CAN WE STUDY INTELLIGENCE USING THE EXPERIMENTAL METHOD

Cronbach (1957), at a famous presidential address to the American Psychological Association, argued that there were two disciplines in scientific psychology that ought to work together, but failed to do so: the experimental and the correlational. I have always argued, along the same lines, that correlational studies were concerned with taxonomy, but that causal arguments required an experimental approach (Eysenck, 1984). These admonitions have had little effect; traditional experimentalists still spurn the application of correlation-related concepts (intelligence, personality) to their studies, whereas correlationists make no effort to make their studies assume a more causal approach by using the experimental method. Boring (195 1) put up the traditional case: “Is a test an experiment? And does the history of mental testing belong in the history of experimental psychology? Not really, although there is no sharp logical line of demarcation” (p. 570). According to Boring, the experimental approach uses an independent variable and determines what function the variation of a dependent variable is of the variation of the independent variable. In the field of testing, the primary variable is a difference of persons, which occurs at random and is not an independent variable, nor can it usually be changed experimentally. Actually, the position of experimental psychologists is much less secure than at first appearance. Physicists can eliminate experimentally all factors other than the independent variable; psychologists cannot. Whatever the experiment, it involves persons, with different IQs, different personalities, different motivations, different attitudes, different emotions. As Gregory Kimble once told me (personal communication, April 7, 197 1): “Ninety percent of the outcome of a conditioning experiment is determined before the subject is strapped in.” That is, the subject’s attitude, knowledge, treatment by the receptionist, and emotions determine his or her reactions over and above the independent variable as manipulated by the experimenter. As an example, consider the experimental studies of Shigehisa and Symons (1978) and Shigehisa, Shigehisa, and Symons (1973) on the