In CrumpLab/SemesterProject7709:

knitr::opts_chunk$set(echo = TRUE)

1. pwr function... compute power for anova.

library(pwr)

pwr.anova.test(k = 3,
               n = 10,
               f = .2,
               sig.level = .05,
               power = )

pwr.anova.test(k = 3,
               n = ,
               f = .25,
               sig.level = .05,
               power = .95 )

library(tibble)

levels <- 4
n_per_level <- 10
# repeat the above many times to compute the F-distribution

alternative_data <- tibble(subjects = 1:(levels*n_per_level),
                           IV = as.factor(rep(1:levels, each = n_per_level)),
                            DV = c(rnorm(n_per_level, 0, 1),
                                   rnorm(n_per_level, 0, 1),
                                   rnorm(n_per_level, 1, 1),
                                   rnorm(n_per_level, 0, 1)
                             )
                      )

aov.out <- aov(DV ~ IV, data = alternative_data)

summary(aov.out)

library(effectsize)
eta_squared(aov.out)

make_a_function <- function(){
  return(1)
}

make_a_function()

replicate( 10, make_a_function() )

run_simulation <- function(){

  levels <- 4
  n_per_level <- 10
  # repeat the above many times to compute the F-distribution

  alternative_data <- tibble(subjects = 1:(levels*n_per_level),
                             IV = as.factor(rep(1:levels, each = n_per_level)),
                              DV = c(rnorm(n_per_level, -.2, 1),
                                     rnorm(n_per_level, .2, 1),
                                     rnorm(n_per_level, -.4, 1),
                                     rnorm(n_per_level, .2, 1)
                               )
                        )

  aov.out <- aov(DV ~ IV, data = alternative_data)
  summary_out <- summary(aov.out)
  eta <- eta_squared(aov.out, partial = FALSE)
  return(eta$Eta2)
}

run_simulation()

replicate(100,run_simulation())

mean(replicate(100,run_simulation()))

run_simulation <- function(){

  levels <- 4
  n_per_level <- 10
  num_questions <- 150
  # repeat the above many times to compute the F-distribution

  alternative_data <- tibble(subjects = 1:(levels*n_per_level),
                             IV = as.factor(rep(1:levels, each = n_per_level)),
                              DV = c(rbinom(n_per_level,num_questions,.75)/num_questions,
                                     rbinom(n_per_level,num_questions,.75)/num_questions,
                                     rbinom(n_per_level,num_questions,.75)/num_questions,
                                     rbinom(n_per_level,num_questions,.80)/num_questions
                               )
                        )

  aov.out <- aov(DV ~ IV, data = alternative_data)
  summary_out <- summary(aov.out)
  return(summary_out[[1]]$`Pr(>F)`[1])
}

run_simulation()

ps <- replicate(100,run_simulation())

length(which(ps < .05))/100

summary_out

CrumpLab/SemesterProject7709 documentation built on May 18, 2022, 9:34 p.m.