I ran a within-subject study with the primary measure of interest being response time. Due to response time being right skewed I used Wilcoxon signed ranks test on the median values. I am having to revisit this work and I am now wondering/checking that using the Wilcoxon signed ranks test is feasible when working with response time.

The study was a within subject design with a single two level factor and the main measurement was reaction time. The total number of participants was 15

  • $\begingroup$ Why are you choosing the wilcox test? I understand that the response time measures are skewed but you are with a central tendency which is not skewed. Please correct me if I am wrong. $\endgroup$
    – knk
    Jan 23, 2018 at 6:00
  • $\begingroup$ My data is showing right skewed. My understanding is that if I had more data then it would be normal based on the clt, however the data I have is right skewed as such I thought I had to either transform or use a non-parametric. $\endgroup$
    – Doc
    Jan 23, 2018 at 17:12
  • $\begingroup$ Alright, I think I understand your line of thought here. The samples you have are not normally distributed but the "population" would be normally distributed? $\endgroup$
    – knk
    Jan 23, 2018 at 21:15
  • $\begingroup$ In this case you should always get more samples so that you have enough power in your test (whichever test you end up using) $\endgroup$
    – knk
    Jan 23, 2018 at 21:18


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.