Confirmation bias and the testing of hypotheses about other people
Critical reviews of the literatures on self-fulfilling prophecies and self-confirming hypotheses uncovered several weaknesses in key works on those topics. In particular two important flaws were revealed. Hypotheses and expectations were confused and confounded and the most important aspect of these effects in person perception, changes in the perceiver's representation of the target, were ignored. Instead these works either made inferences about the perceivers' judgments from other individuals with different perspectives, or claimed to have demonstrated the effect of manipulating the hypothesis whereas their results were probably attributable to manipulating expectancies instead. It was argued that both of these types of inferences are invalid, and reanalyses of data from empirical works showed that the claims were not justified. A series of experiments was conducted in an attempt to find unequivocal evidence of self-confirming hypotheses. Numerous reasons were found as to why the phenomenon was highly unlikely to occur in social interaction. For instance, the asking of biased questions was found not to occur when perceivers generated their own questions to ask instead of selecting from a list given to them. In addition, subjects modified the questions they asked during the course of social interactions in such a way as to eliminate any possible bias in information search. Even when questions searching for confirmatory evidence were asked there was little evidence that interviewers' judgements were biased in favour of confirming their hypotheses. By contrast strong evidence was found for self-confirming hypotheses when subjects used information from their own memories to test hypotheses about aquaintances. These findings were discussed in the light of other paradigms within social psychology. Reasons why social cognition has, at times, so underestimated human rationality were considered and several conclusions were made including the need for greater caution in attempting to emulate and understand social processes in a laboratory setting.