Scores on standardized tests may be expressed in a variety of different modes. Raw scores, grade equivalents, percentile ranks, stanines, and standard scores are the major types in use today. The study investigated whether the mode or scale used to express scores affected the magnitude of group statistics, specifically, t and r (the t-test of the significance of differences between means and inter-test correlations). Raw score data were generated randomly, then translated into grade equivalents, percentile ranks, stanines, and standard scores using norm tables from Metropolitan Achievement Tests. This procedure was replicated five times for three levels of t and r, for three sample sizes, for two tests. Conclusions were drawn regarding when the use of a converted score yielded t or r values that were noticeably different from t or r values obtained when using raw scores. (Author)