Testing the reliability of software tools in sex and ancestry estimation in a multi-ancestral Brazilian sample

Ancestry estimation of three Mediterranean populations based on cranial metrics

“…ancestry estimation methods, traditional and computer-based, problematic, especially taking into account the low accuracy rates reported by numerous validation studies [24][25][26][27]. The main source of bias is the potential lack of the target specimen's source population in the reference sample [24][25][26]. In controlled conditions, such as in identification of individuals of known nationality from a mass accident or casualties from war zone involving known ethnic groups, this problem is minimised.…”

“…The results indicated that Non-standard inter-landmark distances gave the highest classification accuracy. Urbanová et al [24] tested the accuracy of ancestry and sex assessment using four identification software tools (FORDISC 2.0, FORDISC 3.1.293, COLIPR 1.5.2 and 3D-ID 1.0) in a sample of 174 documented human crania of Brazilian origin belonging to different ancestral groups (i.e., European Brazilians, Afro-Brazilians, Japanese Brazilians, and individuals of admixed ancestry) and reported barely 50% correct classification accuracy. Validation studies clearly demonstrate that software platforms (either inter-landmark distance or landmark-based) have a limited ability to correctly and reliably identify subtle biological intra-and inter-population variation which has not been captured in the reference sample [24][25][26].…”

“…Urbanová et al [24] tested the accuracy of ancestry and sex assessment using four identification software tools (FORDISC 2.0, FORDISC 3.1.293, COLIPR 1.5.2 and 3D-ID 1.0) in a sample of 174 documented human crania of Brazilian origin belonging to different ancestral groups (i.e., European Brazilians, Afro-Brazilians, Japanese Brazilians, and individuals of admixed ancestry) and reported barely 50% correct classification accuracy. Validation studies clearly demonstrate that software platforms (either inter-landmark distance or landmark-based) have a limited ability to correctly and reliably identify subtle biological intra-and inter-population variation which has not been captured in the reference sample [24][25][26]. Elliot and Collard [27] also tested FORDISC in a sample of 200 individuals of known ancestry "with and without the test specimen's source population included in the program's reference sample, and with and without specifying the sex of the test specimen" and concluded that no more than 1% of the sample was classified with a high degree of confidence.…”

The vanishing Black Indian: Revisiting craniometry and historic collections

“…Many evaluative studies of forensic discriminant function analysis exist in the literature (Belcher, Williams, & Armelagos, ; Campbell & Armelagos, ; Cross & Wright, ; Elliott & Collard, ; Freid et al, ; Kallenberg & Pilbrow, ; Keita, ; Kosiba, ; L'Abbe, Kenyhercz, Stull, Keough, & Nawrocki, ; Leathers, Edwards, & Armelagos, ; Ross, ; Ross, Slice, Ubelaker, & Falsetti, ; Sejrsen, Lynnerup, & Hejmadi, ; Sierp & Henneberg, ; Spradley & Jantz, ; Williams, Belcher, & Armelagos, ; Wittwer‐Backofen et al, ; Ubelaker, Ross, & Graver, ; Urbanová, Ross, Jurda, Nogueira, ). Detractors argue that these programs are of little use if unknown individuals do not fall within the reference populations' geographic or temporal limits (Belcher et al, ; Campbell & Armelagos, ; Elliott & Collard, ; Keita, ; Kosiba, ; Leathers et al, ; Williams et al, ).…”

Estimativa da ancestralidade em Antropologia Forense por meio do software \"AncesTrees\" em medidas cranianas de uma amostra brasileira

“…Já as Figuras 36 e 37 demonstram o desempenho destes sobre indivíduos brancos e negros, em separado. (BIRKBY, KAJANOJA, 1966;LA BOULINIER, 1968;HENKE, 1973;SNOW et al, 1979;VAN VARK et al, 1982), capitaneadas pelos estudos primordiais de Hanihara (1959) e Giles & Elliot (1962a-b).…”