The major purpose of this study is to further the development of procedures which minimize current limitations of creativity instruments, thus yielding a reliable and functional means for assessing creativity. Computerized content analysis and multiple regression are employed to simulate the creativity ratings of trained judges. The computerized scoring procedure is evaluated in a cross-validation sample. The methodological problems of establishing a reliable criterion and generating parsimonious forced prediction models through predictor stability analysis are emphasized, and possible solutions are explored. Support for the proposed solutions and empirical validation are incorporated in the study. The forced model results are regarded as tentative and should be tested on a new sample. Bibliography and statistical data are included. (Author/AE)