The effects of the z-score transformation on measures of relative erectile response strength: a re-appraisal

In studies of stimulus control of sexual arousal, some researchers transform raw data to z-scores. Earls, Quinsey and Castonguay [Archives of Sexual Behavior, 16, 493-500 (1987)] have argued that the proportion of variance due to stimulus presentations was greater in z-scores than in the raw data or in a percent score. We present analyses of sexual arousal data that show that z-score transformations may distort the information inherent in the raw data and may increase random error. In addition, we present Monte Carlo analyses indicating that the z-score transformation compromises the estimates of Type I error, and, depending on specific circumstances, sometimes increases and sometimes decreases the power of the statistical test. Transformation of raw data to percent of full erection does not distort the information in the raw data, does not compromise the data analyses and, therefore, is the preferred transformation.