Abstract
The Million Clinical Multiaxial Inventory (MCMI) utilizes a relatively large number of shared items among its 20 scales in maximizing external validity and minimizing the length of the test. This study evaluates the impact of this practice upon assessing the internal-structural validity of the MCMI. Analysis discloses that the relationship among the correlations between scales after extracting shared-item effects is not significantly different from the relationship prior to extraction of these effects. However, because the high percentage of shared items among scales does restrict the minimum magnitude of correlation coefficients among all pairs of scales, correlation coefficients obtained in cross-validation studies are not sufficiently free to vary from correlations obtained with the construction sample. Consequently, it is recommended that as a matter of practice cross-validation of inventories utilizing shared items also make comparisons between correlation coefficients after the effect of shared items has been extracted. A method for doing this task is reported herein.
Get full access to this article
View all access options for this article.