0

I just have a very quick question on some data I have. I was asked to collect data for tumor emergence in two groups of mice. One treated with PBS and another treated with PBS plus Matrigel. The one with matrigel had a median day appearance of 183 days post implantation, while the PBS one had a median day appearance of 298 days post implantation.

How would I test if the medians between the two is significant. I know what tests to use for the mean number of days, but my PI wants the median number of days and the corresponding significance of the difference between the two groups.

Searching online it seems like the Mann-Whitney U test would be appropriate, but some are saying that it isn't meant purely for medians.

The data I have is quite simple and I'd like to do everything on excel if possible. See below

enter image description here

Thank You!

N.H.
  • 1
  • why not just compare the two distributions directly instead of their medians? in Excel it would be pretty easy to set up a KS test (Kolmogorov-Smirnov) because you can do it with bins instead of needing the code or software to calculate it with integration – Josh Dec 18 '20 at 22:04
  • 1
    Does this answer your question? [Hypothesis test for difference in medians among more than two samples](https://stats.stackexchange.com/questions/81864/hypothesis-test-for-difference-in-medians-among-more-than-two-samples) – Igor F. Dec 19 '20 at 04:56
  • 3
    I don't follow the logic of this question. By analogy, suppose I needed to cut a vegetable and were to ask, "an online search suggests a petit knife would be appropriate, but some are saying that it isn't used purely for vegetables." Why would that preclude its use in my case? Does every tool need to be hyper-specialized to one particular use case? – whuber Dec 19 '20 at 13:42

0 Answers0