# Untitled

a guest
Mar 17th, 2018
132
Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
1. sim1 = function(n=40, mu=0.6343) {
2.   a  = rnorm(n,   0.0, 1)
3.   b1 = rnorm(n/2, mu - mu/4, 1)
4.   b2 = rnorm(n/2, mu + mu/4, 1) # interaction is mu/2
5.
6.   dat = data.frame(Group = c(rep("C", n), rep("T", n)),
7.                    Sex   = rep(c(rep("F", n/2), rep("M", n/2)), 2),
8.                    Val   = c(a, b1, b2))
9.   fit   = lm(Val ~ Group*Sex, dat)
10.   return(summary(fit)\$coefficients)
11. }
12.
13. sim2 = function(n=40, mu=0.6343) {
14.   a  = rnorm(n,   0.0, 1)
15.   b1 = rnorm(n/2, mu - mu/4, 1)
16.   b2 = rnorm(n/2, mu + mu/4, 1) # interaction is mu/2
17.
18.   dat = data.frame(Group = c(rep("C", n), rep("T", n)),
19.                    Sex   = rep(c(rep("M", n/2), rep("F", n/2)), 2),  # we only exchange labels M & F
20.                    Val   = c(a, b1, b2))
21.   fit   = lm(Val ~ Group*Sex, dat)
22.   return(summary(fit)\$coefficients)
23. }
24.
25. n = 40
26. d = power.t.test(n=n, power=0.8)\$delta
27.
28. # variation 1
29. res1=sapply(1:1000, function(x) sim1(n, d))
30. mean(res1[14,]<0.05) # power main effect
31.
32. # variation 2
33. res2=sapply(1:1000, function(x) sim2(n, d))
34. mean(res2[14,]<0.05) # power main effect
RAW Paste Data