0
$\begingroup$

I have some data that was collected with 2 different devices on a sample of 50 subjects. It was also collected for 50 days in a row.

One of the devices is considered "the gold standard" method of assessment, so I want to use that to evaluate the validity of the data from the other device.

Should I randomly select a single date and use its data for the validation or look at the validity of every single day and average it? Or is there another way that I don't know of?

I'm not sure about collapsing the data across days since that would inflate the sample (n=50 vs n=2500)

Improve this question
$\endgroup$
6
  • $\begingroup$ How was the first device was 'validated'? Do you mean it was calibrated? $\endgroup$
    – Elenchus
    Commented Sep 4, 2020 at 0:00
  • 1
    $\begingroup$ Its was validated against the gold standard via correlation and t-test in multiple publications. But that part of the question isn't important. Even if it was the gold standard measure, how should would validity be evaluated in the current situation? $\endgroup$
    – Chris
    Commented Sep 4, 2020 at 1:55
  • $\begingroup$ *edited to remove confusion about the device known to be valid as that is not the focus of the question. $\endgroup$
    – Chris
    Commented Sep 4, 2020 at 1:58
  • $\begingroup$ That makes sense, cheers. I think the method of validation is important though; it's given me a much better idea of what you're trying to do. Are the subjects the same across all 50 days? If so the data from each day is not independent, and it might be worth incorporating that into your test if you use more than one day $\endgroup$
    – Elenchus
    Commented Sep 4, 2020 at 3:40
  • $\begingroup$ Is there anything that might be interesting in the time-course over the 50 days? $\endgroup$
    – Michael Lew
    Commented Sep 4, 2020 at 6:42

0

Reset to default

Your Answer

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

Browse other questions tagged or ask your own question.