I have one group of participants measured on a health outcome scale at two different time points. I first want to see whether there is a significant difference between time 1 and time 2 on this outcome variable, I then want to see if this outcome variable at T2 is predicted by another dimensional variable (BPD dimensional score). Seems straightforward using a paired samples t-test and then a regression analysis.

However the health outcome variable has an adolescent version and a adult version. At time 1 most of the participants used the HONOSCA (adolescent version) and some used the HONOS (adult version). at the second time point the majority used the HONOS with only a few using the HONOSCA. while there have been numerous studies evidencing that the adult and adolescent versions can be compared, I have not found anything with the use of the different versions used for the same outcome variable. One scale has 13 items and the other 12 items, so total scores cannot be used.

I have transformed the variables into z scores to make them comparable, but is it valid to use the z scores in further analysis? How can I compare T1 and T2 outcomes because the z-scores are not valid for the t-tests?


  • 3
    $\begingroup$ Where did you hear that z-scores are not valid for t-test? z or not z, the t statistics and p-value are the same. $\endgroup$ – Penguin_Knight Mar 29 '13 at 13:30
  • 2
    $\begingroup$ how have you transformed them into z scores - eg which means did you substract, and which standard deviations divide by? If you used different transformations at T1 from T2 you will have removed any structure in the data. $\endgroup$ – Peter Ellis Apr 28 '13 at 23:07

Your Answer

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

Browse other questions tagged or ask your own question.