3
$\begingroup$

I am working with patient experience data where patients answer questions regarding their stay at the hospital. Each question is then given a correlation value as it relates to the final "overall experience" question. The data is in 6 month intervals with each question having different correlation values for each 6 month interval.

How can I get one correlation value for 2 years worth of data from four 6 month intervals, i.e., four correlation values?

$\endgroup$
4
  • 4
    $\begingroup$ You can average the Fisher transformations of r-scores, if the N's are the same, or perform a weighted average, if they are not. But, I suspect you'll want to take the temporal aspects of this situation into account, if you want an accurate answer. $\endgroup$ Aug 22, 2013 at 19:11
  • 2
    $\begingroup$ Why don't just merge the data and compute one correlation value? $\endgroup$
    – O_Devinyak
    Aug 22, 2013 at 19:13
  • $\begingroup$ A prior question is whether the different correlations are really different. If so, averaging may not really help. If so, you could think about looking at different effects in a regression model. $\endgroup$
    – Nick Cox
    Aug 22, 2013 at 19:21
  • 1
    $\begingroup$ Are you perhaps asking the same thing as stats.stackexchange.com/questions/8019/…? If not, how does your situation differ? $\endgroup$
    – whuber
    Aug 22, 2013 at 19:51

0

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.