I am looking at concentrations at 3 sites over 4 years, with 4 samples at each site. Because of the small number of samples, I believe I should use non-parametric statistical tests.

My question is: should I do a Friedman, or should I do 2 separate K-W tests, one for sig. differences between the sites and another for sig. differences over time?

  • $\begingroup$ What are you trying to achieve exactly? As far as i remember the mentioned methods are helpfull to compare different classifiers on different data sets. Is that what you need? $\endgroup$ – Unhandled exception Jul 27 '15 at 22:40
  • $\begingroup$ Thanks for the response. I am trying to assess spatial and temporal trends in heavy metal concentrations in mussel populations. I have 4 samples from each of the three sites over a period of 4 years. $\endgroup$ – Nik Jul 28 '15 at 3:50
  • $\begingroup$ Ok but that doesn't sounds like a classification task for me... what is the output of your algorithm? Do you perform cross validation? $\endgroup$ – Unhandled exception Jul 29 '15 at 6:57

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.