The Null Hypothesis in Rock's One-Trial Learning Experiments

Abstract

Wilson and Miller argue that good experimental design requires support of the experimental hypothesis by rejection of the null hypothesis. Their only examples of designs that support the experimental hypothesis by acceptance of the null hypothesis are the experiments of Rock and his colleagues in paired-associate learning. It is here argued that the provisional experimental hypothesis of Rock, et al. was incremental. Rejection of the null hypothesis would have led to support of the provisional experimental hypothesis. Thus, their design, despite surface appearances, was one of rejection-support, but their results forced them to accept the null hypothesis, which was unfavorable to incremental learning.