Multiple Tests, an Introduction

Last week, a student asked me about multiple tests. More precisely, she ran an experience over – say – 20 weeks, with the same cohort of – say – 100 patients. An we observe some https://latex.codecogs.com/gif.latex?X_{i,t}

size=100
nb=20
set.seed(1)
X=matrix(rnorm(size*nb),size,nb)

(here, I just generate some fake data). I can visualize some trajectories, over the 20 weeks,

library(RColorBrewer)
cl1=brewer.pal(12,"Set3")[12]
cl2=brewer.pal(8,"Set2")[2:7]
cl=c(cl1,cl2)
boxplot(X)
for(i in 1:length(cl)){
lines(1:20,X[i,],type="b",col=cl[i])  }

http://freakonometrics.hypotheses.org/files/2014/09/Selection_585.png

Continue reading Multiple Tests, an Introduction