I am using parallel process growth mixture modeling to estimate bidirectional longitudinal dynamics between pain (0-100 theoretical range of values) and sleep measures (0-20 theoretical range of values). Conceptually speaking if I log transform both of these raw measures and then regress their latent variables (intercept and slope) on each other, should I interpret that model estimates as percentage change in x that is linked to percentage change in y. In other words, should I refer to this relationship as to elastic?

  • 1
    $\begingroup$ If you have 0's then, of course, a pure log transform will not work. Are there any actual 0's? If so, what are you doing with them? $\endgroup$ – Peter Flom Dec 29 '13 at 14:23
  • $\begingroup$ With log transformed data such interpretation is correct. $\endgroup$ – Germaniawerks Dec 29 '13 at 14:40
  • $\begingroup$ @peter thanks for pointing out, I don't have 0's so log transformation did generate reasonable values $\endgroup$ – user3143803 Dec 29 '13 at 14:47

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.