Beyond Compare: Investigating the Effects of Compositional Characteristics and Method Selection on the Results of Skeletal Analyses

Populations are routinely characterized and compared through their prevalence of skeletal lesions. To investigate the effects of both age structure and differential bone preservation on intergroup prevalence comparisons, counts of healed trauma were compared using a variety of statistical approaches, between individuals from two populations with different cultural contexts, age compositions and element preservations: the J.C.B. Grant collection (n=172) and the Luis Lopes collection (n=161). While the bone-years metric addressed the issue of preservation-confounding that skews the person-years metric, it erroneously assumes the stability of fracture risk over the life course. Comparison of age-stratified groups by way of the common odds ratio was found to be the most reliable method for the comparison of fracture data. Workarounds to small sample sizes (extrapolating to larger sample sizes or inserting dummy values into null fields) were both found to cause more problems than they solved. Comparing element aggregates (e.g. total long bones, total cranial bones) is another potential means to detect significant intergroup or intragroup differences, however due to the common cooccurrence of certain fractures and the difficulty in reliably assessing fracture timing, the use of these aggregates also risks introducing bias. Comparing the number of individuals with at least one fracture of interest was found to be the most resilient way of mitigating bias while having the necessary scope to detect different skeletal manifestations of an event of interest. The odds of an individual in the Grant collection having at least one nasal fracture were found to be 3 times higher than in the Lisbon collection, and this overall difference was driven by differences in those 55 years of age or older. The trauma results are consistent with the boxing-etiology hypothesis which posits that the increased visibility and ritualization of specific behaviour through sport (i.e. fighting with one’s fists in boxing and prizefighting, popular activities in British populations) validates that behaviour as socially acceptable, and increases its prevalence in non-sport venues such as physical assaults. To ensure the most robust and defensible interpretations possible, researchers comparing skeletal phenomena within or between groups must ensure that they consider both the preservation and the demographic makeup of their samples, and choose their comparative approach mindfully.Ph.D