# Mean difference

I have two sets of sample data with the first one having the mean of

100.1


and the second set having the mean of

192.1


and if i were to find the mean difference between this two sets, how do i know which of the mean above should subtract with another mean?

For example, i could use the first set to subtract the second set mean

100.1 - 192.1 = -92


while i could use the second to minus the first:

192.1 - 100.1 = 92


I'm confused on which should minus which as i will then need to use the mean difference to calculate the 95% interval. Would love some advice on this.

• It doesn't really make a difference. Some like to keep the mean difference positive. What you are looking for is to see if zero is in the confidence interval as an indicator to reject or fail to reject the null. If that's the case then the CI limits will change signs anyway. – Phil H Sep 21 '18 at 1:43
• It depends on which CI interval you supposed to calculate, $X-Y, Y-X$ or |$Y-X|.$ – callculus Sep 21 '18 at 1:49