Which statistical analysis do I do here?

Question

Hi there,

I have 1 old drug and 4 new drugs and They all have the ability of inducing hormone X in patients. I need help in:

1. how to calculate how well the old and new drugs induce hormone X in patients?
2. Compare if the old drug is better at inducing hormone X than the new drugs?

also, the old drug has a sample size of 37, meanwhile, the new drugs have lower sample sizes (due to being a new drug and is not tested on many people) like 4 or 6. What do I do in this instance?

Answer 1

This is the sort of question that you should take up with your local university. Either their math department or their medical school, if they have one, as the answer will require looking at the data on both the medications and the methods of testing.

However, I can advise this. You should have either before and after serum level measures or a delta (either percentage change or absolute change for each person tested). It is the individual improvements/differences that you want to test between the different medications, not the absolute serum levels as the variance between different patients will cause all sensitivity in the test to be lost. This is key.

Once you have the difference numbers, the type of test should be a one way anova with possible post-hoc Bonferroni or Tukey's HSD test followup. The GLM model will adequately take care of differing sample sizes.