SOURCES OF VARIANCE IN BEHAVIORAL MEASURES OF HONESTY IN TEMPTATION SITUATIONS - METHODOLOGICAL ANALYSES

Reexamined the issue of generality vs. specificity of honesty and compared several alternative methodological approaches to the problem. Both multivariate correlational methods and analyses of variance were applied to assess the extent of intraindividual consistency of 106 6th graders across 6 temptation situations. Correlations among the measures were low to moderate, and the first principal components for the correlation matrices accounted for about 35-40% of the total variance. Mixed- and random-effects analyses of variance indicated that about 15-26% of the total behavioral variance was due to persons, 13-14% was due to tasks, and 60-71% was confounded between error and interaction sources. The results with the different statistical procedures varied in some instances according to different scaling procedures. The correlational and factor-analytic results replicated in substance the findings of earlier studies, which indicated that temptation behavior is only moderately consistent across a variety of tasks. In addition, the factor-analytic and analysis of variance methods were compared on the basis of underlying assumptions, relative advantages and disadvantages, and the potential of each method for offering ancillary information relevant to the generality-specificity issue. (22 ref.) (PsycINFO Database Record (c) 2006 APA, all rights reserved). © 1969 American Psychological Association.