#dans_ce_temps-là_les_hommes_étaient_des_vrais_hommes_et_les_femmes_des_vraies_femmes

avec une précision de 60% (??)

#femmes #historicisation

Je pensais que le sujet était nouveau, bah non.
Plus que les 60%, je me suis bien marré avec " il avait estimé lui que les mains appartenaient à des adolescents. « Ils peignaient ce qui les intéressait : des femmes nues et des gros animaux effrayants. »"

►http://www.journaldelascience.fr/homme/articles/peintures-rupestres-majoritairement-realisees-femmes-3279

60% c’est donc le résultat de son algorithme prédictif testé sur des européens modernes. Mais heureusement pour lui le dimorphisme était plus prononcé à l’époque :

“Twenty thousand years ago, men were men and women were women.”

►http://news.nationalgeographic.com/news/2013/10/131008-women-handprints-oldest-neolithic-cave-art

SAA : Society For American Archaeology Sexual Dimorphism in European Upper Paleolithic Cave Art
▻http://saa.publisher.ingentaconnect.com/content/saa/aa/2013/00000078/00000004/art00008;jsessionid=701i780stb3r6.alexandra

Abstract:
Preliminary research on hand stencils found in the Upper Paleolithic cave sites of France and Spain showed that sexual dimorphism in human hands is expressed strongly enough to allow empirical determination of the sexes of the individuals who made some of them. Further research increased the sample of measurable cases from 6 to 32, a large enough sample to show that persons who made hand stencils in the caves were predominantly females. This finding rebuts the traditional assumption that human hand stencils in European parietal art were made by male artists, either adults or subadults. Findings further suggest that the sexual dimorphism of hands was more pronounced during the Upper Paleolithic than it is in modern Europeans. Attempts to apply the same algorithms to a sample of North American Indian handprints confirms the view that different populations require separate analyses.

Ça avait été repéré par @touti à l’époque (oct. 2013), mais elle n’avait pas vu cette perle… ►http://seenthis.net/messages/185396

#Dans_ce_temps-là_les_hommes_étaient_des_vrais_hommes_et_les_femmes_des_vraies_femmes

EDIT : et l’année d’après par @heautontimoroumenos , le tout fidèlement pointé par @odilon ▻http://seenthis.net/messages/297794

60% de « bien classés » c’est pour la deuxième étape du processus. La première, c’est 79%.

Women leave their handprints on the cave wall | Penn State University
▻http://news.psu.edu/story/291423/2013/10/15/research/women-leave-their-handprints-cave-wall

Snow found he needed a two-step process for the modern hands to successfully differentiate men from women. He first measured the overall size of the hand using five different measurements. This separated the adult male hands from the rest. Snow found that step one was 79 percent successful in determining sex, but adolescent males were classified as female.

Step two compares the ratios of the index finger to the ring finger and the index finger to the pinky to distinguish between adolescent males and females. For the known hands, the success rate, though statistically significant, was only 60 percent. There is too much overlap between males and females in modern populations.
[…]
The first step in the process showed that only 10 percent of the handprints on cave walls in Spain and France were left by adult males. The second step indicates that 15 percent were placed by adolescent males, leaving 75 percent of the handprints female.
[…]
Snow also looked at modern American Indian hands and found that the rules and algorithms developed for Europeans did not work. He notes that different populations require separate analysis.

Je résume, son algo est vasouillard, déjà le truc à 2 étapes successives…, faiblard (60%, ce n’est pas beaucoup mieux que pile ou face…), et il faut l’adapter à chaque population (moyen sur les Européens contemporains, mauvais sur les Amérindiens contemporains). Il est donc évident que là où il marche le mieux c’est sur les Européens préhistoriques…

D’ailleurs, lui-même n’a guère de mal à battre son algo…

“By just eyeballing, I’m more accurate with the modern hands than the formulas I developed,” said Snow. “There are some variables there that I’m not aware of yet. The algorithms are pretty good, but they could be better.”

Enfin, on rappellera que son échantillon est de 32 mains et donc que 10% ça fait 3 et 15%, ça fait 5.