Abstract
Split-half reliability estimates previously reported for the Luria-Nebraska Neuropsychological Battery (LNNB) summary scales have been uniformly high, but their magnitude depends upon the method of item grouping for the half-tests compared. Coefficient alpha was employed in the current study as it gives the mean of all possible split-half values. Alpha reliability coefficients for the LNNB summary and localization scales remained consistently high in current analyses. Alpha reliability coefficients for the LNNB factor scales were quite variable, but most scales have adequate preliminary values for research work and clinical hypothesis confirmation. Four of six receptive speech factor scales were found to be unreliable. Reexamination of their underlying factor structure was suggested.