Dans la serie ortollj coupe les cheveux en quatre :
Je me demande si le fait de passer sous silence le taux de faux negatif ne serait pas deja un biais cognitif ! . Bon ici on s’apercoit que le fait de ne pas le mentionner signifie implicitement qu’il est a zero . Mais je ne pense pas qu’un test medical quelconque puisse exister avec un taux de faux negatif a zero. Donc pour etre plus rationnel il aurait fallu soit preciser que l’on utilise un cas d’ecole ou le taux de faux negatif est egale a zero, ou alors au moins preciser que le taux de faux negatif est négligeable ( T_n< 1%) par exemple.
Parceque avec un taux de faux negatif important disons a 40% on passe a 0.6/49.95=1,2%
la-conversation-scientifique/qu est ce que lesprit critique
Ci dessous un extrait du livre de Gerald Bronner: Déchéance de rationalité
Extrait du livre de Gerald Bronner dans google doc
Arbre sur Sage Cell
le code SageMath:
- Code: Tout sélectionner
p=100000 ;print "Population : "+str(p)
sr=1/1000 ;print "Sick Rate : "+str(sr)
pfr=95/100 ;print "Positive False Rate : "+str(pfr*100)+'/'+"100"
nfr=40/100 ;print "Negative False Rate : "+str(nfr*100)+'/'+"100"
PbNS=(p-p*sr)*(1-pfr);print "Positive but Not Sick : "+str(round((p-p*sr)*(1-pfr),5))
PaS=((p-(1-sr)*p))*(1-nfr); print "Positive and Sick : ",str(round(((p-(1-sr)*p))*(1-nfr),5))
print "chance to be Sick if Tested Positive : ",PaS/(PbNS+PaS)," about :",round((PaS/(PbNS+PaS)),5)," : ",\
round((PaS/(PbNS+PaS))*100,5),"%"
H0=DiGraph({0:[1,2], 1:[3,4], 2:[5,6]})
#H.set_latex_options(edge_labels=True)
H0.set_edge_label(0, 1, "Not Sick : "+str(p-sr*p)+"/" +str(p))
H0.set_edge_label(0, 2, "Sick: "+str((p-(1-sr)*p))+'/'+str(p))
# not sick branch
H0.set_edge_label(1, 3, "Not Sick \n Tested Negative: "+str(pfr*100)+'/'+"100")
H0.set_edge_label(1, 4, "\n\n\n Not Sick \n Tested Positive: "+str((1-pfr)*100)+'/'+"100")
# sick branch
H0.set_edge_label(2, 5, "Sick Tested Negative: "+str((nfr)*100)+'/'+"100")
H0.set_edge_label(2, 6, "\n\n\n\nSick Tested Positive: \n "+str((1-nfr)*100)+'/'+"100")
H0.relabel({0:str(p)+' Peoples' , 1:str(p-p*sr), 2:str(p-p*(1-sr)), \
3:str(round((p-p*sr)*pfr,5))+' \n Tested Negative and Not Sick', \
4:str(round((p-p*sr)*(1-pfr),5))+' \n Tested positive but Not Sick', \
5:str(round(((p-(1-sr)*p))*(nfr),5))+' \n Tested Negative and Sick', \
6:str(round(((p-(1-sr)*p))*(1-nfr),5))+' \n Tested Positive and Sick'})
H0.show(vertex_color='white', vertex_shape='_',edge_labels=True,figsize=[10,10],layout='tree')
ce post est en relation avec le post ci dessous:
lumieres-tenebres