I'm needing help with some of my university work here.
I'm trying to find an appropriate t-test for my data, and I've spent 5 hours slamming my head against a wall trying to get it to work and failing. Not a consecutive 5 hours either, this is me going away, coming back with some hilarious misplaced optimism and a fresh new outlook, then ending a frustrated, stressed out mess every fucking time.
The data's on stick insects. We've got 20 results from 20 stick insects under two different circumstances: The absence of wind, and the presence of wind. The data we got is the number of seconds spent swaying over 10 minutes. These were the same stick insects in both circumstances (eg group 1 did stick insect A with no wind, gets 5 seconds. The same group then used a cold hair dryer to blow air over the same stick insect, resulting in 20 seconds. Repeat over another 19 groups)
The data's not normally distributed, I did a test and pulled up a histogram both confirming it.
I'm lead to believe I either need to do a Kruskal-Wallis test, or a Wilcoxon signed/paired rank test.
First question, am I right and if so which do I use?
Second question, how the actual FUCK do I do these tests (Mine or the new one) in RStudio? The K-W test produced different results if I compared NoWind to Wind or Wind to NoWind, which seems fucking weird to me - I'm comparing the exact same values in reverse order, why would I get wildly different chi squared, DF and P-values?
As for the Wilcoxon test, I just don't know how to input that. I've followed 5 different tutorials and every one shows something slightly different and every one throws out an error.
EDIT: I might have this worked out, if I'm back on the forums screaming at everyone tomorrow you'll know I didn't.