bgall

library(dplyr)


# Wrap data simulation in a function
sim_data <- function(N) {
  ###########################################################################
  # parameters
  ###########################################################################
  
  set.seed(123)

bgall

# Compare the performance of two purportedly-fast ways of generating dummy variables from character vector.
# NOTE: fastDummies retains the original variable by default while modeldb does not

# Dependencies
library(microbenchmark)
library(fastDummies)
library(modeldb)
library(dplyr)

# Simulate data: 1 million rows, 1 variable with 26 unique values

bgall

# Show that social desirability not only produces biased
# "descriptive statistics" (e.g. group means), but can
# produce treatment effects of the opposite sign of
# the true effect. You cannot simply ignore social
# desirability under the assumption that it does
# not affect estimates of causal parameters since
# measures pre-treatment and post-treatment are
# both subject to social desirability.
#
# Note that this arises due to social desirability

bgall

set.seed(123)

##############################################
# Study parameters
##############################################

# Sample size
n <- 63

# Mean of outcome

bgall

################################################
# Define function: create_predictors
# Creates randomly generated data sets containing
# all design variables, attributes, and dummy
# variables for the right-hand side of the conjoint
# analysis (excludes the outcome variable),
# based on specified data parameter values.
# Data are "long," sucht hat each row of the data 
# is a conjoint profile.
#

bgall

#########################################################
# Define function: create_value_dummies
# Create dummy variables indicating if a given
# profile takes on a specific value from its 
# possible values. Create a dummy for the range
# of the variable, independent of whether the actual
# random sampling does (not) draw that value and assign
# it to a profile.
#########################################################

bgall

##############################################################
# Define function: create_attributes
#
# Creates randomly generated vectors of values from sets
# of potential values for a specified number of attributes
#
# *Arguments*
#
# attr_names (optional)
# vector of attribute names of attr_n length. If no values

bgall

#########################################################
# Define function: create_design_df
#
# Description: Generates a data frame with participant
# IDs, choice set ID, profile ID,  for specified numbers of 
# participants, choice tasks per participant, and profiles
# per choice tasks
#
# Arguments:
# N = # of participants

bgall

#########################################################################
# 'loadpkg' function
# Checks if a vector of packages are installed. If not, installs the
# package. Then loads all packages in vector.
#########################################################################

loadpkg <- function(toLoad){
  for(lib in toLoad){
    if(! lib %in% installed.packages()[,1]) {
      install.packages(lib, repos='http://cran.rstudio.com/')

bgall

#########################################################################################
# This gist contains a quick walk-through of several ways to produce scales capturing
# the average value of one or more variables. Each row (observation) gets its own
# value. We'll assume your data are not fully "tidy." What I mean by this is that you
# have an observation for each row and you want to calculate that observation's value
# on the scale, but each variable that should go into your scale is in its own column.
#########################################################################################

#########################################################################################
# Set-up (packages, fake data, etc.)