Inferential statistics was unknown until the twentieth century:

Student’s “t” test for comparing mean scores from two groups appeared in 1908. Analysis of variance tests were devised in the 1920s (Smith et al., 2000) but did not become a common part of psychological research designs until the 1930s (Rucci & Tweney, 1980). With the publication of his Experimental Psychology (1938), R. S. Woodworth “introduced a clear distinction between experimental and correlational research” (Winston, 1990). The critical distinction made between the two kinds of research was that only in experimental work could the cause of behavior be determined by manipulation of an independent variable; the definition “provided one powerful rationale for the animal research of the thirties, forties, and fifties” (Winston, 1990) because manipulations of “causal” variables in animal research provided fewer ethical or practical problems than research with humans. The search for causes of behavior and the theoretical models of learning embodied this definition of the psychological experiment as the means of testing hypotheses. This model of the experiment helped to establish prescriptions for the use of t-tests and analyses of variance as the statistical treatments of choice for the results of experiments, while correlational techniques and regression analyses were utilized by those interested in
individual differences. The methodology of research and standards for analyzing and reporting results of experiments in keeping with psychology’s status as a science is reflected in the standardization of the reports of experiments and the definition of the experiment. The emphasis on hypothesis testing and statistical analyses of comparisons between control and experimental group performance that later came to dominate experimental design and instructions to authors preparing manuscripts reflected the success of Woodworth’s definition of what constituted an experiment in psychology.