Simple t-test Question - Please help - Paired or not? (May/16/2008 )

Sorry if this seems a really simple question, but I've always been a little baffled by maths and after trawling through google I am still confused. In my experiments I have say 2 dishes of cells, of the same cell type. I then apply stimulant X to dish A, whilst dish B remains untreated as a control. I have then done real-time PCR for a variety of targets. My question is.... when doing a t-test do I want to use a paired or non paired test? They are the same cell type, but are obviously different populations of the same cell type, as are in 2 dishes. I understand paired to be used, for example, for taking my temperature, making me run 10miles and then taking it again, ie the SAME subject. Does this still apply to 2 dishes of cells of the same type, one having undergone treatment?

Any help would be greatly appreciated!

Thanks

---

It doesn't matter whether you have the same cells A (after Rx) and B (Before Rx), what matters is that within the A (and within B), the population is homogenous.

In this experiment, you can use paired test, because there are only two populations-A and B, and both have homogenous population.

If you had two populations which had significant difference within each of them, for example, A is a mixture of many diff cells, B is another mixture of same but many different cells, you would use unpaired.

Does it make sense? Another example, if you are studying two groups of mice (Wt and KO) of the same strain, you need to also worry about diferences in their weight, age, sex, etc within each group, in addition to KO effect, and so you use unpaired t-test.

My understanding is this, I may have forgotten over a period of time, and I may stand corrected.

---

Ah, OK, thankyou, that makes more sense now. Its difficult finding answers from google when you a a bit 'mathematically challenged' - too much jargon. Thanks for your help : )

---