- I’m looking at evaluating level of agreement b/w 2 scores taken from the same person, 2 wks apart

- I have produced a bland-Altman plot. Mean differences b/w scores normally distributed

- as expected 95% of the diffs fall within the LoA

- there does not seem to be any systematic error

- the measure is not ‘objective‘ in the sense that there’s a clear threshold at which a difference would be considered too much/too high, so I can’t see how I can nominate an acceptable LoA a priori (as I understand you’re meant to do)

- so, what method/s can I use? could I calculate the Smallest Detectable Change and make this this the LoA?

- in other words, what methods can I use to determine whether the differences are ‘too large’