Date of Award


Document Type


Degree Name

Master of Arts (MA)




The present study investigated the naturalistic use of consis tency, distinctiveness and consensus information, which are the three information variables in Kelley's (1967) covariation model of attribu tion. Newspaper advice columns were used as a naturalistic source of social data.

The first question addressed in this study was whether or not Kelley's (1967) three information variables are ecologically valid. If perceivers in the "real world" actually use covariation information to support or defend their attributions, then it should be possible to reliably locate these pieces of information. If the information is used, it would be possible to examine whether perceivers prefer indi vidual items of information or combinations, as well as which combina tions are most commonly used. Student raters were trained in the use of coding manuals which helped standardize their advice column searches for attributions and the three information types.

The second question addressed in this study concerned per- ceiver's relative preferences for the three information types. A con troversy in the literature regarding the use of consensus information began with several findings that consensus information is underutilized (Kassin 1979a). Researchers have demonstrated that under certain laboratory conditions consensus will be utilized. Up to now, the question of the underutilization of consensus has not been approached naturalistically. For the present study, consensus information was defined as both implicit (e.g., norms, stereotypes) and explicit (e.g., base rates, observed covariation across actors). It was hypothesized that consensus may not be underutilized when such a broad definition is used.

The third area of concern to the present study was an attempt to demonstrate that consensus information, defined as both implicit and explicit, is associated with internal-external attribution in the way that previous theory and research would predict. High consensus should be associated with external attribution and low consensus should be associated with internal attribution.

The results of this study indicate that Kelley's (1967) three information variables may be ecologically valid because at least one of them was used with over half of all attributions examined. The informa tion types were used significantly more often alone than in combination form (p = 0.0004). No one combination was used more than any other. Consensus information was found to be significantly underutilized in terms of perceiver preference, for overall use (p = 0.0005) as well as for individual use (p = 0.0074). A marginally significant result (p = 0.0578) was obtained for the predicted relationship between level of consensus and the direction of attribution. Implicit and explicit forms of consensus contributed equally to the overall use of consensus information, but it was found that consensus was overwhelmingly used in high (e.g., much covariation across actors) as opposed to low (e.g., little covariation across actors) form (p = 0.000002). These results were discussed and various interpretations were offered.