Scoring field dependence: A methodological analysis of five rod-and-frame scoring systems

Abstract

The most consistently used scoring system for the rod-and-frame task has been the total (or average) number of degrees in error from the true vertical, regardless of the initial or final directions of the rod and frame. Since a logical case can be made for at least four alternative scoring systems, a thorough comparison of all five systems seemed appropriate. Comparisons consisted of: (1) an internal consistency/reliability analysis, with split-half and test-retest reliabilities and a multitrait-multimethod matrix analysis of each scoring system, chair, frame, and man position; (2) a repeated measures ANOVA, with ethnic group, sex, and grade as between factors and chair, frame, and man positions as within factors; and (3) correlations of each scoring system with a selected set of external criteria. Results suggest strong support for use of the natural logarithm of the sum of absolute errors as the preferred scoring system, that concern with the confounding of field dependence and the E effect is largely unwarranted, and that all but one of the scoring systems perform adequately.