ABSTRACT
In chapter 2, we argued that it is important to conduct some form of work analysis, and
that such an analysis should be based on an ecological approach. As we pointed out in
chapter 1, various types of work analysis techniques have been developed since the
beginning of this century. In this chapter and the next, we critically examine some of
these techniques to see how well they stand up to the challenges imposed by complex
sociotechnical systems (see chap. 1). Normative approaches to work analysis, particularly
task analysis, are examined in this chapter. To anticipate, we conclude that some form of
task analysis is indispensible in complex sociotechnical systems, but that a work domain
THREE GENERATIONS OF WORK ANALYSIS TECHNIQUES
Because many different types of work analysis techniques have been proposed, it is useful to
categorize them in a way that highlights important similarities and significant differences
as well. Rasmussen (1997a) distinguished between three generic categories of models that can
be adopted for the specific purpose of grouping work analysis techniques. Norma-
tive models prescribe how a system should behave. In contrast, descriptive models
describe how a system actually behaves in practice. Finally, formative models specify the
requirements that must be satisfied so that the system could behave in a new, desired
way.2 Rasmussen used these three model categories to examine the evolution of
theories in a number of diverse areas of research, including engineering design, organi-
zational behavior, decision making, and human error. His review revealed a progression
over time in each of these areas, from normative first-generation models, to descriptive
second-generation models, and only very recently, to formative third-generation models.
