Optimum Weighting to Entrance Subtests and Their Items to Make Composite Score

This research has been accomplished to weight national subtests and their items to make composite score. The aim of the project was to reduce measurement error associated with composite score in classical test theory framework. weighting procedure has been done in three levels; first, 30 multiple-choice-item test in differential calculus course with 3409 sample size was weighted at item choice level (choices popularity percent and formula score), and item level (simple total without weight or item effect weight, item factor weight and item difficulty weight). At subtest level, also, a test battery of medical assistance national test with 5 equally sized subtests, 6 items, which administered on 3572 candidates, has been weighted in different ways (average Pearson product-moment correlation coefficients weights, factor weights and regression coefficients). Another test battery of medical assistance national test with 5 unequal subtest length, 45, 26, 24,6, and 6 items which administered on 3638 candidates were studied without weighting just on their effective weights. This research revealed that formula score method produces more error than other procedures. Only weighting by item difficulty could rearrange examinees ranking in favor of qualified examinees. Other weighting methods are not satisfactory to enhance reliability coefficients and reliability coefficients, ab initio, are affected by appropriate items and well-made subtests with optimum length.