Test Equivalence: Med Student Seeking Help

[awkwardsilence]

Hey you guys! I'm a med student from Sweden in need of some help!

I'm doing a summer course type-a thing in a lab, and I have to present some data from a project I've been doing here. Now, I've tested several genes' expression and I've compared those of an experimental group to a control group. Now that I'm done with the experiments, I've found that some genes show a significant difference in expression between exp. group and ctrl group, and others do not. I think two of the genes are more likely then the others to in reality have equal expression of the gene, in both the groups, but I can't prove a null hypothesis with a t-test...

So, my question now is HOW do I calculate equivalence (and get as close as possible to proving the two samples are equal)?

How do I calculate this in an easy way? (Excel/ online test)

From what I understand, there's one type of test where you calculate two(?) p-values with a regular t-test, then subtract from each other and it should be less p=0.10. I don't really understand it though... Like what two p-values could I get from the same samples? :S

I'd love to get an answer from any of you, if you have an idea about how to solve my problem!

 

[I like Serena]

Hi awkwardsilence! Welcome to MHB! (Smile)

I think you're looking for the Kolmogorov-Smirnov test.
It can test if 2 samples have the same distribution.

We can find online calculators for it.
Excel doesn't have it on its own, but there are plug-ins available.
Or else we can do it directly in excel by introducing an intermediate column.

I'm not aware of a test that involves subtracting p-values from a t-distribution.
As for how that might be possible, I can surmise that we could combine the 2 samples into 1 and deduce a means, standard deviation, and degrees of freedom from it. That would be an approximation of the actual common distribution - if it exists. After that, we can find p-values for each of the 2 samples against that distribution.