# What is a good statistical test for independent replicates?

I have a series of measurements in which something is sampled and placed into either category 1 or category 2. I have 3x2x3 factors for each set of measurements, and each set of measurements is done independently 3 times.

What is a good statistical test to look examine the combined effects of each factor, both independently and interactively with each other?

• To be sure, your design is really 3x2x3 (3 factors, all with 3 levels except the second one), with three replicates, that is 54 values for each statistical unit? – chl Feb 4 '11 at 19:34
• there are 3 factors, each of which have 3 more factors, each of which has 2 more factors (so like DrugA/B/C + DrugX/Y + ConditionA/B), for a total of 18 measurement groups – Dave Feb 4 '11 at 19:38
• Lets see: you are saying -- for each unit, you have the treatment like (., ., .) and the dots represent one of ABC, one of XY, on of AB? and you have 18 such units? For eg: a treatment may look like (A, Y, A) for a particular unit? – suncoolsu Feb 5 '11 at 4:06
• yes, but maybe that was a bad choice of letters. It would be more like ABC/XY/NM for an example of A,X,M – Dave Feb 5 '11 at 13:37