vasishth

<<>>=
# n: no of operations
# x: no of deaths
# N: no of hospitals
dat<-list(n=c(47,211,810,148,196,360,119,207,97,
              256,148,215),
          x=c(0,8,46,9,13,24,8,14,8,29,18,31),
          N=12)

cat("model

vasishth

### R code from vignette source 'recoveringcorrelationsV2.Rnw'

###################################################
### code chunk number 1: recoveringcorrelationsV2.Rnw:98-156
###################################################
new.df <- function(cond1.rt=487, effect.size=123, 
                   sdev=544,
                   sdev.int.subj=160, sdev.slp.subj=195, 
                   rho.u=0.6,
                   nsubj=37,

vasishth

## Our simulated scientist will declare 
## significance only if he/she gets 
## 2 replications with p<0.05:
stringent<-FALSE
## Set the above to FALSE if you want to 
## have the scientist publish a single
## expt. as soon as it's significant:
#stringent <- FALSE

## num of scientists to simulate:

vasishth

## store proportion of false positives
## in one lifetime of 200 experiments:
prop_fps<-rep(NA,1000)
## run k=1000 scientists, each with
## a lifetime of 200 experiments:
for(k in 1:1000){
## number of experiments for each scientist:
nexp<-200
## prob of sampling from a population
## where the null is true:

vasishth

### R code from vignette source '04_ADALecture4.Rnw'

###################################################
### code chunk number 1: 04_ADALecture4.Rnw:67-72
###################################################
vpost<-function(v=0.2609^2,n=1,s=0.15^2){
  return(1/((1/v)+n/s))
}
n<-seq(1,1000,by=1)
plot(n,vpost(v=2600,n=n),type="l",ylab="posterior variance",xlab="sample size")

vasishth

### R code from vignette source '03_ADALecture3.Rnw'

###################################################
### code chunk number 1: 03_ADALecture3.Rnw:85-92
###################################################
data<-read.table("~/Git/Statistics-lecture-notes-Potsdam/AdvancedDataAnalysis/data/gibsonwu2012data.txt",header=T)
## take reciprocal rt to normalize residuals:
data$rrt<- -1000/data$rt
## define predictor x:
data$x <- ifelse(

vasishth

### R code from vignette source '02_ADALecture2.Rnw'

###################################################
### code chunk number 1: 02_ADALecture2.Rnw:66-72
###################################################
library(mvtnorm)
u0 <- u1 <- seq(from = -3, to = 3, length.out = 30)
Sigma1<-diag(2)
f <- function(u0, u1) dmvnorm(cbind(u0, u1), mean = c(0, 0),sigma = Sigma1)
z <- outer(u0, u1, FUN = f)

vasishth

### R code from vignette source '01_ADAIntroLecture.Rnw'

###################################################
### code chunk number 1: 01_ADAIntroLecture.Rnw:134-141
###################################################
## load data:
data<-read.table("~/Git/Statistics-lecture-notes-Potsdam/AdvancedDataAnalysis/data/gibsonwu2012data.txt",header=T)
## take reciprocal rt to normalize residuals:
data$rrt<- -1000/data$rt
## define predictor x:

vasishth

### R code from vignette source 'recoveringcorrelations.Rnw'

###################################################
### code chunk number 1: recoveringcorrelations.Rnw:93-151
###################################################
new.df <- function(cond1.rt=487, effect.size=123,
                   sdev=544,
                   sdev.int.subj=160, sdev.slp.subj=195,
                   rho.u=0.6,
                   nsubj=37,

vasishth

library(MASS)
new.df <- function(cond1.rt=600, effect.size=10, sdev=40,
                   sdev.int.subj=10, sdev.slp.subj=10, 
                   rho.u=0.6,
                   nsubj=10,
                   sdev.int.items=10, sdev.slp.items=10, 
                   rho.w=0.6,
                   nitems=10) {

  ncond <- 2