Abstract

Two experiments were conducted to test the idea that Kelley's(1967) three types of attributional information(consensus, distinctiveness, and consistency information) can be partially implied by one or the other. Sixty subjects received one type of information(Experiment 1), and another 60 subjects received two types of information(Experiment 2) and were asked to infer the remaining information by (1) choosing between the two opposite implications(e.g., high and low consensus implications), on each of the three dimensions, and (2) rating the strength of each implication on 7-point unipolor scales. As expected, subjects were able to fill in the missing information based on the given information. In addition, the pattern of implicational relationships found among consensus, distinctiveness, and consistency information indirectly supported the T-shape attribution space model proposed by Cha and Na(1986). but did not support the schema-based predictions made by Orvis, Cunningham, and Kelley(1975). It was also demonstrated that people would assume a higher baseline for consistency information than for consensus or distinctiveness information. The analysis suggested that the traditional internal-external attribution distinction may have confounded two different attribution choices, one between person and stimulus and another between person and circumstance.