Some Considerations in the Use of Half-List Speech Discrimination Tests
Patti Grubb · Journal of Speech and Hearing Research · 1963
Twenty-five word speech discrimination tests are discussed with respect to part-whole correlations, the interpretation of correlation coefficients, and the concept of validity. The assumptions underlying the interpretation of reliability coefficients are discussed. The conclusions are that the reliability of 25-word tests has not been demonstrated and that the 50-word list is being used as a sort of criterion but is a fallible measure subject to the same kind of error as the half-list.