without including many stakeholders' views (as is made possible with participatory techniques such as the ones described in this paper) and thoroughness requires the process to take time. So if we want to work in line with these basic principles we should start looking at how these decisions support systems and processes could be designed in order to avoid the challenges addressed in this research. Taking research forward, beyond the current complementary/incompatible dichotomy, we would like to see more empirical work on the adaption necessary when decision-making processes change. There is something interesting in taken-for-granted statements such as: “This is not how decisions are made”, “politics work in larger lines”. Such statements can be explained by elements of incompatibility, but they can also be understood as expressions of the inevitable friction involved in transitions from one form of decision-making to another. From “a transitional perspective” (i.e., Klijn and Skelcher, 2007 and McKenna, 2011), the role of research is to contribute to a higher degree of contextual specificity and to a better understanding of the process of transforming traditional structures into hybrid forms. The new processes indicate a different role of elected politicians, a new way of defining the general interest as well as innovative strategies to develop power to get things done. However, we have seen that there are very few studies on the actual use of DSS in political contexts, and therefore there is very little knowledge about these issues. To guide research, as well as practice, we would encourage more studies on the actual use of DSS and less so on the design of new, theoretically very useful, systems. We have shown that DSS use in political decision-making is highly complicated as it conflicts with the politicians' traditional ways of working. However, as DSS holds promises of greater openness in decision-making they may still be worth testing in new decision-making processes in the context of “eGovernment 2.0”, where including the public in decision-making is imperative and efficient and effective ways or doing that are sought. After all, progress in practice is often mostly a matter of ‘muddling through.’