Simulate prosopagnosia data

proso_sim

#devtools::install_github("debruine/faux")
library(faux)
library(afex)
library(tidyverse)

## Null data (no effect)

des_null <- check_design(list(task = c("test", "white", "asian")), 
                         n = 400, 
                         mu = c(100, 100, 90), 
                         sd = c(10, 10, 10), 
                         r = .6)

sim_null <- function() {
  data <- sim_design(des_null, plot = FALSE) %>%
    mutate(proso = test < mean(test) - 2*sd(test)) %>%
    gather(face_eth, score, white, asian)
  
  suppressMessages(
    aov_ez(id = "id", between = "proso", within = "face_eth", 
           dv = "score", data = data, return = "aov") %>% 
      broom::tidy()
  )
}

sim_dat_null <- map_df(1:100, ~sim_null())

sim_dat_null %>%
  group_by(term) %>%
  summarise(power = mean(p.value < .05))



## Experimental hypothesis (ixn between proso and face_eth)

within <- list(face_eth = c("white", "asian"))
between <- list(proso = c("proso", "control"))

des_exp <- check_design(within, between, n = list(proso = 15, control = 385),
                        mu = list(proso = c(70, 40),
                                  control = c(100, 90)),
                        sd = 10,
                        r = .6,
                        dv = "score")

sim_exp <- function() {
  data <- sim_design(des_exp, plot = FALSE, long = TRUE)
  
  suppressMessages(
    aov_ez(id = "id", between = "proso", within = "face_eth", 
           dv = "score", data = data, return = "aov") %>% 
      broom::tidy()
  )
}

sim_dat_exp <- map_df(1:100, ~sim_null())

sim_dat_exp %>%
  group_by(term) %>%
  summarise(power = mean(p.value < .05))