decision support systems
II) Sources of biased access:
3.6 From (in)efflciency to effective decision support
According to classic DSS theory, DSS should promote effective rather than just efficient decision-making (Sprague 1982). In practice, rhetorical justification of DSS frequently highlights efficiency as a key rationale for investment in and application of DSS. Claims of the efficiency of DSS conventionally stem from the capability of computers to speed up the processing and transformation of information. However, as Cleaves (1995:87) notes, “Computerized aids are most helpful in easing the costs and in speeding up the decision process, but don’t necessarily help improve decision quality unless their designs consider the needs and eccentricities of human judgement and decision processes”. In their exploration of the long-term effectiveness of DSS on decision outcomes, Barr and Sharda (1997:144) concluded that while the use of DSS tended to decrease the time decision-makers spent processing information, its effectiveness was more limited in terms of promoting an increased understanding of relationships between decision-making parameters. Barr and Sharda (1997:134,143) also observed that the efficiency gains often encouraged a reliance on DSS such that decision-makers would habitually defer the decision-making process to the computer, decreasing those decision-makers’ effectiveness in future decisions. Hence,
calculative efficiency does not necessarily promote long-term efficacy.
Furthermore, while few DSS developers admit it publicly, privately a number have acknowledged that a mismatch often arises between the DSS that the technical experts provide, and the DSS that intended users are willing to use. Fedra (1995:5) notes that although environmental DSS “have been discussed and advocated for a considerable time... Success stories of actual use in the public debate and
policymaking processes are somewhat more rare”. Clearly, if significant resources are expended on development of a DSS that is disregarded by intended users, then both efficiency and efficacy claims are seriously undermined.
A key question that arises is: what makes an environmental DSS an effective tool for users? The earlier discussion regarding the inadequacies of the technology transfer model suggest that the relevance of the DSS to the decision-making environment of the intended user is critical. This is supported by recent DSS literature. For
example, Moreno-Sanchez et al. (1997:164) suggest that “System development efforts that are technology-driven rather than end-user-demand driven are less likely to succeed”. Ewing et al. (1997:3) argue that “In the past, many DSS-style models have been unattractive because they are either ‘black box’ models that hide critical assumptions about the way the system functions or are so abstract as to ignore many of the political realities of decision making”. The latter comment highlights a
common critique that conventional DSS, premised upon a model of decision-making as an objective, technical exercise, is irrelevant to contemporary environmental decision-making which often entails apparently “eclectic, fuzzy and shifty compromise between competing interests” (Ezrahi 1994:32). As Norgaard (1994:144) notes, “Political choices must be made using criteria other than a
weighing of expected benefits and costs of mechanically predicted, patently unlikely, futures”.
Undermining relevance, and echoing a trend evident in a broad spectrum of computer application and innovation research, conventional DSS development has often
users’ needs rather than supporting the specific characteristics and needs of the decision-making or management environment (Argent and Grayson 1997:199). As participatory DSS development gains currency, the increased participation of users in defining the scope and focus of a DSS emerges as a strategy to confront the
inefficiency dilemma, by enabling management questions to take primacy over software system provision (Argent and Grayson 1997:204).
As discussed earlier, giving primacy to the needs and interests of the users and decision-making environment may lead to the conclusion that a computer-based DSS is redundant for the issue at hand. For example, it may emerge that the existing decision-making processes are adequate in light of the type of decisions, information requirements and the stakeholders involved, and would not be improved through introduction of a DSS. Alternately, the decision requirements may preclude efficient use of a computer-based DSS. For example, occasional or one-off decision tasks tend to be less amenable to efficient use of DSS than routine, frequent and
standardised tasks (cf Walker and Johnson 1996). The extent of knowledge about ecosystem processes and the availability and accessibility of data may also influence whether a computer-based DSS is judged a useful tool. For instance, Gough and Ward (1996:14) concluded that a DSS would not be an appropriate tool to assist management of Lake Ellesmere, in New Zealand, because “Until such time as comprehensive databases containing information about the biological and ecological processes in the lake and the impacts of changes can be established, there are too many sources of uncertainty associated with the types of decisions that are required to be made to make DSS either a useful or viable option”. In other situations, a highly user-friendly DSS incorporating simple relationships may be deemed sufficient for the decision at hand and suited to the literacy characteristics of users, and therefore development of a more complex and sophisticated DSS would be unwarranted and inefficient. Thus, as Argent and Grayson (1997:199) contend, “while software tools can be important components of a decision-making process, their role must be carefully considered in the light of the overall management objectives and the audience for the exercise”. In other words, effective decision support requires cognisance of the pertinent decision-making environment7. In this context, a conversational approach to DSS design, as advocated by Schon
(Interviewed in Binder 1996:56), which engages developers in dialogue with potential users about how they construe decision support, may be useful.
However, engaging professional developers and users in dialogue about the potential relevance of a DSS is not a simple task. On the one hand, if DSS developers are motivated by an interest in refining a particular system, then they may be less inclined to spend substantial amounts of time considering the possibility that an alternate system or process would be more effective. Commitment to a dialogue on relevance would imply a reframing of the role of the DSS developer as supporting a situated decision-making environment by assisting decision-makers in exploring appropriate tools or mechanisms to support their practice. On the other hand, developers frequently note that users “may not fully understand or be able to articulate requirements early in the development cycle” (Moreno-Sanchez et al.
7 Consonant with this point, the case study for this thesis will open by exploring the decision-making environment o f the highlands o f Northern Thailand.
1997:165). Participatory DSS development thus tends to be more akin to Lindblom’s (1959) “muddling-through” than Simon’s (1957) model of rationality. Opportunities for on-going dialogue between users and developers are necessary if both groups are to establish a mutual understanding of users’ needs and of the capability of the DSS to satisfy those needs.
It should be emphasised that the efficacy of a DSS is a dynamic quality, as the
relevance of a DSS may be compromised due to distortions or absences in knowledge which arise through inflexibility of a DSS to respond to changes in the decision making environment. As Torgerson and Paehlke (1990:9) note: "Knowledge is not something which can somewhere be insulated or enshrined, for - to be relevant - it depends always upon the context and dynamics of organizational activity” . Since the decision-making environment, including management objectives, decision options, sociopolitical networks, interested actors, and ecological processes, is not static, the decision support must co-evolve with the environment. As Moreno- Sanchez et al. (1997:164) observe, in the context of multi-media environmental GIS, systems “can easily become ‘snapshots’ of existing conditions and run the risk of becoming outdated shortly after completion”. Against this background, sustained negotiation processes between users and professional developers emerge as useful to enable on-going dialogue about the dynamic efficacy of a DSS, including emerging biases.
As well as relevance, the credibility of a proposed DSS is also an important efficacy consideration. As alluded to during the discussion about expertise, intended users’ cynicism about the purported neutrality and legitimacy of the knowledge or reality embodied within technology often fosters non-use. In reference to the introduction of systems models developed by the Bonneville Power Authority and the U.S. Army Corps of Engineers to inform management of the Columbia River Basin of systems models, McLain and Lee (1996:442) discuss how fish agencies and tribal authorities raised objections to use of the models because they perceived the models to be flawed in favour of hydropower interests. As above, the key is the extent to which users perceive that the functional relationships and other information incorporated within the DSS are pertinent and valid. In this context, the transparency and
comprehensibility of the DSS, and thus literacy or communication biases, emerge as significant. As Laacke (1995:126) argues: “The function and output of each step must be understandable, and it must be credible both to those who will exercise it and to those who are concerned about the information it produces”8. Thus, the
interrogation of potential biases during DSS development emerges as a dimension of satisfying equity and transparency dimensions of credibility, and is thereby important for effective - and efficient - delivery of decision support. While every DSS will be associated with some form of partiality and bias, the key question is whether the biases associated with a particular DSS are significant in light of stakeholders’ concerns.
8 Laacke (1995:126) further argues that “A computer-based DSS that becomes so integrated and self-