Lakens

#Install pwr package if needed
if(!require(pwr)){install.packages('pwr')}
library(pwr)

alpha_level = 0.05 #set alpha level
n = 100 #set number of observations
st_dev = 1 #set true standard deviation
SESOI <- 0.5 #set smallest effect size of interest (raw mean difference)

# calculate lower and upper critical values c_l and c_u

Lakens

#Load packages
library(readxl)
library(mailR)

#Read student data
info <- read_excel("student_names_email.xls",
                   sheet = 1,
                   col_names = TRUE)

#Loop from 1 to the number of email addresses in the spreadsheet

Lakens

---
title: "Observed Alpha Levels"
output:
  word_document: default
  html_document:
    df_print: paged
editor_options:
  chunk_output_type: console
---

Lakens

# Code by Chelsea Parlett, small additions by Daniel Lakens
library(pwr)
library(ggplot2)

#set up vector of effect sizes
es <- seq(0.01,1, length = 1000)
#specify power for test 
pow <- 0.8

#calculate ns (sample size for 80% power in two-sided test)

Lakens

optimal_alpha <- function(power_function, costT1T2 = 1, prior_H1H0 = 1, error = "minimal") {
  #Define the function to be minimized
  f = function(x, power_function, costT1T2 = 1, prior_H1H0 = 1, error = "minimal") {
    y <- 1 - eval(parse(text=paste(power_function)))
    print(c(x, y, x+y)) #optional: print alpha, beta, and objective
    if(error == "balance"){
      max((costT1T2*x - prior_H1H0*y)/(prior_H1H0+1), (prior_H1H0*y - costT1T2*x)/(prior_H1H0+1))
    } else if (error == "minimal"){
      2*(costT1T2*x + prior_H1H0*y)/(prior_H1H0+1)
    }

Lakens

#Standardized p-value
#This is untested alpha software provided with no guarantees - use at your own risk

#Good 1982: P SQRT(N/100)
#Enter the observed p-value and the sample size

p_stan <- function(p, N){
  p * sqrt(N/100)
}

Lakens

if(!require(ggplot2)){install.packages('ggplot2')}
library(ggplot2)

n=20 #set sample size
nSims<-100000 #set number of simulations

x<-rnorm(n = n, mean = 100, sd = 15) #create sample from normal distribution

#95%CI
CIU<-mean(x)+qt(0.975, df = n-1)*sd(x)*sqrt(1/n)

Lakens

require(ggplot2)

#Save downloaded Scopus data in your working directory
scopusdata<-read.csv("scopusPS2010_2015.csv")

plot1<-ggplot(scopusdata, aes(x=Cited.by))  + 
  geom_histogram(colour="#535353", fill="#84D5F0", binwidth=2) +
  xlab("Number of Citations") + ylab("Number of Articles")  + 
  ggtitle("Citation Data for Psychological Science 2011-2015") +
  coord_cartesian(xlim = c(-5, 250))

Lakens

if(!require(meta)){install.packages('meta')}
library(meta)

nSims <- 1000000 #number of simulated experiments
numberstudies<-4 # nSim/numberofstudies should be whole number
p <-numeric(nSims) #set up empty container for all simulated p-values
metapran <-numeric(nSims/numberstudies) #set up empty container for all simulated p-values for random effects MA
metapfix <-numeric(nSims/numberstudies) #set up empty container for all simulated p-values for fixed effects MA
heterog.p<-numeric(nSims/numberstudies) #set up empty container for test for heterogeneity
d <-numeric(nSims) #set up empty container for all simulated d's

Lakens

options(scipen=999) #disable scientific notation for numbers

#Figure 1 & 2 (set to N <- 5000 for Figure 2)
# Set x-axis upper and lower scalepoint (to do: automate)
low_x<--1
high_x<-1
y_max<-2

#Set sample size per group and effect size d (assumes equal sample sizes per group)
N<-50 #sample size per group for indepndent t-test