# Assessing difference observations of two samples

I have two continous samples: background and data. I want to check whether data observations are significantly greater than background. I used U-Test to calculate p-value, but now I want to find a pair of background/data with the biggest background >> data between them.

In practice I have a formula to which calculates some stuff and return one value per observation. I want to show that if I apply this formula the result is significantly bigger for relevant data (non-background) and then among 6 formulas I want to choose the one which gives a biggest difference in results.

-
I do not quite follow your question, do you just need the pair of observations with the largest difference? –  user10525 Jul 12 '12 at 13:14
no amongst formulas I need to find one which produces larger values for data compared to noise. I could just take the one with biggest mean ratio(mean data/mean noise), but I don't know if that is correct –  Sergej Andrejev Jul 12 '12 at 20:27