Categorical Versus Continuous Independent Variables
As knowledge of an area of study increases, research that uses categorical independent variables (e.g., group or personality types) is often replaced by research that employs continuous independent variables that represent the distinctions embodied in the categories.
For example, differences between the sexes in a given investigation may represent power differences, socialization differences, or physical differences; replacing sex—a dichotomous variable—with a continuous variable, such as empathy or achievement orientation, may help clarify the differences that are expected to be at work (see Wood & Eagly, 2002, for a more complete analysis). Similarly, differences between cultures that are treated as dichotomous variables, such as individualism-collectivism or masculinity-femininity, are more informative when treated as continuous (Cai, Wilson, & Drake, 2000; cf. Hofstede, 1980).During this transition from categorical to continuous independent variables, both types of variables are likely to be used so that a model that uses the continuous variable can be shown to not be improved by the addition of categorical variables (e.g., Oetzel, Ting-Toomey, Yokochi, Masumoto, & Takai, 2000). The study of conflict communication is at the point of this transition; analyses that use group both as a categorical variable and as one or more continuous variables can advance theory by showing that the continuous variables are able to capture the group differences successfully.
Moving from categorical to continuous independent variables is more than a measurement issue: It is a theory construction strategy. The methods used in a study reflect the conceptual definitions and the generality, abstractness, parsimony, and completeness of the theory used to investigate the phenomenon of interest. The concepts and theories in conflict communication research are the foundation on which investigations are constructed, and methods are useful only insofar as they ultimately provide tools to answer questions posed by sound theory. Furthermore, methods and theory are interdependent: Without some specific methods, some theoretical questions cannot be posed, and without some specific theory, the choice of methods is arbitrary, or, worse, irrelevant.
For example, to determine whether conflict styles formed a two-dimensional space as proposed by Pruitt and Rubin (1987), Cai and Fink (2002) used multidimensional scaling, and they found that the proposed conflict styles required a space of more than two dimensions: The measures tell us how to advance the theory. On the other hand, Durkheim’s (1951) theory of suicide proposed that there were four kinds of suicide: egoistic, altruistic, anomic, and fatalistic. Without this theoretical insight, examining overall suicide rates would reveal very little of the social mechanisms at work: Theory tells us what to measure.