The Rise and Fall of Social Psychology: The Use and Misuse of the Experimental Method (review)

Social scientists naturally assume that more scientific rigour is better than less. As a result, suggestions for improvement within disciplines move in predictable directions: longitudinal studies should replace cross-sectional designs, ever more complex multivariate techniques should replace dated statistical devices, and, of course, whenever possible, the most scientific of methods, the experiment, should take precedence over all alternatives.

In these terms, The Rise and Fall of Social Psychology is in part a case study, the lowest of the low. It challenges the contributions to knowledge of a discipline based on the experimental method, the highest of the high. A fair judge, it seems to me, would declare a TKO, and warn Goliath to stay away from impressionable students until his head has cleared.

Brannigan begins by summarizing the remarkable virtues of the true experiment. There is no hint of trouble in his account. Even generalizability, often said to be the Achilles heel of this design, is of little concern when the focus is on general processes present in all segments of a population (p. 3). He then turns his attention to the golden age of social psychology, the decades following World War II, when giants walked the earth and experimental methods were increasingly applied to intrapsychic phenomena, reflecting —or causing —a split in the field between its sociological and psychological practitioners. Then, in the 1990s, came decline. What caused it?

The quick answer is the institutional review board. Designed to protect human subjects from unethical treatment, these boards looked askance at the deception common to the social psychological experiment, and rejected the argument that advancement of knowledge required it.

How great was the loss? Where would we be had an established experimental science remained free of the shackles imposed by puritanical busybodies? Brannigan's answer is that the loss is minimal, that deception of a few students in the laboratory is as nothing compared to the deception of millions of students in the classroom.

Brannigan's assessment of the scientific accomplishments of social psychology is shared by many of its practitioners. Taken one a time, the field's most famous experiments turn out to be not experiments at all, but demonstrations or dramatizations ("happenings"), and its most influential concepts and "effects" fail to survive careful analysis or critical examination of the evidence alleged to support them. The casualties include Festinger's cognitive dissonance, Milgram's obedience to authority, Zimbardo's mock prison, Sherif's and Asch's conclusions about social influences on perception, the Hawthorne effect, Rosenthal's Pygmalion effect, and, more generally, the allegedly pernicious effects of pornography and TV violence.

The mystery is not so much how these findings have managed to survive, but how they have managed to prosper. Enshrined as they are in uncritical textbooks, pervasive educationist philosophy, and even in law, all of them remain enormously influential. Their appeal, Brannigan notes, stems in part from their moral subtext. Many can be seen to derive from "political aspirations" rather than, as it should be, the other way around. But the larger part of their appeal appears to stem from their grounding in experimentation. Who can quarrel with the method of Galileo? Augustine Brannigan manages to do so by putting two and two together. If a method consistently produces defective results, however glorious its attributes, something must be wrong with it. And, sure enough, experiments, as executed by social psychologists, turn out to have a fatal flaw. They are incapable of falsifying theories. Not to be outdone by rats, the truth machine somehow detects the expectations of the truth-seeker!

Brannigan briefly considers possible fixes. One possibility is to follow the lead of physics and divide the duties of theoreticians and experimentalists. The resulting "triple-blind" experiments would reduce the tendency to corroborate rather than falsify theories, increase the likelihood of rigorous replication, and encourage publication of negative findings (pp. 164–65).

Of the several inadequacies...