# Including the variation within a sample when comparing between samples

I want to include the variation within a sample when analysing the variation between samples.

Say I have sample A, B, C.

Sample A was analysed three times with the values: 2398, 2383, 2411 Sample B: 2530 Sample C: 2391

The SD of sample A is 14.0 The SD of sample A, B, C with the values 2398, 2530, 2391 is 78.3

How do I make sense of this? At what threshold value can I determine if a sample differ from one another? How do I include the variation within a sample when calculating the variation between samples?

• You should first start by stating your objective. Are you trying to study differences in mean or variance? – Arun Jose Sep 27 '16 at 11:01
• Probably both, I have to find a good parameter to compare the samples be it mean or variance.. – Lennart Sep 27 '16 at 11:03
• In that case test for each with a separate test, one for verifying if variances are the same, and the other if the means are the same. – Arun Jose Sep 27 '16 at 11:06
• Yeah, and how do I determine that? They will differ to some degree but at what point can I say that they differ or is that threshold arbitrarily set up as a hypothesis by myself? – Lennart Sep 27 '16 at 11:09
• There is no universal rule on determining the value of alpha. The only thing you can go on is if you have any criteria specified about the false positives or negatives that can be tolerated. – Arun Jose Sep 27 '16 at 11:16