A TECHNIQUE OF PATTERN ANALYSIS WHICH EMPHASIZES THE DEVELOPMENT OF MORE EFFECTIVE WAYS OF SCORING A GIVEN SET OF VARIABLES WAS FORMULATED. TO THE ORIGINAL VARIABLES WERE SUCCESSIVELY ADDED TWO, THREE, AND FOUR VARIABLE PATTERNS AND THE INCREASE IN PREDICTIVE EFFICIENCY ASSESSED. RANDOMLY SELECTED HIGH SCHOOL SENIORS WHO HAD PARTICIPATED IN THE WASHINGTON PRE-COLLEGE TESTING PROGRAM WERE ADMINISTERED A SECOND BATTERY THREE MONTHS LATER. CRITERION DATA WERE AVAILABLE ON MEN AND WOMEN WHO LATER ENTERED THE UNIVERSITY OF WASHINGTON. MEASUREMENTS INCLUDED TESTS OF ACADEMIC ACHIEVEMENT AND APTITUDE AND FIVE CRITERIA OF COLLEGIATE PERFORMANCE. THE PATTERNS INCREASED MULTIPLE CORRELATIONS IN THE WEIGHTING SAMPLE BUT CROSS VALIDATION SHRINKAGE WAS CONSIDERABLE EXCEPT WHERE THE ORIGINAL VARIABLES WERE MOST HEAVILY WEIGHTED. SUBPATTERN SCORING, IN COMMON WITH OTHER TECHNIQUES OF PATTERN ANALYSIS, MADE NO CONTRIBUTION TO PREDICTION OF ACADEMIC ACHIEVEMENT. THIS FAILURE IS SIGNIFICANT BECAUSE OF ATTENTION PAID TO THE SELECTION OF DIFFERENTIATED CRITERIA AND RELEVANT PREDICTORS. PATTERNS OF PERSONALITY NEEDS ASSOCIATED WITH ACADEMIC ACHIEVEMENT WERE ALSO STUDIED, BUT THE SAME LACK OF PREDICTIVE STABILITY WAS DEMONSTRATED. THERE WOULD SEEM TO BE SMALL ROOM FOR CONTINUING THE CONJECTURE THAT PATTERNS CAN GO ABOVE AND BEYOND PREDICTION FROM SIMPLE LINEAR FUNCTIONS OF ORIGINAL VARIABLES. (PR)