3
$\begingroup$

I have two datasets and want to know what analyses to perform and how to implement them (GraphPad Prism is preferable, but SPSS could also help).

  • Case 1: I have 3 treatment groups and 5 time-points. At every time-point I have a score from 1 to 10 from the same animal. I want to check the difference between 3 drugs at every time-point (i.e get the p value for the means in 3 treated-groups at every time-point). Should I use two-way Repeated Measures ANOVA for this purpose or just two-way is also fine?

  • Case 2: I have 2 genotypes and 2 (not 3) treatments but also 5 timepoints. How should I analyse this data?

$\endgroup$
  • $\begingroup$ It sounds like you want to run 15 t-tests, 3 at each time interval (which, themselves, wouldn't be very informative about the drugs effect over time). Don't you have any theory that might guide this analysis? From what you described it's just a fishing expedition. In order to plan a proper analysis one needs to know why you ran the study, what you were looking to find. $\endgroup$ – John Sep 23 '11 at 17:59
  • $\begingroup$ Thank you. It's quite hard for me to answer your question. I actually did 15 t-tests before and have some idea about p-value for every time-point. Now I need to publish this data and want to show it in more attractive manner. You know when you run experiments you don't have time to study statistics. Now I'm trying to understand at least basics. I really don't know how to describe my task better. In the 1st case i need to show that score in the group treated with 1st drug is significantly higher than in 2 others starting from 2d time-point (this significance i got with t-tests). $\endgroup$ – Grey bear Sep 23 '11 at 18:08
  • 3
    $\begingroup$ Anybody "running experiments" should take the time to study statistics! $\endgroup$ – pmgjones Sep 24 '11 at 2:55
1
$\begingroup$

For case 1 it sounds like a 3 by 5 mixed ANOVA would be suited. You could then perform follow up tests such as analysis of simple effects, post hoc tests, contrasts, etc. depending on the status of initial tests. This scenario is a simple variant of another questions involving a 2 by 5 mixed design. See my detailed answer to that question here for further detail.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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