A useful device for the systematic definition of a theoretical construct is Guttman's facet design (Guttman, 1954). Facet design defines a universe of observations by classifying them using a scheme of facets with elements subsumed within facets. Facets are different ways of classifying observations; the elements are distinct classes within each facet. The universe of observations is classified using three kinds of criteria: (a) the population facets that classify the population, (b) the content facets that classify the variables, and (c) the common range of response categories for the variables. The facet design approach can be expressed graphically as follows:
In this representation, [X] is the population of objects (respondents, research participants), [A], [B]... [N] are content facets, and R is the common response range. Roskam (1990) emphasized the importance of the response range because it defines the domain of observations. Thus, if the range is defined as "correct/wrong by an objective criterion," we are investigating intelligence behavior, and if the range is defined as "ordered as very positive/very negative toward that object," we are investigating attitude behavior (Roskam, 1990, p. 189).
For our present goal, we concentrate on the facet structure of the variables. The various content facets can be viewed as a cross-classification, analogous to an analysis of variance design that specifies the similarities and dissimilarities among questionnaire items. Each facet represents a particular conceptual classification scheme that consists of a set of elements that define possible observations. The content facets must be appropriate for the construct that they define. In selecting the most appropriate facets and elements, the objective is to describe all important aspects of the content domain explicitly and unequivocally. For example, for many constructs, it may be useful to distinguish a behavior facet that defines the relevant behaviors and a situation facet that defines the situations in which the behaviors occur. An example of a facet design is Gough's (1985, p. 247) design for reasons for attending weight-reduction classes. Gough defined the person facet [X] as "married women attending slimming groups." There are two content facets: source and motive. The facet design can be summarized as: To what extent does the person [X] feel that Source [S] led her to believe that she would achieve Motive [M] if she lost weight, as rated using Response [R], The source facet [S] has four elements: (a) own experience, (b) husband, (c) doctor, and (d) media. The motivation facet [M] has seven facets: (a) feel healthier, (b) feel fitter, (c) be more physically attractive, (d) have fewer clothing prob lems, (e) suffer less social stigma, (f) be less anxious in social situations, and (g) feel less depressed. The response range [R] is defined on a 7-point scale ranging from 1 (not really at all) to 7 (very much indeed). In facet design, the facet structure is often verbalized by a mapping sentence, which describes the observations in one or more ordinary sentences. Figure 19.1 presents a mapping sentence for the reasons for attending weight-reduction classes.
In this facet design the first facet (source) refers to the source of the belief, and the second facet (reason) refers to a specific consequence of losing weight. A facet design such as the one described can be used to generate questionnaire items. The [X] facet points to a specific target population of individuals. The source facet has four elements, the reason facet has seven, which defines 4 x 7 = 28 questions. For example, combining the first elements of the source and reason facets leads to the survey question, "Did your own experience lead
The extent that person [x] feels that
Was this article helpful?