Abstract

Two related experiments, each based on 60 college students, tested the idea that people can draw informational implications from a single piece of information (consensus information, distinctiveness information, or consistency information) and also make attributions regarding each behavior description they read. As expected, subjects experienced no difficulty drawing implications out of the information given and making attributions of the behavior. The information implications produced supported the predictions made by Orvis et al.(1975) in almost every detail. The fact that terminal information (the information which is implied by but does not imply other kinds of information) did produce attributions is inimical to Orvis et al.'s (1975) position which posits schemata and also to Kelley's (1967) cube theory of attribution. Attempts were made to find laws governing the observed information-implication relations and ways to make predictions without invoking the concept of schema. The notion of an attribution space was introduced, and it was argued that an attribution can be driven not only by confirming information but also by disconfirming information favoring a competing attribution.