# t-test on the difference of two independent observations

I want to check if there is any significant difference in my two sets of observations (which are not dependent). Now my first observations have two stages stag1_before and stage2_after same is with the second set of observations. What t-test should I perform? I want to perform between the difference of set of observation one and difference of set of observation two.

here is some sample

## 1 Answer

In principle you can perform a t-test for mean comparison between your two samples (dif1 and dif2)

You can find more info here http://www.stat.yale.edu/Courses/1997-98/101/meancomp.htm where variances are unknown and we assume $$\mu_1 = \mu_2$$ as the null hypothesis.

However, I am concerned that not all the observations you've shown follow the same distribution (for the test to be valid, each individual should be independent and identically distributed with the rest of individuals of the population), so take everything your test tells you with a grain os salt

• In such unequal distribution can I use t-test with unequal variances? – alo Jun 13 '19 at 21:49
• It's definitely the one that fits the best, but this test still assumes equal distribution within each group – David Jun 13 '19 at 22:48