M.A. Theses: Marina Elliott, 2008

Craniometric ancestry determination and FORDISC 3.0

FORDISC 3.0 is a computer program that provides ancestry determinations from discriminant anaylses of the cranium.  It is widely used, but has been criticized for inconsistent performance.  FORDISC's creators suggest that the failures are due to inappropriate sample selection and variable choice, or misinterpretation of results.  Specifically, they warn against testing individuals whose populations are not represented in the database.  They also claim that using too many variables reduces success.  Other research suggests that the cranial region may be influencing the results.  Since the face may be influenced by climatic/dietary adaptations, the inconsistencies may be due to the preponderance of facila measurements in FORDISC's dataset.

The study selected 200 individuals from 5 geographic populations from W.W. Howells worldwide database and processed them in FORDISC 3.0.  Analyses were conducted with both the source population included and excluded.  Analyses were also run using both sexes and only the relevant sex.  Test individuals were analyzed using 56 variables as well as sets of 10 variables to isolate the basicranium, neurocraium and facial regions.

FORDISC 3.0 performed well only when the source population was included and the number of variables maximized.  When the source population was excluded, FORDISC 3.0 could not consistently place an individual into a closely related group.  Constraining the sex improved FORDISC 3.0's discriminant ability.  The cranial region datasets performed very poorly and the results were not consistent between populations.  Overall, FORDISC 3.0 performed to a lower level than expected.  The results suggest that caution must be used when applying this program in all but the most restricted contexts.