Hi all,

I have some data which I am trying to compare; 3 experimental datasets and 2 control datasets (positive and negative control). Each set consists of 6 percentages, each from a different repeat of the same experiment. I basically want to compare each experimental set to the positive and negative control (separately) to confirm which ones are similar.

In other words, 2 experimental datasets should be very similar to the negative control and 1 very similar to the positive control. I can see visually that this does appear to be the case, but I want to confirm it somehow if I can.

I have done this already using Mann-Whitney U tests, but now I realise that this doesn't particularly help. For example, in one case I got a P-value of 0.361. That tells me they're not significantly different, but are they 'significantly similar'? Is significant similarity even a meaningful concept?

If anyone can help it would be very much appreciated, statistics aren't really my greatest skill.

Thanks, Malik.

I am a bit confused by your data, and what you are trying to achieve.

Suggest you do some googling of non-inferiority tests, and possibly bland-altman plots to see if either of them works for you.

Thanks Thesisfun.

Sorry, I don't think I explained it very clearly.

I am looking at the viability of mutant bacteria in response to a compound, and this is recorded as a percentage of the viability observed in normal conditions. In the positive control mutant and one of the test mutants, viability is almost 100%. In the negative control mutant and two of the test mutants, viability is almost 0%. I want to show whether each test mutant is as resistant/sensitive to the compound as the control is.

I think a Bland-Altman plot may work, I haven't had a chance to look thoroughly yet though so any other suggestions from anyone in the meantime would be very welcome.

Thanks again, Malik.