Controls for Group Differences
We typically create statistical models that employ variables that are causally related and that correspond to our conflict structure and process. For example, think of a few variables that could be used in an SEM.
But variables also may be used to establish boundary conditions for the conflict system to be investigated or to control for unwanted or extraneous sources of variability. To conduct research comparing groups or cultures, we should measure and then attempt to control for these kinds of variables, which typically include socioeconomic factors such as education level, income relative to the per capita income of the nation the sample is from, occupational prestige, and the like (see, e.g., Massett, 1999). If these factors are not controlled, they may mask effects, making it difficult to know whether the empirical relationships among variables reflect the causal relations of interest.In cross-cultural investigations, controlling for these factors allows researchers to at least attempt to differentiate cross-cultural from within-culture factors, thereby improving the ability to make valid cross-cultural comparisons. If such controls are not employed, claims about culture are likely to be masking the influence of class, population density of one’s geographic area (urban vs. suburban vs. rural), education, and so on.
This methodological discussion has significant implications for theory development. If we are concerned with belief and behavior systems of a population (e.g., the norms for handling conflict in the workplace, in the family, or among friends; the sanctions for transgressors of these norms; and the conditions under which specific requests are appropriate), then many factors make a difference, but which of these factors do we classify as cultural? Because developmental effects (e.g., worldwide urbanization, increased participation in formal education) modify cultures, any differentiation between cultural and noncultural factors is a snapshot of one point in time. We try to get a handle on this differentiation by two methods: (1) using comparable samples, which are difficult to obtain and even more difficult to define, or (2) relying on statistical controls. The former involves using samples that may be unrepresentative of the larger population but similar in some characteristics to the other culture with which its members are compared. The latter is likely to “over control” by removing effects that are or will be tied to cultural differences. This conundrum requires not better methods but better theory to investigate processes and attribute effects to their appropriate sources.