In lieu of an abstract, here is a brief excerpt of the content:

  • Who Needs Data? I've Got Experience!
  • Dawnie Wolfe Steadman
key words

Forensic Anthropology, Quantitative Methods, Medicolegal Casework, Human Identification, Skeletal Analysis

In 2009, the National Academy of Sciences released a blistering report on the current state of forensic science oversight and practice in the United States. The report skewered the current scientific integrity and standards of a number of forensic subdisciplines, including fingerprints, blood spatter, bite marks, and ballistics, and offfered specific recommendations to improve forensic science. Anthropology was not specifically mentioned in the report, but the same criticisms leveled at the other disciplines, including poor training in quantitative methods, little attention to standards, and the abject lack of errors associated with methods, were obviously relevant to forensic anthropology. To the credit of the discipline, anthropologists have produced scores of articles that focus on quantitative analyses (e.g., Algee-Hewitt 2016, 2017; Hefner et al. 2014; Kooi and Fairgrieve 2013; Megyesi et al. 2005; Slice and Algee-Hewitt 2015; Stoyanova et al. 2015, 2017), validation studies (e.g., Jooste et al. 2016; Kenyhercz et al. 2017a, 2017b; Kim 2016; Milner and Boldsen 2012; Savall et al. 2016; Suckling et al. 2016), and cognitive bias in our methods (Nakhaeizadeh et al. 2014a, 2014b, 2018). The rise of "computational anthropology," the use of advanced computing to produce models, simulations, and predictive modeling to address complex problems in anthropology, offfers unlimited opportunities for forensic anthropology. The technological explosion of quantitative computing and big-data analysis tools has formed an important space in anthropology, including demography, ethnography, and past and present migration studies, yet forensic anthropology has remained largely unchanged by these advancements. Despite the warnings of the National Academy of Sciences and recent computational effforts in the field, those who are most likely to practice forensic anthropology continue to rely on traditional techniques and personal experience rather than quantitative approaches. Here I explore some of the historical and structural reasons behind the slow acceptance of new methods based on large (and/or simulated) data sets and computational approaches highlighted in these special issues of Human Biology and offfer some guidance for moving forward.

Forensic anthropology is the application of the principles of skeletal biology to medicolegal questions, such as the identification of unknown remains and interpretation of trauma to the bones. Forensic anthropologists are called into the most challenging forensic contexts in that they work with human remains that are decomposed, fragmentary, burnt, cremated, incomplete, and otherwise visibly unidentifiable. Unlike other forensic fields that focus on a single set of techniques and tasks (e.g., fingerprints, DNA, ballistics), forensic anthropologists can be asked to complete four disparate tasks in a single case: search and recovery of human remains, personal identification, trauma analysis, and estimation of the postmortem interval. Other [End Page 77] specialized tasks may include facial reproduction techniques to garner leads from the public concerning the identity of unknown remains and isotope analysis of human remains to assess geographic origins and migration history that may also assist with identification. These distinct responsibilities preclude the ability to develop a single set of best practices and lead to a crowded playing field of methods. For example, the task of personal identification begins with assessing the biological profile—estimations of sex, ancestry, age, and stature from the skeleton, each of which may consist of multiple methods. While stature estimates rely on regression formulas derived from various sized reference samples, most of the other methods are historically based on qualitative observations of skeletal morphological traits.

Comfort levels with methods developed in the 1980s and 1990s based exclusively on macroscopic morphological age-, sex- or population-based variation (called here "traditional methods") seem to preclude the inclusion of important new techniques that use large sample sizes (including simulated data and machine learning), quantitative data such as three-dimensional imaging techniques, and sophisticated statistical analyses. For instance, transition analysis is a maximum-likelihood approach developed for bioarchaeological samples using three skeletal indicators and choice of priors, providing an output of a maximum likelihood estimate and 95% confidence interval for each indicator, as well as for the combined evidence (Milner and Boldsen 2012), whereas Slice and Algee-Hewitt (2015) and Stoyanova et al. (2015, 2017) apply digital methods to...

pdf