In my dataset, I have 2 columns. Both of them are a percentage. However, they only take the following values : (0,10,20,30,40,50,60,70,80,90,100). I would like to know if I could do a t-test on those 2 columns. I have more than 300 rows in my dataset.

  • 3
    $\begingroup$ Thanks to the size of the samples and the "short tailed" distributions, a t test is an option. But maybe it is easier to use a distribution-free test (e.g. Wilcoxon test or a permutation t test) to avoid boring discussions like "your data are not normal, so the t test is wrong". Make sure to pick a test that fits to the samples being paired or not. $\endgroup$ – Michael M Nov 15 '18 at 18:40
  • 1
    $\begingroup$ Thanks for your answer !I've seen that using a non parametric tests such as Wilcoxon test requires a huge size of dataset. So that's why I was wondering if I could do something like a t-test or not. $\endgroup$ – Syndo rik Nov 15 '18 at 18:45
  • 2
    $\begingroup$ If n=4 is huge, you are right. Otherwise not ;-). $\endgroup$ – Michael M Nov 15 '18 at 19:04

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.