A Comparison of Multi-Attribute Generalizability Strategies

Abstract

Two generalizability theory strategies are appropriate for analyzing multiple attributes: the first is an integrated design that includes the attributes as a facet; the other involves a strategy of analyzing variance and covariance components on "specific" attributes. Sums of squares, mean squares, and variance components of the integrated design are shown to be simple functions of their specific design counter parts. Of particular interest is a relationship showing that integrated design variance components not involving attributes are identical with the average respective covariance components across attributes. An illustration is provided to demonstrate the relationships, strengths, and weaknesses of each strategy.

Get full access to this article

View all access options for this article.

References

Bartko, J. J. On various intraclass correlation reliability coefficients. Psychological Bulletin, 83, 1976, 762-765.

Cronback, L. J. , Gleser, G. C. , Nanda, H. , and Rajaratnam, N. The Dependability of Behavioral Measurements: Theory of Generalizability for Scores and Profiles. New York: Wiley, 1972 .

Farrell, A. D. , Mariotto, M. J. , Conger, A. J. , Curran, J. P. , and Wallander, J. L. Self- and judges' ratings of heterosexual-social anxiety and skill: a generalizability study. Journal of Consulting and Clinical Psychology, 1979, 47, 164-175.

Wessberg, H. W. , Mariotto, M. J. , Conger, A. J. , Farrell, A. D. , and Conger, J. C. The ecological validity of role plays for assessing heterosocial anxiety and skill of male college students. Journal of Consulting and Clinical Psychology, 1979, 47, 525-535.