Methods and tools in user centred design for information technology

Methods and tools in user centred design for information technology

BOOKS Methods and tools in user centred design for information technology M Galer, S Harker and J Ziegler Elsevier, Amsterdam, 1992, 436 pp, ISBN 0 4...

258KB Sizes 0 Downloads 75 Views

BOOKS Methods and tools in user centred design for information technology

M Galer, S Harker and J Ziegler Elsevier, Amsterdam, 1992, 436 pp, ISBN 0 444 89301 6, 436 pp, $85.00 For many ergonomists, the presentations made by HUSAT researchers, concerning the H U F I T toolkit, to the annual Ergonomics Society conference a couple of years ago had an air of "bringing coals to Newcastle"; indeed, there was a feeling abroad that the coals were being delivered to properties which had already been converted to central heating! So my first response on seeing this book was what does the current H U F I T toolkit look like, and my second response was will it be useful to ergonomists? Before I answer these questions, it is worth pointing out that this book contains far more than a single "toolkit"; it represents a collection of reports from a large-scale ESPRIT project involving the following institutions: Bull, Olivetti, Siemens, ICL, Philips, HUSAT, Fraunhofer IAO, Westf~lisehe Wilhelms Universit/it and the University of Piraeus. Thus the project incorporated both academic and industrial collaborators, and contributions from across Europe, ie Greece, the UK, Germany, the Netherlands, Italy and France. The link between the various institutions was the ESPRIT project "Human Factors in Information Technology". Consequently, this is both a final report of the overall project, with contributions addressing multimedia, speech, direct manipulation, and a collection of methodologies. With reference to the latter point, there are three original methods in this book, each delighting in its own acronym (PAS, QED, SANE), together with several unnamed collections of techniques which were brought together for specific projects, such as one on product usability and one on IDSSs. Thus, although the book is divided into five sections (Supporting the design process from conception to use; Developments in interface design; Software tools to support user centred design; Exemplar of the application of user centred design; Conclusions), in this review I shall focus on two aspects: the methodologies and the discussions of interface design. Allison et al (Chapter 2: Human factors for designers of information technology software) n o t e a number of obstacles to employing human factors in the design process, cited by designers. From the table on p 14, it would seem that the most common objections were that human factors knowledge was not available, and that the methods would take too long to use and would not be cost-beneficial. Consequently, this leads to the old chestnut that it is necessary to " . . . match human factors methods more closely to design practice . . ." (p 14). Naturally this is not an easy exercise; there are considerable variations in design processes, and in organizations. This leads to a restatement of another Vol 23 No 5 October 1992

familiar concept: " . . . any approach which links the human factors contributions to a single one of these settings is unlikely to apply and be successful in all other settings" (p 15). While this work starts from a position close to the heart of many ergonomists, it progresses beyond a simple statement of fact and proposes a range of tools which offer a way beyond the somewhat defeatist attitude of ergonomists, that their work is ignored. The main goal of this work seems to be the provision of " . . . structured methods to help designers take account of user needs without adding significantly to the timescale or skill requirements of the work" (p 15). Thus the main beneficiaries of the proposed tools would be designers with little or no previous experience of human factors. In one sense, this might answer the second response above: the tools may be of limited value to ergonomists simply because they are designed for the non-ergonomist. This raises a number of issues for the ergonomics community. I wonder how useful the much-vaunted (here and elsewhere) goal of "giving ergonomics away" to non-ergonomists will actually be. By definition, ergonomics knowledge is nebulous, touching on a wide range of disciplines, and its skills are heavily dependent upon the experience and expertise of its practitioners. Consequently, I wonder whether researchers from other disciplines would necessarily assimilate information or view problems in the same way as an ergonomist. If this is the case, then simply elucidating 'key concepts' is insufficient; in order for the concepts derived from human sciences to be useful to engineers it requires, I feel, a paradigm shift on the part of the engineers. I am not sure that such a shift would produce benefits, particularly as it could require a loss or restructuring of existing knowledge and skills. In other words, why burden researchers and practitioners, who are highly competent in one field of endeavour, with information and methods from quite different fields? It would seem more useful for ergonomists to 'crusade' in terms of the benefits the discipline can offer, and for design projects to incorporate an ergonomist as part of the team. This is quite different from the avowed aims of the book (although, paradoxically, appears to be the approach in some of the projects reported). A further extension of this argument is that providing minimal knowledge of ergonomics increases the risk of 'tokenism' in project development; by allowing the non-ergonomist access to some, but not all, relevant information, aren't we in danger of condoning, or even promoting, 'bad' ergonomics? Surely one of the strengths of ergonomics is its ability to maintain a focus on the end-user, and to situate problems within a wider context of end use rather than immediate technical hiccoughs? Consequently, I feel that giving ergonomics to designers is naive and quite simply wrong. Having made this declaration, I would now like to proceed with a review of the tools this book aims to supply to non-ergonomists. In the introduction, Galer and Harker propose that PAS, "the planning, analysis and specification tools are 359

BOOKS

for use during the early stages of design . . ." (p 7). This particular toolkit comprises six tools (incidentally, I found it a little confusing that PAS could be both a tool and a toolkit, and would have preferred some consistency in the use of terminology here). The PAS tools are: • • • • • •

user mapping; user characteristics; task characteristics; user requirements survey; USE - usability specification for evaluation; functionality matrix (p 17).

Although I sympathize with the aims of these 'tools', I found it very difficult to derive sufficient information from this chapter to enable me to operationalize them; ie while it was clear from my reading what the aims of this approach were, it was unclear how these would be met. There are, however, seminars run by H U S A T which may perhaps help me (pp 28-30)! This chapter also contains a description of Q E D (quick ergonomic design), which functions as a 'primer' on human factors issues. Rather than discuss this in detail, I refer the reader back to my point concerning "giving ergonomics away" above. In Chapter 3 (An ergonomics framework for user activity centred software design), Jeffroy and Lambert present and develop a "course of action" framework which has been used by Bull in France. As they present them, the main principles of the framework are: • analysis focusing on the genesis of user actions, with identification of the meaning of actions from the user point of view; • clinical analysis of a set of actions, through verbal and action protocols; • analysis conducted in situations as close as possible to real work conditions (p 43). Again, while I am sympathetic to these proposals, I can find little in them which I do not regard as conventional ergonomics. There is, however, an argument, presented on pp 46-47, regarding "the paradox of ergonomics in design", which I would like to discuss. The point rests on the following quotation from Pinsky and Theureau (1984): "To make an accurate statement about the effect on the user activity, a new system has to be completely designed, but then it is too late to intervene in the design process." I found this argument both compelling and irritating. On the one hand, I would accept that there is quite simply insufficient (good, practical and applicable) theory in the field of ergonomics, which means prediction can be difficult. On the other hand, I am not convinced that a system needs to be complete before it can be evaluated (as this quotation implies). Jeffroy and Lambert suggest a route around the paradox via prototyping, but this is not without its own pitfalls and problems (Lim and Long, 1992). I felt that the 'paradox of ergonomics' argument rests on a definition of ergonomics derived less from ergonomics itself than from engineering; is it perhaps possible that the paradox would only arise if there was insufficient attention to the end user in the initial product specification? 360

In Chapter 4 (The SANE toolkit for cognitive modelling and user centred design), Brsser and Melchior present what appears to be the flagship of the book. The SANE (Skill Acquisition NEtwork) toolkit is a modelling language for defining cognitive operator procedures in HCI. The main aim underlying this approach is to permit product evaluation to be performed in conjunction with development, a goal I am sure all ergonomists would love to see achieved. Following a brief but very readable review of typical approaches used in HCI, ie prototyping, task analysis, job/work analysis, models of user knowledge etc., Brsser and Melchior present the fundamental principles of SANE together with a worked example based on the use of a cash dispenser. SANE is a " . . . representation of user knowledge and p r o c e d u r e s . . . Its basic unit is the o p e r a t i o n . . . " (p 105). B/Ssser and Melchior emphasize that their concern is with the relationship between user information requirements and device functions. This feels like a combination of TOTE and ETIT; it extends some of the previous approaches reviewed, appears to be very attractive and I would certainly like to see it used in anger. However, Brsser and Melchior also note that SANE assumes that the " . . . functions of the device, and the operations of the user of the device . . . are isomorphic" (p 109). I am not sure whether I read this correctly, but it certainly seems somewhat dubious, if impossible to substantiate; given the assertion, repeated throughout this book, that users often have difficulty in defining appropriate responses to computers, I found the notion that functions and operations are (rather than ought to be) isomorphic a little difficult to accept. Having said, this, I found the SANE work both original and interesting and would certainly advise anyone working in HCI to read this chapter. Chapters 5 (Product usability - Bertaggia et al), and 7 (Design of dialogues - Haubner) basically present well-written, readable reviews of central topics for HCI. Chapter 6 (The computer human factors database - Phillips and Galer) describes a relatively new database of HCI references. Again, although laudable, this chapter does not present anything new, particularly when one considers the size of the database in the Ergonomics Information Analysis Centre which is used to compile Ergonomics Abstracts. Finally, I found the chapters relating to applications, in general, very disappointing. Not only was there insufficient information concerning how the research was conducted, but there also appeared to be highly selective reporting of results so that it made interpretation of the work reported quite difficult. In conclusion, then, I am not sure that there is much in this book which would be of interest to either ergonomists or HCI practitioners (with the exception of B6sser and Melchior's chapter on SANE). As far as the non-ergonomist is concerned, the discussion of topics is very variable - sometimes assuming little or no knowledge, sometimes assuming a wealth of prior knowledge - which would, I feel, make application of many of the proposed 'tools' quite difficult.

Chris Baber Applied Ergonomics