The present study suggests that although the reliability of an academic aptitude test given under formula-score condition can be increased substantially through empirical option weighting, much of the increase is due to the capitalization of the keying procedure on omitting tendencies which are reliable but not valid. (Author)