Assessing fluctuating odontometric asymmetry among fossil hominin taxa through alternative measures of central tendency: Effect of outliers and directional components on reported results

Abstract

Preliminary inquires into the distribution and expression of fluctuating odontometric asymmetry (FOA), among selected fossil hominins, have revealed results that may be serviceable within studies that assess, among others, palaeobiological, evolutionary processes and events. Though several intricate statistical applications have aided in the advancement of FOA to the hominin fossil record, little is known regarding the influence of outliers and directional components on reported results. Moreover, most methods employed to test homogeneity among FOA datasets are sensitive to the assumption that underlying samples reflect Gaussian distributions. Because this assumption is often violated, alternative formulations of Levene's test statistic, which have been shown to be robust under non-normality, have been suggested. Unfortunately, previous FOA studies have failed to address their potential. Given this, we considered two areas that may influence interpretations of FOA among fossil hominin studies. Firstly, we assessed distributions of signed data ( d u) among samples of Australopithecus africanus, Paranthropus robustus and Homo habilis for outliers and directional asymmetry to evaluate their influence on reported heterogeneity. Secondly, in an attempt to decrease the probability of falsely rejecting H 0 due to non-normality, we considered alternative estimates of central tendency for comparisons of FOA. Our study confirms the need for intrinsic scrutiny of data, as the removal of one extreme value within the buccolingual H. habilis sample produced statistically significant outcomes at the sample level, while directional asymmetry was exposed within an expanded buccolingual P. robustus sample. However, though servicing alternative measures of central tendency remains informative, except for the buccolingual P. robustus sample before the correction of directional asymmetry, replacement of the mean was not required herein. Consistent with previous investigations, significant differences between buccolingual values in apposing arcades were unique among A. africanus and P. robustus, with the latter expressing greater FOA overall. Finally, our results strengthen the assertion that the individuals sampled among H. habilis may indicate an episode of developmental compromise where external and/or internal noises are lessened through internal homeostasis.