Concomitant Variable

Experimental Control

Controlling the effects of concomitant variables is generally desirable. In addition to random assignment of subjects to experimental conditions, methods can be applied to control these variables in the design phase of a study. One approach is to use a small number of concomitant variables as the inclusion criteria for selecting subjects to participate in the study (e.g., only eighth graders whose parents have at least a high school education). A second approach is to match subjects on a small number of concomitant variables and then randomly assign each matched subject to one of the treatment conditions. This requires that the concomitant variables are available prior to the formation of the treatment groups. Blocking, or stratification, as it is sometimes referred to, is another method of controlling concomitant variables in the design stage of a study. The basic premise is that subjects are sorted into relatively homogeneous blocks on the basis of levels of one or two concomitant variables. The experimental conditions are subsequently randomized within each stratum. Exerting experimental control through case selection, matching, and blocking can reduce experimental error, often resulting in improved statistical power to detect differences among the treatment groups. As an exclusive design strategy, the usefulness of any one of these three methods to control the effects of concomitant variables is limited, however. It is necessary to recognize that countless covariates, in addition to those used to block or match subjects, may be affecting the dependent variable and thus posing potential threats to drawing appropriate inferences regarding treatment veracity. In contrast, randomization to experimental conditions ensures that any idiosyncratic differences among the groups are not systematic at the outset of the experiment. Random assignment does not guarantee that the groups are equivalent but rather that any observed differences are due only to chance.

...