Who Needs Data? I’ve Got Experience!

1 Department of Anthropology, University Tennessee, Knoxville, USA. *Correspondence to: Dawnie Wolfe Steadman, 1621 Cumberland Avenue, Strong Hall Room #502A, TN 37996 E-mail: [email protected] KEY WORDS: forensic anthropology, quantitative methods, medicolegal casework, human identification, skeletal analysis. Human Biology, Winter 2018, v. 90, no. 1. doi: 10.13110/humanbiology.90.1.05. Copyright © 2018 Wayne State Press, Detroit, Michigan 48201 invited commentary Who Needs Data? I’ve Got Experience! Steadman1 * In 2009, the National Academy Sciences released a blistering report on current state science oversight and practice in United States. The skewered scientifijic integrity standards number subdisciplines, including fijingerprints , blood spatter, bite marks, ballistics, offfered specifijic recommendations to improve forensicscience.Anthropologywasnotspecifijically mentioned report, but same criticisms leveled at other disciplines, poor training little attention standards, abject lack errors associated with were obviously relevant anthropology. To credit discipline, anthropologists have produced scores articles that focus analyses (e.g., Algee-Hewitt 2016, 2017; Hefner et al. 2014; Kooi Fairgrieve 2013; Megyesi 2005; Slice 2015; Stoyanova 2015, 2017), validation studies Jooste 2016; Kenyhercz 2017a, 2017b; Kim Milner Boldsen 2012; Savall Suckling 2016), cognitive bias our methods (Nakhaeizadeh 2014a, 2014b, 2018). rise “computational anthropology ,”theuseof advancedcomputingtoproduce models, simulations, predictive modeling address complex problems offfers unlimited opportunities for . technological explosion computing big-data analysis tools has formed an important space demography, ethnography, past present migration studies, yet remained largely unchanged by these advancements Despite warnings recent computational effforts fijield, those who are most likely continue rely traditional techniques personal experience rather than approaches. Here I explore some historical structural reasons behind slow acceptance new based large (and/or simulated) data sets approaches highlighted special issues Biology offfer guidance moving forward. Forensic is application principles biology questions, such as identifijication unknown remainsandinterpretationof traumatothebones. called into challengingforensiccontextsinthattheyworkwith humanremainsthataredecomposed,fragmentary, burnt,cremated,incomplete,andotherwisevisibly unidentifijiable. Unlike fijields single set tasks fijingerprints, DNA, ballistics), can be asked complete four disparate case: search recovery remains, identifijication, trauma analysis, estimation postmortem interval. Other 000 ■ Steadman specialized may include facial reproduction techniquestogarnerleadsfromthepublicconcerning identity remains isotope assess geographic origins history also assist identifijication. These distinct responsibilities preclude ability develop best practices lead crowded playing fijield methods. For example, task begins assessing biological profijile—estimationsof sex,ancestry,age,andstature from skeleton, each which consist multiple While stature estimates regression formulas derived various sized reference samples, historically qualitative observations morphological traits. Comfort levels developed 1980s 1990s exclusively macroscopic age-, sex- or populationbasedvariation (calledhere“traditionalmethods”) seem inclusion use sample sizes (including simulated machine learning), three-dimensional imaging techniques, sophisticated statistical analyses. instance, transition maximumlikelihood approach bioarchaeological samples using three indicators choiceof priors,providinganoutputof amaximum likelihood estimate 95% confijidence interval...