Converting Group Centroids into Original Measurement Units in DFA

[ckevinbarrett]

I used SPSS to perform a discriminant function analysis on leg length which I am using to differentiate males and females in my sample. New measures are classified as male if the discriminant value is above the mean of the Group Centroids, or female, if the value is below.

What I would like to have is the "cut-off" (mean of the Group Centroids) in the original measurement units (mm), so that I can classify new cases as male or female without multiplying each measure by the coefficients and comparing it to the mean of the Group Centroids. Is there an easy way to do this?

 

[Dason]

Is leg length the only variable you're using? Is the function you're using to compute the discriminant value linear? If so then yes this would be relatively easy.

 

[jrai]

Is leg length the only variable you're using? Is the function you're using to compute the discriminant value linear? If so then yes this would be relatively easy.

Dason: Can this exercise be seen as converting Mahalanobis distance to Euclidean distance? Even I'd be interested to know how we can exactly do this? Thanks.