The purpose of this research was to develop and test a method of identifying groups of students participating in a mandated statewide norm-referenced testing program who have undergone highly specific training in preparation for, or coaching during, test administration. The implementation of the analysis procedure was intended not only to identify such groups but also to discourage school staff from engaging in student test preparation. A Chi-square based analysis was developed which compared classroom, school, and district test score distributions against empirically derived "expected" distributions anchored to the district median score for a given subtest. The statistical model for generating the expected distributions was built using the data from the 1983 administration. The fit between the expected distributions created by the model and a sample of empirical observed distributions was exceptionally close. Significance levels were set, so that only groups with highly discrepant score distributions would be identified. The analysis performed in 1984 was replicated without modification in 1985. Results showed that since the analysis and related notification procedures were implemented, there has been some reduction in the number of groups exhibiting score distributions that may be associated with improper test preparation. (Author/JAZ)