Biology Reference
In-Depth Information
Over the last two decades, the scientific rigor of research methods within skeletal biology
(as applied to forensic anthropology) has been called into question by important legal proceed-
ings. As a result, there is now a need for more objectivity and standardization of approaches.
6
Expert witnesses giving court testimony in the United States must now demonstrate the
reliability and relevance of the scientific methods implemented in order to analyze evidence
according to the Federal Rules of Evidence (2001) and the ruling from the court case Daubert
vs. Merrill-Dow Pharmaceuticals (1993). This reliability is determined by error rates that
quantify observations and statistically report confidence levels (
Adams and Byrd, 2002;
Christensen, 2004
). For example, Christensen (2004:2) stated, “scientific knowledge [must]
be grounded in the methods and procedures of science” more than subjective opinion. This
knowledge must be empirically tested, falsifiable, and subject to peer review through publica-
tion (
Christensen, 2004
). Asmany forensic anthropologists are alsoworking in bioarchaeology,
this has raised the standards across both specializations and has served to sharpen the skills of
the researcher (
Ubelaker, 2000
). Most sex estimation research has since taken on a metric
approach using discriminant function analysis or other multivariate quantitative methods.
Discriminant Function Analysis
Discriminant function analysis (DFA) is a statistical procedure that classifies unknown
individuals and the probability of their classification into a certain group (such as sex or
ancestry group). Discriminant function analysis makes the assumption that the sample is
normally distributed for the trait. The posterior probability and typicality probability are
applied to calculate the classification probabilities (
Albanese et al., 2008
).
7
The posterior prob-
ability is the probability that an unknown case belongs to a certain group based on relative
Mahalanobis' distances measuring the distance to the center or centroid of each group. The
typicality probability is how likely the unknown case belongs to a group based on variability
within all groups. The discriminant function procedure has been programmed into most
standard statistical packages for greater applicability.
Not all skeletal measurements are equally effective for sex estimation using DFA and the
skill of the researcher plays an important role; practice and exposure to population variation
are still crucial.
Adams and Byrd (2002)
compared 13 different measurements taken by
68 researchers. They discovered high interobserver variability in all measurements by
researchers with less than 5 years of experience in osteometrics, but there was no significant
improvement after 5 years of experience. Clearly, there is some level of subjectivity even
within metric sex estimation, requiring some training by the researcher. This can be
compared with the many years of training and experience necessary to become familiar
with visual sexual dimorphism within a single population, especially to accurately assess
sex from the cranial morphology (
Buikstra and Ubelaker, 1994
). Krogman explains the
dichotomy between metric analysis versus descriptive analysis as “experience versus
6
As further evidence of this, refer to the recently formed Scientific Working Group for Forensic
Anthropology (SWGANTH), whose goals are setting best practices and minimum standards for forensic
anthropology. See
www.swganth.org
.
7
An excellent explanation for how to interpret the results from a discriminant function analysis is available
in FORDISC 3.1 under the help menu (
Ousley and Jantz, 2005
:12).
