Dr. Levitt's thoughtful evaluation1 of our paper "Statistics, Science, and Psychiatry"2 was welcomed by us. It is evident that our sobering conclusions that statistical evaluations as they are generally applied to many psychiatric problems are unwarranted and misleading do require careful consideration.
He is correct in his statement that many of our arguments revolve around the meaning of the word "population" in research, and that if the population is large it is necessary to have recourse to a sample from which to draw inferences about the population itself. An inference concerning the population may not be 100% accurate, but statistical sampling theory will enable us to determine the probable amount of error attached to it.
So far, so good. The important point is, however, that the sample must represent the population from which it was drawn. Such samples are usually either random