Critical Difference Table for Word Recognition Testing Derived Using Computer Simulation

Abstract

Purpose To construct a table for upper and lower limits of the 95% critical range for changes in word recognition scores obtained with monosyllabic word lists (of lengths 10, 25, 50, and 100 words) using newly available methods. Although such a table has been available for nearly 30 years (A. R. Thornton & M. J. M. Raffin, 1978), the earlier table was constructed by calculation and used an approximation to the variance of the difference score between 2 administrations of word lists of identical size. It has been used clinically, reproduced, and recommended for use by clinicians in handbooks and textbooks. Method The new table was created using computer simulation of the relevant distributions and a direct estimate of the variance of the difference score between 2 tests, calculated using the simulated results. Results The new table differed from the previous table in 23% of entries. Critical ranges were both narrowed (82%) and expanded (18%). No range changed by more than 1 word correct in any direction. The original table was most accurate for list sizes of 25 words each. Conclusion Using the new table will provide more accurate estimates of the 95% critical range for successive administrations of word recognition tests.