Abstract
Converting raw scores to T-scores simplifies the interpretation of test results by providing a standardized metric. A simple linear conversion of raw scores to T-scores can produce misleading T-scores if the raw score distribution deviates significantly from normality. The aim of the present research was to provide normalized T-scores and percentile rank scores for the Test MultiDimensionnel de Personnalité (TMDP). T-scores were established by first transforming raw scale scores to a normal distribution with either an approach based on the Item Response Theory or with an approach based on Rankit percentile-based normalization. The three resulting T-scores (Linear-, IRT- and Rankit-based T-scores) were compared. Results revealed that both normalization approaches yielded similar normalized T-scores, which deviated substantially from linear T-scores for most scales, especially in the higher range of the scores. Benefits of normalized T-scores and the need for gender- and age-based norms are discussed.