ABSTRACT This study investigates biases in sex estimation using the DSP (Diagnose Sexuelle Probabiliste—Probabilistic Sex Diagnosis) method in bioarchaeological contexts. The aim is to assess whether this method overperforms in estimating one of the sexes and to explore the implications for palaeodemographic interpretations. A meta‐analysis was conducted using data from 32 archaeological and current populations ( n = 2303), spanning prehistoric to modern times. DSP‐based sex estimations were compared with results from medical analytical, genetic, and alternative osteological methods. In the DSP reference data, morphological variability was analyzed through principal component analysis of pelvic measurements, and classification methods were evaluated using both linear discriminant analysis and random forest modeling. Findings reveal a consistent overestimation of female representation in sex ratios when using DSP, particularly in prehistoric and contemporary populations. This bias is linked to population‐level differences in sexual dimorphism, especially in the IIMT and SPU pelvic dimensions. The proportion of indeterminate individuals—most often male—varied by population, with European samples showing more male indeterminates and North American samples showing more female indeterminates. The study confirms a systematic sex classification bias within DSP applications, driven by both statistical and morphological factors. These distortions can mislead interpretations of population structures, funerary practices, and mortality patterns. Corrective approaches, such as redistributing indeterminate individuals or calculating confidence intervals for sex ratios, are proposed. Further research into population‐specific dimorphism and complementary methods is recommended to improve reliability in sex estimation for population‐based studies.
Bédécarrats et al. (Sun,) studied this question.