r rHE SUBJECT of this article is bias in a science, mainly as caused by method, but also including such others as those caused by the social circumstances under which research is carried out. Ways in which some of these effects rnight be rectified will be considered in a later article. The science I shall be considering is nonmedical psychology, that is, the psychology that is taught in the universities and which uses as its main methods expersments and tests, and which therefore includes clinical psychology. I believe that every subject has certain methodological biases and is organized in a certain way, which gives a characteristic stamp to the knowledge it purveys Therefore my examination of psychology is intended to have a general significance. I choose the subject because it is the one I know most about and because of the practical importance to me of clanfying my ideas about method. Psychology has been institutionalized and psychologists form a group, which I shall contrast with the community as a whole, and with another larger group, namely that of all scientists. I would distinguish Science fi"om Scholarship, perhaps, and say that for psychologists, as for many linpsts, and for some sociologists and political scientists, Science forms a reference group. This cannot be said of all who call themselves 'scientists'. Obviously it doesn't apply to a moral scientist, nor necessarily to one who calls himselfa social scientist; and it doesn't seem to me to apply to the conscious devotees of historical science. The picture of the reference group of Science that is widely held in psychology is a stock one, in which the practices of the most successful scieniists, that is the physical scientists, play a predominant part, as do also the ideas of the philosophical expounders of scientific method. I refer to it as a picture of a reference group to emphasize that it is different from the reality. There is, I think, much seleciion from, and simplification of, what really obtains, and there is sometimes evidence to suggest that the extremer identifiers go further than their models, particularly, for instance, in their atiitude to introspection. The consequence of this