I have calculated two standard errors for two fishers z correlations from the same study , both measuring the same variable. Is it acceptable for me to average the two standard errors into one? If so , how do I go about this please?



If combining makes sense (*) (**), then two things:

  • work with variances and then take square root;
  • a weighted average should be used,

just like in the independent two-sample t-test where the pooled standard deviation is calculated as

$$ s_p = \sqrt{\frac{(n_1 - 1) * s^2_1 + (n_2 - 1) * s^2_2}{n_1 + n_2 - 2}} $$

(*) not enough detail is provided

(**) cf. exchangeability

  • $\begingroup$ Here is why I would need to combine them - Originally , I had to calculate the standard error and variance for both combined before entering the data into SPSS to calculate a meta analysis. I have originally combined the two fishers z to ensure that they are an independent sample for meta analysis , yet when it comes to calculating a standard error and variance I believe that I need one sample size and I have no idea how to combine the two sample sizes? $\endgroup$ – laboh Aug 6 '18 at 14:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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