# Thread: How to convert data for Bland Altman plot?

1. ## How to convert data for Bland Altman plot?

I have data from two different tests, one is an ultrasound and the other is an antibody based test. My aim is to validate the efficacy of the latter over the former or at least say that it is as good as as the ultrasound test. I have seen that other research papers apply Bland-Altman plot to see how well two tests agree. I am trying to do the same. The aim of these two tests is to establish stage of disease, so I converted their outputs into stages of 0,2,4 (or 10,20,40) based on given ranges. Now when I apply the BA plot, it gives a kind of output which I dont think is right at all.
Can anyone suggest what I should do to implement BA on this data and if not possible is there any other test that can help in establishing validity of the antibody based test?

2. ## Re: How to convert data for Bland Altman plot?

hi,
this is fairly little to go on, but it would seem that the two methods kind of agree for low stages and wildly disagree at the medium stage . Based on 1 point it is hard to say, but there is a hint of agreement at high stage again.

So, maybe you could look a bit more into why you have a such a wide range at middle stages - and you would probably need more data points.

Without knowing the details, I would think that this is plausible - generally it is much easier to recognize the high and low values than the middle ones.

Regards

3. ## The Following User Says Thank You to rogojel For This Useful Post:

Tabi (03-22-2017)

4. ## Re: How to convert data for Bland Altman plot?

Thank you so much for your comments. Agree, cant conclude much with this output.
Originally Posted by rogojel
hi,
this is fairly little to go on, but it would seem that the two methods kind of agree for low stages and wildly disagree at the medium stage . Based on 1 point it is hard to say, but there is a hint of agreement at high stage again.

So, maybe you could look a bit more into why you have a such a wide range at middle stages - and you would probably need more data points.

Without knowing the details, I would think that this is plausible - generally it is much easier to recognize the high and low values than the middle ones.

Regards

 Tweet

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts