Assume 2 groups, n=30, d=0. You are allowed to choose any 6 (i.e. 10%) of the values from either group to remove. Goal: manipulate / increase any existing difference due to chance. How often will you hit p=0.05? Regular vanilla btw.subj. t-test.
Intuitively I guessed it would be high. Just tested it on a range of datasets, I haven't failed to get the p<0.05 once so far as long as I can remove 10% of the values
-
-
I can't reproduce that.
-
Actually, the corollary is almost as much fun - I can get all of the p<0.05 above the 0.05 limit if I delete 10% of the values as well, so I can change the direction of "significance" either way for most outcomes
- Show replies
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.