Bulletin, October/November 2006
Cognitive Work Analysis
Special section of the Bulletin edited by Patricia Katopol
An Ecological Approach to the Design of Information Systems
by Raya Fidel
Raya Fidel is a professor and director of the Center for Human-Information Interaction at the Information School, University of Washington. She can be reached at fidelr<at>u.washington.edu.
Information systems are built to support people in their activities, whether on the job or in their personal lives. It makes sense, therefore, that the design of information systems should be based on an understanding of the activities people carry out. These activities, however, are not the same for all people at all times both because different jobs call for different activities and the same person may be involved in different activities at different times. This diversity introduces the idea that each type of activity may require its own information system.
Consider Nora, a hypothetical person, who is a professional meteorologist. Her job requires heavy use of information systems, but so does her private life. For example, she likes to do her shopping on the Web and to get information about movies she is planning to watch. Most search engines on the Web are general ones – they are supposed to work well for finding any type of information, be it expert meteorological data, shopping links or movie reviews. The search engines also provide the same functionality and interface for all purposes, whether Nora’s search is in support of analyzing data, writing a report or looking at pictures. When Nora uses the Web on the job, she retrieves items that are geared for the average citizen as well as those for the professional. Moreover, she cannot manipulate or display the information in ways that would be most useful to her.
To make Nora’s professional search more effective, she needs an information system that is customized for meteorologists – that is, a set of information sources, a search engine and an interface that fit the work that professional meteorologists do. Such a search engine, however, would not be helpful to Nora when she looks for movie reviews. A customized system for movie buffs would be much more effective for this purpose.
More generally, to design effective information systems for professionals requires analysis and understanding of the work they do. This dictum applies not only to professionals but to any other people involved in work that requires decision making – what we call “cognitive work.” Each type of cognitive work requires its own analysis if it is to guide the design of a customized information system. This way, design engineers, high school students, shoppers on the Internet, archivists, Web designers and other groups of people would each have their own information system tailored to fit their work. In other words, work analysis is a crucial step in the development of customized information systems.
The Cognitive Work Analysis Framework
As its name implies cognitive work analysis (CWA) is a framework that guides the analysis of cognitive work. Its purpose is to provide an understanding of the work people do in a way that is relevant for the design of information technology. The framework was developed by Danish researchers at Risoe National Laboratory in the early 1980s and already has been applied to the design of technology in various work environments, such as nuclear power plants, hospitals and manufacturing. Annelise Mark Pejtersen, who is one of the developers of CWA, was the first to apply it to the design of information systems when she developed a fiction retrieval system called Book House. The system provides various options for how to look for “a good book to read,” depending on how much a person knows about what he or she wants. It also supports the work of catalogers of fiction.
In addition to the work activities themselves, CWA analyzes the environment in which they take place. More specifically, it analyzes factors in the environment – which we call “constraints” – that shape the activities. Consider the activity of Web searching by high school students. Their environment includes many factors, but not all of them shape their activities in Web searching. Examples of factors that might do so are the level of funding the school receives, the level of interest students have in what they do or the technical support available. Other factors such as the publicity the school gets in the press, while important, might be less likely to shape how students search the Web. The goal of CWA is to identify these constraints through empirical analysis – that is, investigating actual activities and situations (rather than speculating what they are) – and to understand how they shape the way activities are performed. That is, CWA is an ecological approach – one that focuses on how environmental constraints shape activities.
An important object in CWA is the people who carry out the activities. As carriers of activities they are called “actors.” The analysis of actors is based on the assumption that people who belong to a certain group – be it meteorologists, Web designers or shoppers on the Internet – have common characteristics that make it possible for them to perform the activities effectively. So in an analysis of the work high school students do, for instance, CWA would investigate what is required for one to be an operational high school student.
As a framework, CWA provides a general structure and tools for work analysis. The structure identifies the major areas that should be investigated. These are called “dimensions of analysis” and are briefly described below. For each dimension, CWA also provides tools to support its analysis. The most common of these tools is the means-ends analysis (also called the abstraction hierarchy), and it, too, is presented here in brief.
Dimensions of Analysis
CWA proposes seven areas that require analysis when one wants to understand the work actors do. These areas are presented graphically (see the article by Jens-Erik Mai in this issue). At the center of the analysis are the actors themselves, their characteristics – such as their work experience, education and experience with the technology – and the values they hold. The actors are nested in six dimensions that shape their activities. On the outside is the environment of the work place. It includes all the factors that shape how the work place is run. In the high school case, state graduation requirements and the availability of technology are examples of such factors. Nested within the environment is the work place itself, or work domain, which is the arena in which the actors operate.
Zooming in to understand the actors’ interaction with information technology, the next dimension is the actual task (or tasks) that the actors carry out, and the decisions they have to make to perform their tasks. To study the interactions between the actors and their work setting one also performs an organizational analysis, looking at how work is divided, how collaboration is taking place, the prevalent management style and similar issues. Based on an understanding of these dimensions for a particular group of actors and on empirical evidence, the analyst can then identify what strategies actors can use to perform their tasks.
Investigations along these seven dimensions will create a comprehensive understanding of the constraints that shape the activities of the actors under study and how these constraints affect their activities.
Means-ends analysis is particularly effective when investigating the work domain, the tasks and the decisions that are made. As illustrated in Table 1, means-ends analysis consists of five interrelated levels of analysis, starting from the most abstract level (the goals and constraints) and concluding at the most concrete (the actual tools that are used).
|Table 1 The Five Levels of Means-Ends Analysis|
|Levels||Questions to be Asked|
|Goals/Constraints||What are the ultimate goals and purposes? What affects the work domain/task/decision but cannot be changed?|
|Priorities||What did the actors decide is the best way to achieve the goals given the constraints?|
|Functions||What is done in general terms?|
|Processes||What actual activities take place?|
|Resources||What is being used to perform the activities?|
The relationships among the levels of abstraction are also means-ends relationships because each level explains how the level above is taking place and the reasons for the level below to take place. For instance, referring to Table 1, “Processes” are performed through the use of “Resources” (the means) and for the purpose of carrying out the “Functions” (the end). Using this tool provides not only a comprehensive understanding of the work domain/task/decision, but also an explanation why activities are carried out in the way they are.
This Special Section
The articles in this special section bring together examples of ways CWA could be applied. Patricia Katapol and Ammy Phuwanartnurak describe two cases in which CWA was employed to study organizational culture and information sharing, while Kari Holland discusses how it guided her approach in a study about searching and cataloging publications of conferences, organizations and other corporate bodies. Finally, CWA can also be useful in developing conceptual constructs, and Jens-Erik Mai illustrates how it might be applied in this mode to the design of controlled vocabularies.
Improving the Web
The idea that information systems be customized for defined groups of users might seem unrealistic in the information-intense World Wide Web environment we know today. It does deserve consideration, however, because it is already taking place.
Electronic commerce on the Internet, for instance, is facilitated by a host of customized information systems whether for tangible items, travel or consulting. Many websites support access to information for particular groups of people. Most of them, however, employ a general search engine without tailoring its functionality and interface to the needs of their searchers. At the same time many general search engines are crawling the Web. While each is supposed to be unique, the differences among these engines do not contribute directly to the effectiveness of searching, and in essence they provide the same service.
It makes sense to customize some of these engines for particular audiences. There could be a search engine for high school students, another for lawyers, another for design engineers, one for high-level managers and so on. While such an approach will require coordination and collaboration among vendors, it has the potential to improve access to information on the Web in a meaningful way.
For Further Reading
Fidel, R., et al. (1999). A visit to the information mall: Web searching behavior of high school students. Journal of American Society of Information Science, 50, 24-37.
Fidel, R., Pejtersen, A.M., Cleal, B., & Bruce, H. (2004). A multidimensional approach to the study of human-information interaction: A case study of collaborative information retrieval. Journal of the American Society for Information Science, 55(11), 939-953.
Fidel, R., & Pejtersen, A.M. (2004). From information behavior research to the design of information systems: The Cognitive Work Analysis framework. Information Research: An International Electronic Journal, 10(1).
Pejtersen, A. M. (1985). Implications of users’ value perception for the design of a bibliographic retrieval system. In: J. C. Agrawal & P. Zunde (Eds.), Empirical Foundations of Information and Software Science. (pp. 23-37). New York: Plenum.
Pejtersen, A.M. (1992). The Book House. An icon based database system for fiction retrieval in public libraries. In Cronin, B. (Ed). The marketing of library and information services 2. (pp. 572-591). London, Aslib.
Rasmussen, J., Pejtersen, A.M., & Goodstein, L.P. (1994). Cognitive systems engineering. New York: Wiley.
Vicente, K. (1999). Cognitive Work Analysis: Toward safe, productive, and healthy computer-based work. Mahwah, NJ: Lawrence Erlbaum Associates
Articles in this Issue
An Ecological Approach to the Design of Information Systems