Assigned Resampling Method: A new method to estimate size sexual dimorphism in samples of unknown sex

This paper proposes a new method, Assigned Resampling Method (ARM), to estimate the degree of size sexual dimorphism in samples of unknown sex. ARM resamples with replacement pairs of observation from the mixed-sex data and after applying a filter, generates a distribution of dimorphism estimates by transforming the pair values into ratios. The mean of the distribution is proposed as the ARM estimate of sexual dimorphism. Using 40 metric variables from comparative data sets of known sex (91 humans, 46 chimpanzees, and 56 gorillas), the ARM estimates were compared with the observed sexual dimorphism. Results show that the difference between the ARM estimates and the observed sexual dimorphism is within 5% for most of the variables examined. ARM is shown to perform reliably under various conditions of unequal sex ratios and small sample sizes. By directly comparing fragmentary materials without estimating body size, and by using an algorithm that does not rely on sex diagnosis of individual specimens, ARM addresses the challenges of studying sexual dimorphism with fossil samples.