Hi everyone,

I am trying to carry out a model II linear regression on a simple x,y scatter plot. This is different from a standard linear regression in that measurement error is assumed in both the x and y variables, rather than just one of the two variables, when calculating the least squares best fit. I have pasted a link to some information about the two below, in case it helps.

(Introduction to Model I and Model II linear regressions | MBARI)

Model II regression is also referred to as 'reduced major axis' or 'least squares cubic' method. In graphpad it's referred to as 'Deming' regression.

If anyone knows if I can do this in SPSS I would be very grateful if you could let me know. Thanks.

Michaeljohn

That is also referred to as orthogonal regression. That is not available in SPSS directly , but you can do this with principal components via the Factor procedure, or you might try instrumental variables via the TSLS procedure,

Hi Jon,

Thanks for taking the time to reply. I will see if I can carry out the procedures you suggested.

Kind regards,

Michaeljohn