Sex estimation using continuous variables: Problems and principles of sex classification in the zone of uncertainty

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F49777513%3A23330%2F20%3A43960625" target="_blank" >RIV/49777513:23330/20:43960625 - isvavai.cz</a>

Výsledek na webu

<a href="https://www.elsevier.com/books/statistics-and-probability-in-forensic-anthropology/obertova/978-0-12-815764-0" target="_blank" >https://www.elsevier.com/books/statistics-and-probability-in-forensic-anthropology/obertova/978-0-12-815764-0</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1016/B978-0-12-815764-0.00016-2" target="_blank" >10.1016/B978-0-12-815764-0.00016-2</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Sex estimation using continuous variables: Problems and principles of sex classification in the zone of uncertainty

Popis výsledku v původním jazyce

Discriminant function analysis using continuous bone measurements provides a highly efficient tool for sex estimation in forensic practice. We show that three aspects are especially important for a reliable application of DFA in sex estimation: (1) the accuracy of sex classification should be calculated by using cross-validation, as non-cross-validated accuracies usually overestimate the true classification rate. Given that an accuracy greater than or equal to 80% is part of the Daubert criteria for admissibility, an unbiased estimate of the classification rate is of high relevance for forensic practice. (2) The sex of an unknown individual should be estimated by a population-specific discriminant function that was derived from the population to which the individual is thought to belong. The application of a discriminant function based on an unrelated population may result in misclassifications and may produce a large sex bias, that is, large differences between the classification rates for females and males. (3) Sex estimation in the zone of uncertainty, that is, the overlapping area between the sexes, should be avoided to prevent misclassifications. Sex should be assigned only to those individuals with a posterior probability of being female or male higher than 0.95. Although such an approach limits the practical applicability of DFs, because some portion of the individuals remains unclassified, it allows a high classification accuracy to be maintained at the individual level. It is worth considering extending the Daubert criteria by adding a posterior probability threshold to the sex estimation procedure, as a reliable sex estimation of every single individual, and not only the overall accuracy across the entire sample, is of utmost importance in forensic practice.

Název v anglickém jazyce

Sex estimation using continuous variables: Problems and principles of sex classification in the zone of uncertainty

Popis výsledku anglicky

Discriminant function analysis using continuous bone measurements provides a highly efficient tool for sex estimation in forensic practice. We show that three aspects are especially important for a reliable application of DFA in sex estimation: (1) the accuracy of sex classification should be calculated by using cross-validation, as non-cross-validated accuracies usually overestimate the true classification rate. Given that an accuracy greater than or equal to 80% is part of the Daubert criteria for admissibility, an unbiased estimate of the classification rate is of high relevance for forensic practice. (2) The sex of an unknown individual should be estimated by a population-specific discriminant function that was derived from the population to which the individual is thought to belong. The application of a discriminant function based on an unrelated population may result in misclassifications and may produce a large sex bias, that is, large differences between the classification rates for females and males. (3) Sex estimation in the zone of uncertainty, that is, the overlapping area between the sexes, should be avoided to prevent misclassifications. Sex should be assigned only to those individuals with a posterior probability of being female or male higher than 0.95. Although such an approach limits the practical applicability of DFs, because some portion of the individuals remains unclassified, it allows a high classification accuracy to be maintained at the individual level. It is worth considering extending the Daubert criteria by adding a posterior probability threshold to the sex estimation procedure, as a reliable sex estimation of every single individual, and not only the overall accuracy across the entire sample, is of utmost importance in forensic practice.

Druh

C - Kapitola v odborné knize

CEP obor

—

OECD FORD obor

10103 - Statistics and probability

Projekt

<a href="/cs/project/EE2.3.30.0038" target="_blank" >EE2.3.30.0038: Nová excelence lidských zdrojů</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2020

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název knihy nebo sborníku

Statistics and Probability in Forensic Anthropology

ISBN

978-0-12-815764-0

Počet stran výsledku

28

Strana od-do

155-182

Počet stran knihy

418

Název nakladatele

Academic Press

Místo vydání

Amsterdam

Kód UT WoS kapitoly

—