The preceding chapters drew a clear distinction between research classified as experimental and research that is correlational. In correlational research, the investigator's role is that of an observer. All variables of interest are permitted to vary freely in their natural context. In a real sense, all the variables in correlational studies are dependent variables. The researcher's job in these research contexts is to assess this variation and to tease out the patterns and interrelationships that exist among the critical measures. On the other hand, in experiments, the researcher actively intervenes in the normal pattern of covariation, systematically controlling variation in the independent variable (or variables) to assess its causal impact. For purposes of internally valid cause—effect analysis, controlled manipulation of the causal variable and random assignment of subjects to the manipulated conditions are the necessary hallmarks of true experiments.
In many research contexts, the distinction between experimental and correlational studies may not be all that clear-cut. For example, in our discussion of field experiments, we mentioned studies in which the researcher selects rather than creates the levels of the independent variable, or cases in which “random” assignment occurs naturally rather than by experimental intervention.1 Such studies preserve the logic of experimental design but lack the degree of experimenter control that characterizes “pure” experiments. By the same token, some correlational studies are conducted in the context of interventions into a given social situation (e.g., studies that investigate the reactions of an established group to the introduction of a new member), thus mixing aspects of experimental and correlational design. The distinction, then, between experimental and correlational research should be seen as a continuum rather than a strict dichotomy.
Somewhere between true experiments and pure correlational research studies are those research situations in which some systematic intervention in a social setting has been made for the purpose of assessing its causal effects, but the exposure of participants to this____________________