... of its strong participation. Journal of Information Technology (2000) 15, 39â51 .... to human factors such as 'needs' and 'job satisfaction' and 'technical factors'.
Journal of Information Technology (2000) 15, 39–51
Participatory sociotechnical design of organizations and information systems – an adaptation of ETHICS methodology PETER ADM AN Computer Centre, University of Hull, Cottingham Road, Hull HU6 7RX, UK
L ORR AINE WA RREN Lincoln School of Management, Lincoln University Campus, Brayford Pool, Lincoln LN6 7TS, UK
This paper examines a practical adaptation of the ETHICS methodology used in redesigning an information technology (IT) support service in an academic setting. The purpose of the project was to design appropriate organizational structures and functions and an accompanying information system (IS), to increase the effectiveness of the existing service. A participative sociotechnical approach was adopted for the entire design process which was carried out by the practitioners themselves. The staff’s views were elicited during informal participatory group sessions as well as in one-to-one informal discussions. While ETHICS was the overall guiding methodology for the design, QUICKethics was used as a complementary means of analysing the requirements of the new IS. This paper describes the methodology used and the design process; it re ects on the adaptation and its match with the ETHICS methodology, exploring the claimed association with the viable systems methodology and concludes with suggestions for further research.
Introduction The focus of this paper is the redesign of an existing service organization and the congruent IS. While the ETHICS methodology was used as the overall guiding methodology of the design process, QUICKethics was used as a complementary means of analysing the requirements of the new IS. Methodologies for affecting organizational change in the IS area have received great attention since the early 1980s. Of these, sociotechnical methodologies claim to give as much attention to the needs of the people involved as to the demands of the technology. ETHICS (Effective Technical and Human Implementation of Computer Based Systems), a design methodology developed by Enid Mumford and her colleagues, evolved from the sociotechnical stable; it promotes user participation as a major element in the system design process. Mumford (1994) summarized ETHICS as ‘a means of increasing [user] participation and facilitating good communication’ (p. 303). It is widely claimed that user participation in the design of systems is an important factor in their successful implementation (Mumford and Weir, 1979; Mumford, 1983, 1993, 1994, 1995, 1996; Baroudi et al., 1986; Franz and Robey, 1986; Tait and Vessey, 1988; Avison and
Catchpole, 1992; Gartner and Wagner, 1994; Mumford and Beckman, 1994; Wong and Tate, 1994). According to Lucey (1995) participation has ‘a wide range of meanings; from mere consultation at one extreme to full control at the other’ (p. 116) in terms of those affected by the situation. Yet, the notion that participation is always bene cial is arguable. For example, Lucey (1995) stated that ‘Although there is an intuitive feeling in us all that participation is always good, it may not always produce bene cial results in practice unless certain criteria are met’ (p. 116). Empirical research on participation seems to indicate that the support for the ‘success’ claim is modest (Franz and Robey, 1986) or tentative (Baroudi et al., 1986). Indeed, Cavaye’s (1995, pp. 317–18) recent comparative study showed that only one-third of such research apparently supports a positive link between user participation and system success. Westrup (1997), who has closely examined the concepts of user and user participation in the IS context, concluded that ‘Despite the arguments raised, user participation in systems development is both desirable and useful’ (p. 200). Notwithstanding the reservations expressed in the literature, we chose ETHICS as our methodology for the project because of its strong participation
Journal of Information Technology ISSN 0268–3962 print/ISSN 1466–4437 online © 2000 The Association for Information Technology Trust
40 element, which is considered very important in this case. For the purposes of this paper the word ‘participation’ is used in the sense of a means to an end (i.e. means to meeting another objective, e.g. ef ciency or effectiveness gains) rather than an end in itself (for the enskilling of people) (Brown, 1996, p. 212). The aim of our project was to redesign the organizational functions and processes of a university’s (central) information technology (IT) support service. The rationale is that, with ever-increasing use of IT, both in terms of the number of users and the diversity of applications, providing IT support on an institutional basis has become a major problem. Another dif culty is that hundreds of thousands of freshers (potential new users) enter higher education every year and the entire student population changes every 3 or 4 years. Moreover, externally funded national initiatives set up to promote the use of computers in teaching and learning activities create additional IT support requirements. To exacerbate these dif culties, student numbers have doubled in the last 5 years without a proportionate increase in support staff. The institution in this particular case was experiencing the aforementioned pressures and wanted to overcome these dif culties by increasing the ef ciency and effectiveness of its IT support service. With these objectives, one of the authors was asked to explore the problem situation with a view to formulating a suitable new model. The authors took the critical systems thinking (CST) approach; the choice of methodology was guided by total systems intervention (TSI), a meta-methodology devised by Flood and Jackson (1991). The operationalization of CST through TSI was discussed by the authors in a previous paper (Warren and Adman, 1997). The potential value of CST to information systems is discussed in Jackson (1997). In the early stages of the project it became clear that any organizational change had to be accompanied and supported by an appropriate IS; the design of that system therefore also became part of the project. The totality of the project was then undertaken as a major exercise in which the existing system was analysed, redesigned and implemented by the professionals for themselves (as much as for others). The indications are that participation was a key factor in the favourable (initial) acceptance of the new system by the staff concerned. The primary aim of this paper is to test the principles of participative sociotechnical design of organizations and their IS in a practitioner-led action research case study and, within the same context, examine the extent to which the ETHICS methodology is underpinned by viable systems methodology.
Adman and Warren Research methodology Case history The setting is a typical medium-sized university in the UK where the student population has grown from 6000 to aproximately 12 000 students in recent years. The academic computing services unit (the ACSU) offers a campus-wide IT/IS support service to all staff and students. It employs between 30 and 40 staff who are of a mixture of academic-related, technical and clerical grades. The trigger for this project was the problematic situation concerning the ACSU which had arisen gradually over the years. Within the ACSU the problems were perceived as follows. The volume and variety of requests for IT support, particularly at the end-user level, had reached saturation point, straining already over-stretched resources. On occasions, this resulted in lengthy (physical) queues at public access points in the ACSU, i.e. the reception and the help desk. Exacerbating the situation was the lengthening of ‘virtual’ queues of problems raised by telephone or e-mail. Users seemed to be unclear about how best to raise their technical problems with the ACSU and who or which section to contact for the best results. Most users followed the recommended route when raising their problems with the ACSU by approaching the reception or help desk, while some ‘powerful’ users would attempt to jump the queue(s) by contacting the ‘expert’ directly. In some cases, users would raise the same problem with other ‘expert(s)’ while it was already being dealt with by another, without the ACSU being aware of it. In this scenario it was dif cult for the ACSU to monitor the trail (or the ownership) of the problems brought into the ‘system’. When the long-serving director of the ACSU decided to retire and the university created in its place a new, more senior post – dean of information systems (DIS) – it seemed appropriate to have a fresh look at these perceived problems and their likely causes and possible solutions. To that end, a project team was formed consisting of four ACSU staff, with one of the authors of this paper designated as its head. The team members conducted informal interviews with various staff within the ACSU and a few users outside. The project team also observed the help desk in action from the viewpoint of the users approaching the ACSU for help. These observations were conducted unintrusively over a 6-week period during which a team member took random snap-shots of the dynamics of the support activities around the reception and the help desk. The perceptions of the ACSU staff seemed to be that appropriate technical skills were already in place to deal with user problems – thanks largely
Adaptation of ETHICS methodology to an extensive and continuing internal training programme. They felt that it was the combination of staff shortages and excessive user demands and expectations which was causing the queues. The lack of mechanisms for tracking the progress of the problems through the system was also recognized as a contributing factor. They felt that the ACSU’s staf ng levels and organizational framework needed rethinking to enable them to overcome the dif culties experienced. With little likelihood of a signi cant increase in staf ng levels, it was decided to examine the ACSU’s support activities more closely and, if necessary, redesign its structures around (and driven by) its front line service. Approach The next stage involved the selection of an appropriate approach for further exploration of the problem situation and, if necessary, to facilitate the redesign of it. The authors chose a methodological approach for reasons of added legitimacy and accountability. We drew on – a meta-methodology devised by Flood and Jackson (1991). For the debate on the philosophical underpinning of TSI, the reader should see, for example, Ormerod (1992, 1996), Tsoukas (1992), Mingers and Brocklesby (1995) and Midgley (1996). The known characteristics of the institution led to the choice of the ‘brain’ and ‘culture’ metaphors as most appropriate for the case study, which is discussed more fully in Warren and Adman (1997). The subsequent use of the system of system methodologies (Jackson and Keys, 1984) pointed towards a ‘sociotechnical design’ methodology and more speci cally to ETHICS.
Figure 1
Methodological interrelationships
41 ETHICS was developed by Professor Enid Mumford (Mumford et al., 1978; Mumford and Henshall, 1979). It is a problem-solving methodology aimed at identifying cause–effect relationships for solving problems and could be classed as positivist by implication. ETHICS incorporates the philosophies of both participation and sociotechnical design. The aim in ETHICS is not necessarily to produce a computer-based solution; the emphasis is on participative design with an appropriate balance between the social and technical aspects of the complete system. Our interpretation of the relationship between the main strands is shown in Fig. 1. Figure 1 implies a slight nal slant in ETHICS towards softer approaches, which is acknowledged by Mumford and Beekman (1994). They con rmed that ETHICS is more useful for non-technical aspects of systems design. We now consider the two main strands. The principal objective of the sociotechnical approach is to make work more satisfying for all concerned while achieving a high level of technical ef ciency. In this approach equal importance is attached to human factors such as ‘needs’ and ‘job satisfaction’ and ‘technical factors’. The sociotechnical design process involves setting objectives and specifying alternatives separately for the social and technical components before bringing those alternatives together and ranking them; then consideration is given to the resource implications and constraints before nally selecting the best sociotechnical solution. The second main strand of ETHICS is participation. Mumford (1994) saw participation as assisting freedom by sharing information, discussing alternatives and making choices. Effective, open communication enables sharing to take place amongst participants.
42
Adman and Warren
Ideally, all participants have an equal opportunity to raise issues, to call into question ‘the truth, appropriateness and sincerity of what is said’ (Mumford, 1994, p. 304) and ‘express attitudes, feelings, concerns and doubts’ (Hirschheim and Klein, 1994, p. 90). ETHICS addresses participation by setting up a steering group to provide guidelines and a design group to carry out the design process – led by a speci c person identi ed as the ‘facilitator’. Mumford (1983) distinguished between the three essential elements of participation: structure: (i.e. participation mechanism), content (i.e. remit: issues and boundaries) and process (gaining of relevant information and knowledge). She identi ed three levels of participation: consultative (lowest level of participation – traditional), representative (mixed group of users and analysts) and consensus (attempts to involve all user staff – long process). ETHICS consists of a set of logical steps which, in Mumford’s (1983) version, amounted to 15; a more elaborate version was later described by Hirschheim (1985), which consisted of 25 steps, showing a greater separation of the social and technical issues. Although ETHICS has changed over time, the main steps remain as summarized in Table 1. ETHICS adopts an analytical framework which was developed by an American sociologist (Parsons and Shils, 1951) for describing and measuring job satisfaction (steps 2 and 3 in Table 1). In contrast, QUICKethics, the ‘front end’ to ETHICS (Mumford, 1995, p. 79), uses fewer steps and is often deployed for requirement analysis. Avison and Fitzgerald (1995) explained that these steps are sometimes organized as a four-act drama (as shown in Table 2). In our study both ETHICS and QUICKethics were used in different stages, adapted suitably for the Table 1 1. 2.
3. 4. 5. 6. 7.
Design Values During the preliminary stage it became apparent that there were shared values and beliefs amongst the ACSU staff; this was not totally unexpected since most of them had worked there for a long time. These values and beliefs (about their role in the provision) seemed to be more consistent within each of the two main sections of the ACSU than perhaps between them. Boundaries The ACSU is probably not very different from most IT departments in local government, the civil service or the health service. What makes the ACSU different is its environment, which affects the overall aims of the service offered in support of academic activities – research and learning. Most of these activities involve the use of IT, needing varying support from the ACSU. Such is the nature and frequency of the interaction between the users and the ACSU that the users are not (or not seen as) totally external. They are almost part of the system but not quite (i.e. ‘outer boundary’). And, unfortunately, they are a variable component since some 2000–3000 new students present themselves as new users every year. The design group therefore needed to take special care in handling the ACSU’s boundary with its users. There was also another boundary to consider, that is the ‘inner boundary’ between the design group and the rest of
The steps in ETHICS
Description of mission and key tasks. Why does the department exist? What are the important tasks, main constraints and critical success factors? Diagnosing user needs and problems: day-to-day, ef ciency needs, effectiveness needs, job satisfaction needs (related to personality knowledge needs and psychological needs, related to competence, control and ef ciency and related to employee values, i.e. ethical needs) Setting ef ciency and job satisfaction objectives for the new system from 1 and 2 above Developing a number of alternative design strategies and matching them against the objectives Choosing the strategy which best achieves both sets of objectives Implementing the new system Evaluating it once it is operational
Table 2 1. 2. 3. 4.
particular problem situation. How well our practice matches the principles of ETHICS will be discussed later.
The steps in QUICKethics
Self-re ection: description of the individual managers work, information needs – ‘essential’, ‘desirable’ and ‘useful’ Self-identi cation; mission, key tasks, critical success factors and major problems Group discussion Group decision: ‘essential’ needs which are common become a core module for the IS system and ‘desirable’ needs help guide future development
43
Adaptation of ETHICS methodology
Figure 2
The ACSU’s environments, boundaries and interactions
ACSU who were not directly (or wholly) involved in the design but who nonetheless would be affected by the new system in a signi cant way. The exible nature of participation in the design process of this project meant that this particular boundary became very blurred. Figure 2 illustrates these boundaries. Structure A two-tier structure was established to manage the design process. The steering group, comprising the DIS and two senior members of the ACSU, was formed to provide overall guidance to the project as and when necessary. The design group consisted of four permanent members (the original project team) and several other staff attended one or more meetings either by invitation or they asked to attend themselves. The DIS designated the facilitator (indicated by F in Fig. 3), who (a departmental head) in turn selected the other three members and chaired the meetings of the design group. Inevitably, the structure of the participation was to determine the extent to which certain groups were able to in uence decisions. The design group met on several occasions, while the steering group met three times. As can be seen in Fig. 3, the design group was also represented in the environment group which is responsible for the accommodation of the ACSU and the physical conditions relating to it. Any design issue which could have an ‘environmental’ consequence was also discussed there.
Figure 3
Composition of various groups
Process The main stages of the design process were identifying the mission, key tasks, variances and job satisfaction; these all required information which was collected over a period of time and in a variety of ways (shown in Table 3). The facilitator then collated the information gathered and brought it to the design group for discussion. The mission of the ACSU was identi ed by the design group with some guidance from the steering group. The next stage involved identifying the key tasks performed (or to be performed) followed by the diagnosis of needs (Mumford, 1993), as summarized below. (1) Separate lists of the key tasks were produced by each staff member within the ACSU. They were speci cally asked to highlight those tasks which crossed over (linked) to the other section(s) or subsections. (2) Ef ciency needs were established through analysis of ‘variance’. Mumford (1995) de ned variance as ‘a tendency for a system to deviate from some desired or expected norm or standard’ (p. 80). Each ACSU staff member was asked to prepare an itemized ‘list’ of their thoughts on the matter. The design group then drew up two composite lists: one for the ‘key’ variances (with a view to controlling them) and the second for operational variances (with the intention of reducing them).
44
Adman and Warren
Table 3
Information gathering
Process
Information source
Gathered by
Identify the mission Key tasks Variance analysis Ef ciency needs (problems) Effectiveness needs (CSF)a
Design group and steering group discussions Design group discussions
DIS and facilitator Design group members and facilitators
Design group discussions, lists, interviews Selective interviews with ‘key’ ACSU staff and users Questionnaires and interviews
Design group members Design group members
Job satisfaction a
Design group members
CSF – Critical Success Factors
(3) For effectiveness needs the focus was the critical success factors (CSFs) (Rockart, 1979). Firstly, the design group considered (a summary of) the CSFs from the viewpoint of the ACSU (designated as CSF1) which were based on individual responses given to the relevant questions. The same questions were put to some ‘key’ users; their responses were taken to be representative of the (wider) user viewpoint (designated as CSF2). The design group found substantial overlap between CSF1 and CSF2. Some consideration was also given by the design group to the opinions expressed by both groups (the ACSU and ‘key’ users) on appropriate outcomes (as well as inputs, i.e. the CSFs). From the combination of all these sets of factors emerged the desired characteristics of the new system such as the ‘one-stop shop’, ‘problem logging’ and ‘problem ownership’. Eventually these were to form the kernel of the new system. Job satisfaction needs were assessed on the basis of ‘ t’ in ve categories: task structure, knowledge, ef ciency, psychological and ethical. Job satisfaction was measured as the gap between the needs of the ACSU staff and their experience of the system; the larger the gap the worse the ‘ t’. If the ‘ t’ was bad, then one would need to question as to why that is. The job satisfaction questionnaires were useful in identifying problems but not giving the reasons for problems and bad ts (Mumford, 1993). Follow-up interviews or
Figure 4
A viable information system (from Mumford, 1995)
group discussions were used by the design group to help to provide explanations. Many of the solutions were incorporated into the model. On the whole, the use of questionnaires was avoided because of the ACSU staff’s known dislike of lling out questionnaires, except for the ‘job satisfaction’ information. For the information requirements of the (new) system the design group used QUICKethics – a front end to ETHICS which claims it is based on the viable systems methodology. QUICKethics identi es ve levels of activity, as illustrated in Fig. 4. Each ACSU staff member was asked to list the information sources or devices that they thought would help the ACSU to function more effectively. Ranking high on their lists was the on-line availability of sources such as, for example, frequently asked questions (FAQs) and frequently performed tasks (corresponding to levels 1 and 2 in Fig. 4). Other suggestions included the provision of an ‘electronic moticeboard’ (for level 1) and ‘problem log’ and ‘caller-queue’ (for levels 2–4).
Implementation and evaluation ETHICS does not seem to offer much practical help on the implementation of the model. In our case the implementation was led by the project team with the help of section leaders. It amounted to a number of staged ‘dry run’ sessions undertaken over 3 weeks. Each dry run was preceded by a short
45
Adaptation of ETHICS methodology brie ng talk and followed by feedback. The evaluation of the system in place is a continuing process with a summary report presented to the DIS at 6-monthly intervals. Table 4 summarizes the various phases of the project, associated activities and the groups involved in those activities.
Model A single model gradually emerged from the discussions of the design group. It went through several iterations to accommodate various suggestions made in consultations along the way. In the end the group were reasonably con dent that the model would be well received because the likely reactions to the model of the ACSU staff and the users had been anticipated. This was achieved by widely disclosing the central themes of the model at various intermediate (prerelease) stages and encouraging informal debate over the relevant issues. The model has as its core a single access and coordinating point for all ACSU activities. It provides various internal mechanisms to facilitate the rapid and seamless escalation of problems through different sections within the ACSU. To work successfully the model requires more a change of perception or emphasis within the ACSU than a dramatic change in work practices. It brings together the reception and help desk including their staff and relocates them physically in a more central location. This is the central hub which the users access through a variety of ways (in person, by telephone or by e-mail). The matter raised (e.g. problem, inquiry or request) is recorded in the problem log (see Fig. 5) and taken on board as a ‘job’ by the ACSU. Thereafter, it is followed up until its successful closure – when the user is informed of the outcome. The central (single) contact point is named the ‘advisory point’, the change of name re ecting the new role. It is staffed by the adviser and Table 4
three IT assistants (ITAs). The latter were previously the reception staff but they were recast (and regraded) as ITAs after being given training to provide additional rst-level support. In addition, a hardware technician is on call nearby. The system (but not the staff) is monitored continuously by a ‘coordinator’ to ensure that the ‘jobs’ in the problem log are progressed rapidly and are not held-up. Their responsibilities cut across sections and subsections within the ACSU. This new position was lled by internal redeployment (also creating a chance for career development for the individual concerned). A crucial element of the model is the software for supporting the entire operation. This software had to be written in-house to t the special requirements. The whole ACSU is positioned to work around the software and the problem log. Each activity concerning a job is recorded there and remains visible to and observed by every ACSU staff member continuously throughout the day. While performing other duties for the ACSU, the staff are alerted automatically when a job arrives via the log for their attention. The ow of jobs from one ‘expert’ to another through the system is facilitated primarily by the software and overseen by the coordinator. The main features of the model are as follows. (1) (2) (3) (4) (5) (6) (7) (8)
A single point of contact (one-stop shop). Call logging (for all problems). Problem tracking. Problem ownership. Report back to the user. Support for remote users. De ned levels of response. Service performance indicators.
On release, the model was described and explained to all ACSU staff in three staged workshops, noting any suggestions made along the way. Finally, the decision was taken unanimously to accept it and implement it as described – but not without some trepidation!
Project phases, activities and groups involved
Weeks
Phases
Activities
Groups involved
1–6
Preliminary
Project team
7–34
Design process
Exploring the problems: interviews and observations Meetings, interviews and questionnaires
10–13 14–34 35 36–39 40–43 47+
Information requirements IS Release of the model Training Implementation Evaluation (ongoing)
QUICKethics Design, develop and test Workshops Training sessions Brie ngs, ‘dry runs’ and feedbacks Questionnaires and observations
Design group, steering group and environmental group Project team and section leaders IS development team DIS and project team Project team and training section Project team and section leaders Project team
46
Figure 5
Adman and Warren
Layers of support and problem escalation
Re ection Our research interest in this study was to test the principles, tools and techniques involved in the participative, sociotechnical design in a real problem situation through ETHICS methodology. In this section we rst present a brief critique of ETHICS, then go on to consider the extent to which the principles of sociotechnical design were upheld in our adaptation. More speci cally the degree of ‘match’ between the theory and practice is assessed. Subsequently, we highlight some of the areas covered inadequately by ETHICS, which could be addressed in further research. Brief critique of ETHICS Much has been written about ETHICS over the last 15 years or so. It originated from the eld of organizational behaviour and is classed as a process-oriented methodology. The three key elements discussed above – participation, human and technical factors – are recognized in the literature as its main strengths. Also recognized, however, are its weaknesses. For example, it is claimed that ETHICS assumes that (1) consensus can somehow be reached through participation, (2) the ‘client’ is the management, (3) ‘organizational needs’ are known or drawn by the management and (4) the boundary of the design is determined by the management. These seem to be valid points. Mumford (1994) talked about ‘organizational needs’, but Jayaratna (1994, p. 152) argued that organizations do not have needs, objectives or goals; these are formulated by powerful groups or individuals (management) for the organization. He concluded that ‘By de ning these as organizational needs, the methodology effectively removes any debate or discussion of the relevance, validity or viability of these needs’ (Jayaratna, 1994, p. 153).
It is also claimed that ETHICS does not provide (1) discussion on the context and conditions suitable for its use, (2) explicit models for managing social processes (Jayaratna, 1994, p. 155) and (3) guidelines for implementation. These points also seem valid, although it has to be said that there is no perfect methodology for any given situation. On that, Hornby et al. (1992) pointed out that ‘Experienced analysts can be described as “pragmatists”, using different methods as and when it suits them and their needs. They tend to “satis ce”, being heavily in uenced by the particular constraints under which they work, including the expectations of the others and the time, human resources and money available’ (p. 164). Jayaratna (1994) further pointed out that ‘ETHICS’ use of the cybernetic models of Beer does not extend to the more sophisticated notions of homeostasis or their recursive interlocking nature’ (p. 157). We shall examine this point in detail later in the paper. According to Westrup (1997), QUICKethics does not pay suf cient attention to the selection of users. He explained that ‘they are assumed unproblematically as either future users or representatives of future users’ (Westrup, 1997, p. 192). Westrup (1997) added that ‘[it] treats the issue of power in a naïve (as empowerment) and often contradictory way (sometimes as emancipatory at other times like business process reengineering)’ (p. 200). He further claimed that ‘What ETHICS/QUICKethics does not acknowledge is the capacity for technology to transform organizations’ (Westrup, 1997, p. 195). Case study experience Besides the boundary issues discussed earlier, the authors’ adaptation varied from the de ned methodology in a number of other ways. For example, the
47
Adaptation of ETHICS methodology Table 5
Design group’s remit
Criteria
Design group’s remit (ex-post)
Ability to take decisions? Membership?
Recommendation only (Selective) representation only Vague Vague Unde ned Frequent and informal
Design limits? Organizational restructuring? Implementation responsibility? Wider consultation?
participation and information gathering was somewhat selective and constrained timewise. Nor were there any alternative models developed as recommended in step 3 of ETHICS. Instead, a single model began to emerge and was developed iteratively from an initial, broad outline. Participation Participation is the touchstone in ETHICS and is therefore addressed extensively. Yet the methodology seems to allow enough exibility to re ne or modify the guidelines given or even revise them completely. Much recent work has been done in this area using varying approaches. For example, Newman and Robey (1992) used social process modelling for dealing with participation in IS design, while Wong and Tate (1994) used a simple process modelling technique (i.e. Table 6
data ow diagrams or DFDs). Tait and Vessey (1988), on the other hand, took a ‘contingency’ approach towards user involvement and measured the impact of variables. In our case study, no attempt was made to formalize the participation process. Instead, the facilitator simply set out to avoid the most common problems. These were identi ed by Mumford (1995) as ‘trust’, ‘election versus selection’, ‘con icts of interest’, ‘stress’, ‘communication and consultation’ and ‘the roles of the designers and the managers’. Of these communication and consultation was in our control and could be affected fully and ‘trust’ was considered a crucial ingredient, although obviously dif cult to create (and maintain). These then came to signify our outlook on participation. Barki and Hartwick (1989) argued for an alternative de nition; they separated the constructs user participation (‘a set of behaviors or activities performed by users in the system development process’ (p. 53)) from user involvement (‘subjective psychological state’ (p. 53)). As practitioners we see a range of ‘participation’ and ‘involvement’ in any design project rather than a uniform one. Moreover, we are aware that differing cultural in uences/values (be they racial, occupational or whatever) can have a signi cant effect on the priorities for participation and how it might be structured. In our case study, participation was partly direct and partly indirect (‘representative’); overall, it was more ‘consultative’ than ‘democratic’. The facilitator was a
Sociotechnological problems
Sociotechnological principles (theory)
Case study (practice)
The principle of compatibility The design process involved shared problem solving and decision taking through group discussion at design group meetings and communicating with the other groups (see Fig. 3) – but note Table 5 The principle of minimal critical speci cation
The model was designed with minimal job speci cation
The principle of variance control
Groups were given the responsibility for identifying and correcting problems that might arise from their (group’s) work
The principle of multiskilling Wherever possible the staff were designated jobs in which they could use their multiskills. Additional training opportunities were provided towards further multiskilling The principle of boundary management
Particular attention was paid to this aspect to ensure that the various boundaries identi ed earlier are managed properly (achieved mainly by inclusion)
The principle of information ow
IS were designed (e.g. problem log, caller queue, etc.) so that information goes rst to the group that would use it or have produced it
The principle of design and human values
The objective of providing a high quality of working life for all participants was considered throughout the design process
The principle of incompletion The design process went through several iterations; at each stage a current working version was produced for further discussion. The new system will be evaluated every 6 months and modi ed appropriately.
48 departmental head (rather than independent as recommended by Mumford (1993)) and chaired the design group’s meetings; this was considered a drawback as it inevitably led to a rational slant and hierarchical dominance rather than a strictly collegial approach. In retrospect, it became clear that a number of questions should have been asked (and answered) about the design group’s remit before the project started as recommended by Mumford (1995) (see Table 5). This was an oversight and was probably attributable to the general uncertainty in the institution at that time. Pragmatically, the ultimate choice of ETHICS was partly owed to our familiarity with the methodology and our preparedness to accept (broad) management goals and boundaries. Most of all, it appeared to us that the designers’ role in ETHICS can also be seen as that of teacher, adviser and learner – an appropriate approach given the nature of the organization in this case. We were not disappointed with the experience or the outcomes: the biggest gain is the readiness and enthusiasm with which the model seems to have been received by all. The apparent realization of the (hoped for) increased ownership of it could well be attributable to the participatory nature of the design process and the sociotechnical principles. To what extent the same (or similar) outcomes would have been achieved through an informal design process is only conjecture – but there are indicators. For example, the feedback from those concerned highlighted that having methodological underpinning was instrumental in leaving them with increased con dence in the process which had taken place. The rigour of a methodological approach made it more acceptable to them than any unstructured approach. In particular, they thought that through participation ETHICS offered them a very good way of exploring the problem situation (as experienced by them) right down to their personal level. They felt that at least they had a hand in shaping their own ‘futures’ rather than leaving it to an ‘expert’ (internal or external) who would not be likely to understand the issues involved in the particular situation. In the end, what seemed to matter most was not the solutions offered but actually being part of nding the solutions and participating in working towards them. Notably, participation was seen by the ACSU staff as having fostered good supporting relationships within and between the different sections. In doing so it (more than) compensated for the length of time taken to complete the project. Design principles Socio-technical design incorporates a number of work design principles (Mumford, 1996). In summary these are the principles of ‘compatibility’, ‘minimal
Adman and Warren critical speci cation’, ‘variance control’, ‘multiskilling’, ‘boundary management’, ‘information ows’, ‘design and human values’ and ‘incompletion’. Table 6 summarizes the actions taken in our case study to conform to these principles. We have learnt an important lesson from this study, which has in fact a more general relevance than the speci c methodology employed. That is, any new organizational design has to work mostly (if not exclusively) with the people already in place and people amount to more than their ‘needs’ and ‘job satisfaction’ requirements – they are also ‘personalities’. In our adaptation, the design group also decided to consider the ACSU staff as ‘personalities’. Somewhat subjectively we discussed what would and what would not work well with these ‘personalities’ in place and then steered the design to work around them. Effectively, ‘personalities’ became another critical factor in our design ‘equation’. This was not seen as shifting the balance against change and towards maintaining the status quo, rather it was taken as tting the design around the personalities (real) rather than tting the personalities into the new design (abstract). It could be that the nature and maturity of the organization in this case necessitated the slant in our approach. We feel that further research is needed on the issue raised in this paragraph.
Exploring ETHICS’ association with the viable systems methodology The methodological framework of ETHICS is said to be based on the viable systems methodology (Mumford, 1996), a cybernetic model developed by Beer (1979, 1981, 1984, 1985; see also Espejo and Harnden, 1989). The viable systems methodology is also said to underpin QUICKethics which is used mostly (but not exclusively) for the analysis of information requirements. We believe that the nature of the associations between the viable systems methodology, ETHICS and QUICKethics are complex and remains unquestioned in the literature. To open up the debate, we brie y consider the main characteristics of the viable systems methodology, what ETHICS has to say about them and the various steps we have taken in the case study to incorporate those characteristics into the design. Firstly, we observe that the viable systems methodology is de ned tightly; its theory has been examined rigorously and its practice has been reported widely for peer review. It consists of ve functional elements which are interconnected through various information and control loops. The ve elements are ‘implementation’, ‘coordination’, ‘control’, ‘development’ and ‘policy’;
49
Adaptation of ETHICS methodology these correspond to ‘systems’ 1–5, respectively. In contrast, ETHICS is de ned loosely; it is described more in terms of concepts and philosophies than methodologies. The organizations and IS designed using ETHICS are often described as having ve levels of activity, named ‘levels’ 1–5 to mirror the viable systems methodology’s ve systems. However, the levels are depicted hierarchically (hence the name level), which is not the case in the viable systems methodology. The difference between the viable systems methodology and ETHICS in this regard is more than super cial. Jayaratna (1994) examined ETHICS closely under his own meta-methodology (called Normative Information Model-based Systems Analysis and Design (NIMSAD)), highlighting as drawbacks the lack of details and guidance beyond its early stages and, above all, ETHICS’ stance of ‘taking the problem as given’. Here we take a different line and concentrate on the characteristics of the viable systems methodology which differ from ETHICS, in particular ‘variety’, ‘performance measurement’, ‘recursion’ and ‘linkage’. Variety (engineering) ‘Variety’ is one of the building blocks of cybernetics and can be de ned as the number of possible ways a system is capable of exhibiting from a given point of view. To be viable a system must exhibit at least as much variety as its environment (‘the law of requisite variety’; Ashby, 1964). In other words, the variety balance between the system and its environment is important for viability. The variable systems methodology offers two devices which can be used to redress any variety imbalance; these are variety reduction (of the environment) and variety ampli cation (of the system being designed). Brie y, the following are possible ways of achieving these (Jackson, 1990). (1) Variety reduction (or ltering) – of its operational and environmental variety: (i) structural (e.g. divisionalization and functionalization), (ii) planning (e.g. setting priorities), (iii) operational (e.g. management by exception) and (iv) rule setting (e.g. instructions, etc.). (2) Variety ampli cation – of a system’s own variety: (i) structural (e.g. integrated teamwork), (ii) augmentation (e.g. recruit experts and employ consultants) and (iii) informational (e.g. management IS). While the viable systems methodology addresses ‘variety engineering’ extensively as intimated above, in contrast there seems little in ETHICS on the issue (as distinct from the ‘variance analysis’ referred to earlier under the information gathering stage in ETHICS). In our case study, variety engineering of
various types was exercised – points (1(i) and (iv)) and (2 (i) and (iii)) in particular and point (1 (iii)) to a lesser degree. For example, by discarding some of the functions, streamlining others, increasing automation, changing priorities (for variety reduction) and facilitating additional training, introducing a new management IS and team working (for ampli cation), we were able to create a new system capable of matching environmental variety. Measures of performance This issue does not seem to be addressed at all by ETHICS. However, in the viable systems methodology performance measurement is crucial (Beer, 1984) and is addressed by adopting three levels of achievement: ‘actuality’, ‘capability’ and ‘potentiality’. From these three indices one can obtain productivity, latency and performance, which are de ned as (1) productivity = actuality/capability, (2) latency = capability/potentiality and (3) performance = actuality/potentiality (equivalent to latency 3 productivity). These measures give a better indication of a system’s performance than cost alone. However, they are dif cult to quantify with any degree of accuracy. The way in which our design group addressed this problem was to produce a cumulative (weighted) list from each participant’s weighted lists of what they considered the ACSU actually did, what it was capable of doing and what potential there was for other functions/activities. We were then able to estimate the performance expected from the system roughly; while doing so we were concerned about the subjectiveness of the weights associated with each element and, for that matter, the crude method of combining them. We believe that more research is needed to discover methods that can be applied in this context and whose validity and reliability can be judged. Recursion and linkage Recursion is inherent in the viable systems methodology and is essential for a system’s viability, as are the information channels which link systems 1–5 and the system and its environment. ETHICS, on the other hand, does not seem to provide any guidance on these. In our case, an element of recursion was introduced by creating design subgroups to replicate the design process and the system functionality, appropriately at the subsystem level. Clearly this was a scant attempt to achieve recursion, but it did show a possible way forward in ETHICS in this regard. On linkage, our ad hoc solution was to introduce crosssystem link members in the design and management
50 of information linkages across systems 1–5; similarly a senior manager was designated to do the same between the system and its the environment. We believe that, by incorporating these we managed to create an adaptation of ETHICS which has a stronger association with the viable systems methodology than the methodology currently possesses.
Conclusions In this paper, we set out to provide further insight into the participatory sociotechnical design of organizations and IS through a case study using ETHICS. We have also presented an examination of the claimed association between ETHICS and the viable systems methodology. To us, ETHICS seems more an embodiment of a philosophy – a position. The case study has shown that it is feasible to adapt ETHICS and deviate from some of its speci c guidelines while remaining within its overall principles. However, there is an argument that the theoretical underpinning of ETHICS, in particular its connection with the viable systems methodology needs further attention. In the ETHICS literature, references to the viable systems methodology do not seem to offer much in the way of grounding the argument. Yet the evidence suggests that one can extend the viable systems methodology and, indeed, add other arguments to support the practices of ETHICS. Some possible avenues for further exploration have been identi ed in this paper.
References Ashby, W.R. (1964) An Introduction to Cybernetics (Methuen, London). Avison, D. and Catchpole, C.P. (1992) Information systems development: the importance of participation. University of Southampton Discussion Papers in Accounting & Management Science 38, 1–11. Avison, D. and Fitzgerald, G. (1995) Information Systems Development (McGraw Hill, Maidenhead). Barki, H. and Hartwick, J. (1989) Rethinking the concept of user involvement. MIS Quarterly 13(1), 53–63. Baroudi, J.J., Margrethe, H.O. and Blake, I. (1986) An empirical study of the impact of user involvement on system usage and information satisfaction. Communications of the ACM 29, 232–8. Beer, S. (1979) The Heart of the Enterprise (Wiley, Chichester). Beer, S. (1981) The Brain of the Firm (Wiley, Chichester). Beer, S. (1984) The viable system model: its provenance, development, methodology and pathology. Journal of the Operational Research Society 35, 7–25.
Adman and Warren Beer, S. (1985) Diagnosing the System for Organisations (Wiley, Chichester). Brown, M. (1996) A framework for assessing participation, in Critical Systems Thinking – Current Research and Practice, Flood, R.L. and Romm, N.R.A. (eds) (Plenum Press, New York and London) pp. 195–213. Cavaye, A.L.M. (1995) User participation in system development revisited. Information and Management, 28, 311–23. Espejo, R. and Harnden, R. (1989) Cybernetics and Organizational Theory: A Critical Review (Wiley, Chichester). Flood, R.L. and Jackson, M.C. (1991) Creative Problem Solving: Total Systems Intervention (Wiley, Chichester). Franz, C.R. and Robey, D. (1986) Organizational context, user involvement, and the usefulness of information systems. Decision Systems 17, 329–51. Gartner, J. and Wagner, I. (1994) Systems As Intermediaries: Political Frameworks of Design and Participation, Proceedings of the III Participatory Design Conference (Chapple Hill, Carolina), pp. 37–46. Hirschheim, R. (1985) Information systems epistomology: a historical perspective, in Research Methods in Information Systems, Mumford, E., Hirschheim, R., Fitzgerald, G. and Wood-Harper, T. (eds) (Elsevier Publishers BV, Amsterdam). pp. 13–36. Hirschheim, R. and Klein, H.K (1994) Realizing emancipatory principles in information-systems development – the case for ETHICS. MIS Quarterly 18(1), 83–109. Hornby, P., Clegg, C.W., Robson, J.I., Maclaren, C.R.R., Richardson, S.C.S. and O’Brien, P. (1992) Human and organizational issues in information systems development. Behaviour and Information Technology 11(3), 160–74. Jackson, M.C. (1990) Organisational Design and Behaviour – Study Guide (Lincoln School of Management, Lincoln). Jackson, M.C. (1997) Critical systems thinking and information systems research, in Information Systems: An Emerging Discipline Mingers, J. and Stowell, F. (eds) (McGraw Hill, London), pp. 201–38. Jackson, M.C. and Keys, P. (1984) Towards a system of systems methodologies. Journal of Operational Research 35, 473–86. Jayaratna, N. (1994) Understanding and Evaluating Methodologies – NIMSAD: A Systemic Approach (McGraw-Hill Book Company, London). Lucey, T. (1995) Management Information Systems (DP Publications, London). Midgley, G. (1996) Mixing Methods: Developing Systemic Intervention (Hull Centre for Systems Studies, Hull). Mingers, J and Brocklesby, J. (1995) Multimethodology: Towards A Framework For Mixing Methodologies (Warwick Business School, Coventry). Mumford, E. (1983) Designing Participatively (Manchester University Press, Manchester). Mumford, E. (1993) Designing Human Systems for Health Care – The ETHICS Method (Eight Associates, Stockton Heath). Mumford, E. (1994) Technology, communication and freedom: is there a relationship? IFIP Transactions A – Computer Science and Technology, 49, 303–22.
51
Adaptation of ETHICS methodology Mumford, E. (1995) Effective Systems Design and Requirements Analysis: The ETHICS Approach (MacMillan Press Ltd, Basingstoke). Mumford, E. (1996) Systems Design – Ethical Tools for Ethical Change (MacMillan Press Ltd, Basingstoke). Mumford, E. and Beekman, G.J. (1994) Tools for Change & Progress: A Socio Technical Approach to Business Process ReEngineering (CSG Publications, Leiden, The Netherlands). Mumford, E. and Henshall, D. (1979) A Participative Approach to Computer Systems Design (Associated Business Press, London). Mumford, E. and Weir, M. (1979) Computer Systems at Work Design – The ETHICS Method (Wiley, New York). Mumford, E., Land, F. F. and Hawgood, J. (1978) A participatory approach to computer systems. Impact of Science on Society 28(3), 235–53. Newman, M. and Robey, D. (1992) A social process model of user–analyst relationship. MIS Quarterly 16(2), 249–66. Ormerod, R. (1992) Combining Hard & Soft Systems, (Warwick Business School, Coventry). Ormerod, R. (1996) The Design of Organisational Intervention: Choosing The Approach, (Warwick Business School). Parsons, T. and Shils, E. (1951) Towards a General Theory of Action (Harvard University Press, Cambridge, Mass.) Rockart, J.F. (1979) Chief executives de ne their own needs. Harvard Business Review 57(2) 81–93. Tait, P. and Vessey, I. (1988) The effect of user involvement on system success: a contingency approach. MIS Quarterly 12(1), 91–107. Tsoukas, H. (1992) The Road to Emancipation is Through Organizational Development: A Critical Evaluation of Total Systems Intervention (Warwick Business School, Coventry). Warren, L. and Adman, P. (1997) The use of TSI in designing a system for a university IS user support service, in
The Third Australia New Zealand Systems Conference Proceedings, Queensland, 1–4 October, pp. 189–98. Westrup, C. (1997) Constituting users in requirements techniques, in Information Systems and Qualitative Research – Proceedings of the IFIP TC8 WG8.2, Amsterdam 1994, pp. 182–204. Wong, E.Y.W. and Tate, G. (1994) A study of user participation in information system development. Journal of Information Technology, 9, 51–60. Zmud, R.W. (1984) Design alternatives for organizing information systems activities. MIS Quarterly 8(2), 79–93.
Biographical notes Peter Adman is the assistant director of the computer centre at the University of Hull. He has extensive experience of the IS eld having worked in IS strategy formulation, user support and systems development for over 20 years. His interests extend to research, consultancy and teaching in designing organizations and their IS. Lorraine Warren is a senior lecturer in the department of business strategy at the Lincoln School of Management of the University of Lincolnshire and Humberside. She is currently active in research, consultancy and teaching in the elds of management and IS. Her research interests include soft methodologies for IS development and the management of new technologies in small businesses. Address for correspondence: Dr P. Adman, Computer Centre, University of Hull, Cottingham Road, Hull HU6 7RX, UK.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
