Social scientists have long recognized the possibility that research data may be affected by seemingly theoretically irrelevant contextual factors such as the ordering of response items in multiple-measure designs. There are many situations in research in which the ordering of measures or manipulations is theoretically uninteresting, but the results obtained may nevertheless be significantly affected by these contextual variations. Although an obvious concern in survey research (e.g., Fienberg & Tanur, 1989; Schuman & Presser, 1981; Sudman & Bradburn, 1982), inadvertent context effects can arise in most social and psychological research, including such diverse topic areas as personality testing (Knowles et al., chap. 15, this volume), lateralization of brain function (van Eys & McKeever, 1988), and countless others. One obvious motivation in seeking to understand the causes of context effects is the desire to control their influence or eliminate their occurrence. Indeed, the procedure of counterbalancing measures and manipulations is a staple of experimental design that is intended to control order effects. Although they have frequently been regarded as a nuisance factor in research, context effects can also be viewed as providing information about the processes involved in the generation of behavioral responses in experiments and surveys. From this perspective, they constitute a potentially important source of data about human thought processes (cf. Loftus, Feinberg, & Tanur, 1985). Understanding the psychological underpinnings of context effects thus becomes more than just a methodological issue.

In this chapter, I explore one approach to understanding context effects in psychological research, based on the concept of generic knowledge structures, a commonly invoked theoretical construct in cognitive and social psychology. Generic knowledge structures (GKSS) are organized sets of beliefs about the social environment that summarize, in a general (abstract) and functional way, previous direct and vicarious experience with the stimuli encountered in this environment. These knowledge structures reside in long-term memory and are thought to be organized by stimulus domain. For any given topic, the GKS includes a specification of the elements or basic attributes of the stimulus domain and their interrelationships (for comprehensive reviews, see W. F. Brewer & Nakamura, 1984; S. T. Fiske & Taylor, 1990). In essence, GKSs are the means whereby the social perceiver constructs the meaning of current experience. They provide a basis for making sense of the world by providing information about the elements of the environment and their interrelationships, as well as their significance for the perceiver. GKSs occupy a central position in most theories of social cognition and include such well-known theoretical entities as scripts (Abelson, 1981), schemas (Rumelhart, 1980), categories (Lingle, Altom, & Medin, 1984), and stereotypes (Hamilton & Trolier, 1986). Although the content and structure of these types of mental representation have been a matter of heated debate (see, e.g., E. E. Smith, 1988; Solso, 1989), there is abundant evidence suggesting that the idea of abstract knowledge structures is a useful theoretical tool for understanding human information processing, especially information acquisition and retrieval (see Alba & Hasher, 1983, and S. T. Fiske & Linville, 1980, for respectively pessimistic and optimistic analyses of the usefulness of the GKS approach).