Psychological experimentation rarely meets the assumption of random sampling that conventional statistical hypothesis-testing procedures are generally believed to require. A rational is presented to justify the current procedure of testing statistical hypothesis with nonrandom samples. Random assignment of Ss to treatments allows statistical inference apply only to the Ss actually used in the experiment. Randomisation tests can be used as exact tests for this purpose and parametric tests can be used as approximations to randomisation tests. Nonstatistical inferences about persons not used in an experiment can be made on the basis of logical considerations. (A) + Research on selective attention in humans is discussed in this paper. Emphasis is placed on an analysis of the nature of the selectivity that is demonstrated in 4 experimental tasks: recognition of tachistoscopically presented materials, listening to one of several simultaneous auditory messages, speeded classification of multidimensional objects, and searching through complex visual fields. The results of these studies are interpreted in terms of coding strategies that Ss use in processing sensory input. (A)
Abstract