Hi all,
Does anyone know how to implement the signal detection theory (e.g., measures of sensitivity, ROC curve etc) to analyse data in Psychology field using JMP?
Thank you.
Hi @Peter_Bartell,
Thank you for your comment and link. I am novice user in JMP Basic and some details are shown, as follows:
- My analysis is based on repeated-measures design (e.g., between-subjects = T1, T2, ...)
- In general, you have a matrix M_2 x 2 which rows are represented by Stimulus event = {s,n} and columns Response event = {S, N}.
Matrix M:
S N
s M[1,1] = "CA" = P(S/s) M[1,2] = "FR" = P(N/s)
n M[2,1] = "FA" = P(S/n) M[1,2] = "CR" = P(N/n)
where "Correct Accept", "False Rejection", "FA" = "False Alarm" and "CR" = "Correct Rejection", and P(*/*) is a conditional probability.
So I need to analyse that matrix M (e.g., CA vs FA) using the signal detection theory (SDT) through ROC curve and sensitivity index in a repetead-measures design.
Thank you.
Measures of model sensitivity and the means to visually communicate them (ROC curves, Confusion matrices, etc.) are widespread throughout many JMP modeling platforms for categorical type responses. A partial list of the platforms are nominal logistic regression, Partition, and if you are a JMP Pro user, Bootstrap Forest, etc. So if you can provide some more details of the use case in psychology perhaps community members can provide additional guidance. Here is the general JMP online documentatioon link for ROC curves...additional details can be found in the specific platform documentation.
https://www.jmp.com/support/help/13/ROC_Curve.shtml
