THE PURPOSE OF THIS RESEARCH WAS TO STUDY THE EFFECT OF ERROR OF MEASUREMENT UPON THE POWER OF STATISTICAL TESTS. ATTENTION WAS FOCUSED ON THE F-TEST OF THE SINGLE FACTOR ANALYSIS OF VARIANCE. FORMULAS WERE DERIVED TO SHOW THE RELATIONSHIP BETWEEN THE NONCENTRALITY PARAMETERS FOR ANALYSES USING TRUE SCORES AND THOSE USING OBSERVED SCORES. THE EFFECT OF THE RELIABILITY OF THE MEASUREMENT AND THE SAMPLE SIZE WERE THUS DEMONSTRATED. THE ASSUMPTIONS OF CLASSICAL TEST THEORY WERE USED TO DEVELOP FORMULAS RELATING TEST LENGTH TO THE NONCENTRALITY PARAMETERS. THREE METHODS OF ESTIMATING POWER FOR DIFFERENT CONDITIONS OF SAMPLE SIZE AND TEST LENGTH WERE STUDIED. THE COST OF AN EXPERIMENT WAS ANALYZED IN TERMS OF A FIXED COST PER SUBJECT AND A VARIABLE COST DEPENDENT UPON TEST LENGTH. COMPUTER PROGRAMS WERE WRITTEN TO USE THE LEAST SQUARES APPROXIMATION AND THE APPROXIMATION BASED ON PATNAIK TO ESTIMATE THE POWER UNDER ALL PERMISSIBLE ALLOCATIONS OF RESOURCES TO SAMPLE SIZE AND TEST LENGTH. THE PROGRAM RESULTS INDICATE WHICH OF THE PERMISSIBLE ALLOCATIONS WILL RESULT IN MAXIMUM POWER. TO DEMONSTRATE EMPIRICALLY THE EFFECT OF ERROR OF MEASUREMENT ON THE POWER OF STATISTICAL TESTS, SAMPLES OF PERSONS AND ITEMS WERE RANDOMLY DRAWN FROM A LARGE POOL OF DATA. TESTS OF 10, 20, AND 40 RANDOMLY DRAWN ITEMS WERE SCORED FOR SAMPLES WITH FOUR AND EIGHT PERSONS PER GROUP. THE EXPECTED TRENDS WERE PRESENT BUT NOT DEFINITIVE. (AUTHOR)