Interpreting the color effect of melanin on cocaine and benzoylecgonine assays for hair analysis: brown and black samples compared.

This paper examines the hypothesis that cocaine and BE assays of hair demonstrate a putative bias effect for darker color hair samples. Although such an effect has been reported in the literature, no one has examined this claim outside the bounds of simple significance of mean difference. In this paper a large number of black and brown hair samples are compared for cocaine and BE concentration values, and this comparison is evaluated for both significance and effect size. Two innovative measures are used to assess this relationship - a calculation of effect size using Cohen's d, and the use of an ROC curve to evaluate the potential for a dark color bias. The paper reports mixed results for significance, but consistent results for effect size. There does not appear to be any significant effect for cocaine. While BE demonstrates a significant mean difference, both the effect size and the ROC analysis show the effect to be trivial.