I have 200 species of animals and I want to check whether giving them a medicine can affect them in a significant way. I study 20 days before administering the medicine and 20 days after, such that an example of my data is as follows:

Species     |    Date     |     Value
   A        |    ....     |      4
   A        |    ....     |      10
  ...            ....           ...
   B        |    ....     |      50


I know that by using a two-way t-test for each individual species I can determine whether the medicine had any effect by comparing the 20 days before to the 20 days after, but since I have 200 species it means I have to run 200 separate two-way t-tests, so I was wondering whether there is a one single statistical test which can determine whether the medicine had a significant effect. I thought about using something like an ANOVA test but that doesn't seem appropriate (or at least what I found online doesn't seem to support it) so I thought I should ask here to find out what I can do. Thanks!

  • $\begingroup$ A linear (mixed effects) model, if you examined the same animal before and after. If you have two different animals for before and after, a regular linear model. $\endgroup$ Commented Sep 16, 2019 at 6:41

1 Answer 1


This would be most suitable as a comment but I don't have comment privileges yet. Do you want to look at how each species reacts to the medicine? If not, you can just look at the overall before and after, treating the species as subjects and running one paired samples t-test.

  • 1
    $\begingroup$ Yeah at the moment I've started to do that since it seems like a good idea to split everything up that way. I'm examining the difference between the values pre- and post-medication and then running a paired t-test to check if the change is significant $\endgroup$ Commented Sep 15, 2019 at 4:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.