Non-parametric Limits of Agreement plot (Bland-Altman)

[Edsel Ing]

I have skewed data but want to do a LoA plot.
Is there a stata plug-in that uses the median instead of the average?

[daniel klein]

I do not fully understand the request. There is more than one average involved in the LoA plot (Bland and Altman 1986). Do you want the median of the two raters on each subject, i.e. the median instead of the mean rating? If so, there is no point to it as the two are identical for any pair of ratings.

Or, do you want the median instead of the average of the differences in ratings? If so, would still want the limits to be based on the SD of the difference?

Best
Daniel


Bland, M. J., Altman, D. G. (1986). Statistical Methods for Assessing Agreement between two Methods of Clinical Measurment. Lancet, 1, 307-310.

[Nick Cox]

Like Daniel, I am not clear on what you want instead. This machinery is based on means and differences between them. I have never heard that you can just switch to medians. Perhaps you should be comparing on some transformed scale.