general: many points raw: one point
Gaussian
sensitivity A = area under ROC-curve for A’ = area under curve for 1 point Az Ad’ = area d-prime = area under ROC
many points curve by using d’
counting squares AZ < 0.5 means sampling error or theoretical maximum = 1; high
response confusion sensitivity, perfect discrimination
theoretical maximum = 1; signal
AZ = 0.5 means guessing
perfectly distinguishable from noise
AZ > 0.5 the higher, the more theoretical minimum = 0; low
sensitive the subject and/ or the sensitivity, not able to discriminate
A’ = 0.5; guessing, so signal cannot be
easier the test
distinguished from noise
d’ = d-prime (parametric) = distance
between signal and noise means in
theoretical minimum = 0; sampling error/
terms of standard deviation
response confusion
or d’ = ZH - ZF
use when
1. the signal and noise
distributions are both normal
Gaussian
sensitivity A = area under ROC-curve for A’ = area under curve for 1 point Az Ad’ = area d-prime = area under ROC
many points curve by using d’
counting squares AZ < 0.5 means sampling error or theoretical maximum = 1; high
response confusion sensitivity, perfect discrimination
theoretical maximum = 1; signal
AZ = 0.5 means guessing
perfectly distinguishable from noise
AZ > 0.5 the higher, the more theoretical minimum = 0; low
sensitive the subject and/ or the sensitivity, not able to discriminate
A’ = 0.5; guessing, so signal cannot be
easier the test
distinguished from noise
d’ = d-prime (parametric) = distance
between signal and noise means in
theoretical minimum = 0; sampling error/
terms of standard deviation
response confusion
or d’ = ZH - ZF
use when
1. the signal and noise
distributions are both normal
