Hello,
I am searching for a statistical test to compare two proportions of responsive cells.
I know that Khi2, z-test and Fisher’s exact test allow this kind of comparison but the issue I have is that I have several proportions in each cases (ie : several samples on the same condition that all contain a proportion).
For example:
Drug A = [12/25, 1/100, 40/70]
Drug B = [17/35, 40/200, 20/50]
I suppose that I cannot use a t-test because the data is discrete (1 if the cell respond, 0 if it does not).
I also does not want to do the mean of my data because I want to consider the variation between each measures.
Should I use a GLM with a Bernoulli law or something like that?
Really looking forward to your inputs!
Thank you very much in advance :)
Angel
Hello Angel,
If the samples were systematically different in some way(s), then you'd have to model the additional variable(s) in whatever analysis was conducted. For example, if one sample came from a warmer room/lab than another, or if different substrates were used.
If, on the other hand, you consider the samples to be otherwise interchangeable within a given drug, then perhaps a simple odds ratio or logistic regression on the summated cases (e.g., using your sample values, drug a = 53 / 195; drug b = 77 / 285) might suffice to answer your question about drug differences.
Good luck with your work.
Hi David and Jos,
Thank you so much for your help !
David :
Actually, the several measure represent several replicates of the same experiment. What you propose seem to be a really good option but I would like to take into account the inter-sample variability and the relative importance of each sample. For exemple in the case of Drug A, the second sample (1/100) is really different from others samples but contain a hudge amount of cells. If I sum all the sample, this sample is going to take a huge importance and minimize the others results. Or maybe should I do a "Weighted proportion" ? What do you think about it ?
Jos :
Thank you for make me discover the Cochran-Mantel-Haenszel test ! This kind of stratified approach seems perfect. If I understand correctly, the CMH tests my proportion in function of a first variable (here, each of my drugs) which contain another variable (here my replicates)? (see attached table)
Bests,
Angel
- Badges
- Science topic
- Similar topics
- Chemistry
- Pharmacology
- Pharmaceuticals
- Drugs