In AmandaPMurphy/SemesterProjectLabII:

knitr::opts_chunk$set(echo = TRUE)

1. Consider a 2x2 design. Assume the DV is measured from a normal distribution with mean 0, and standard deviation 1. Assume that the main effect of A causes a total shift of .5 standard deviations of the mean between the levels. Assume that level 1 of B is a control, where you expect to measure the standard effect of A. Assume that level 2 of B is an experimental factor intended to reduce the effect of A by .25 standard deviations.

A. create a ggplot2 figure that depicts the expected results from this design (2 points)

library(tidyverse)

df <- data.frame(A = c("A1","A1","A2","A2"),
                 B = c("1","2","1","2"),
                 DV = c(0,0,.5,.25)
                 )

ggplot(df, aes(y = DV, x = B, color = A)) +
  geom_point(stat= "identity", position = "dodge")

# N per group
N <- 120

A_pvalue <- c()
B_pvalue <- c()
AB_pvalue <- c()
for(i in 1:1000){
  IVA <- rep(rep(c("1","2"), each=2),N)
  IVB <- rep(rep(c("1","2"), 2),N)
  DV <- c(replicate(N,c(rnorm(1,0,1), # means A1B1
                        rnorm(1,0,1), # means A1B2
                        rnorm(1,.5,1), # means A2B1
                        rnorm(1,.25,1)  # means A2B2
          )))
  sim_df <- data.frame(IVA,IVB,DV)

  aov_results <- summary(aov(DV~IVA*IVB, sim_df))
  A_pvalue[i]<-aov_results[[1]]$`Pr(>F)`[1]
  B_pvalue[i]<-aov_results[[1]]$`Pr(>F)`[2]
  AB_pvalue[i]<-aov_results[[1]]$`Pr(>F)`[3]
}

length(A_pvalue[A_pvalue<0.05])/1000
length(B_pvalue[B_pvalue<0.05])/1000
length(AB_pvalue[AB_pvalue<0.05])/1000

AmandaPMurphy/SemesterProjectLabII documentation built on May 23, 2022, 3:16 p.m.