2
$\begingroup$

I am wishing to test the equivalence, or not of two measurement methods. Some background, we have two test instruments (old and new) that measure radiation dose distributions compared to what was planned which gives a gamma value, a percentage. I have performed 40 measurements of different dose distributions on the new device and want to compare it with distributions of the old device. Here is a sample of the data:

enter image description here

I would like some advice as to a method to show that both measurement devices are equivalent. I have looked at the Mann-Whitney U Test, is this appropriate?

Thanks

$\endgroup$
2
  • $\begingroup$ Maybe look at intraclass correlation and bland-altman plots. I don't think the Mann Whitney test is what you want. $\endgroup$ Aug 30, 2023 at 11:01
  • 1
    $\begingroup$ Thanks for that. $\endgroup$
    – James
    Aug 30, 2023 at 11:50

1 Answer 1

4
$\begingroup$

Measurement agreement is best assessed on an absolute scale that is easy to interpret. I suggest computing the mean absolute discrepency between the two measurements, and using the bootstrap to get a confidence interval for that. This is discussed in more detail, with R code, here. You can supplement that with median absolute difference and a histogram of all the absolute differences.

$\endgroup$
0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.