In 1979-80, the Arkansas minimum competency tests were administered to a sample of 5,000 students in grades 3, 6, and 8. To determine how well test objectives matched the curriculum, their teachers estimated how many of the four items per objective a randomly selected student would answer correctly. Because chi square test comparisons of teacher estimates with actual responses were statistically questionable, a probability model was applied instead. The probability of obtaining 0, 1-2, or 3-4 correct answers by chance alone for the four items per objective was calculated, to obtain a validity coefficient. Results were presented by a 3 by 3 contingency table. Two applications of the model yielded validity coefficients of .78 and .98. An objective was considered invalid only when a large number of teachers predicted 0-1 correct and many students achieved at that level. (CP)