@ajordannafa @PhDemetri @camjpatrick Of course. Andy Gelman to the rescue! Also, I found this paper which shows that type I error rates go up when outliers are removed. Don’t like it because I prefer thinking about type M/S errors but this might suffice
Type I error > 20% after removing outliers with a threshold value of Z = 2 http://t.co/RS3yv0DPvr
Type I error > 20% after removing outliers with a threshold value of Z = 2 http://t.co/RS3yv0DPvr
Type I error > 20% after removing outliers with a threshold value of Z = 2 http://t.co/RS3yv0DPvr