battenr

# Title: Model Diagnostics - Generalized Linear Model (GLM)

# Description: Demonstrating how we can explore model diagnostics 
# using simulated data. For this specific use case, we are going 
# to explore diagnostics for a generalized linear model (glm). 

# Setup ----

#... Libraries ----

battenr

# Title: Average Treatment Effects (ATE) vs Individual Treatment Effects (ITE)

# Description: Demonstrating how it's important to distinguish between individual treatment effects 
# and the average treatment effect. 

# Setup ----

#... Libraries ----

library(tidyverse) # ol faithful

battenr

# Title: Splines for Causal Inference 

# Description: Demonstrating how splines are helpful for causal inference. 
# Specifically, they are helpful when we have a non-linear relationship. 

# The goal with causal inference is to block backdoor paths. To do this, we
# adjust for confounders. We need to model the relationship between the 
# confounder and exposure, between confounder and outcome or both. 

# Sometimes these relationships are non-linear. So what do we do? 

battenr

# Title: Stepwise Selection is a Bad Idea for Causal Inference 

# Description: Demonstrating how using stepwise selection can lead to 
# a biased estimate when the goal is causal inference

# Setup ----

#... Libraries ----

library(tidyverse) # ol' faithful

battenr

# Title: Estimating Survival Curves Using IP Weights

# Description: Demonstrating how we can create survival curves with inverse 
# probability weights for causal inference. 

# Notes: Used the book Causal Inference: What If by Hernan & Robins and accompanying 
# R code for guidance. Thanks to everyone involved for making it freely available!

# Setup ----

battenr

# Title: Same Method, Different Answers - Causal Estimands

# Description: Demonstrating how you can use the same method (IPW), but end up with
# different answers. Each of them is valid, but each answer is for 
# a different question. 

# This code simulates data for two study designs: 
# - An RCT
# - An observational study (RWE)

battenr

# Title: Different Methods of Modelling Survival Data 

# Description: There are a wide variety of methods than can be used, however 
# we need to be careful about what assumptions we are making. 

# The goal of this code is to demonstrate how survival curves can look 
# different depending on different methods. 

# This code simulates data based on a Weibull distribution, then 
# attempts to fit different models to the data, highlighting how 

battenr

# Title: Confounding as a Pathway

# Description: Confounding is sometimes mixed with the term confounder. This
# code demonstrates how we can adjust for either variable on a confounding pathway 
# to reduce bias (assuming there are no colliders)

# Setup ---- 

#... Libraries ----

battenr

# Title: Demonstrating Bootstrapping Using a Visual (gif)

# Description: Demonstrating how bootstrapping works, repeatedly resampling then 
# calculating the statistic of interest. 

# Note: ChatGPT was used to assist with creating this code

# Setup ----

#... Packages ----

battenr

# Title: Descendant of a Collider - To adjust or not adjust? 

# Description: Should we adjust for a variable that is a descendant of a collider? 
# It is commonly accepted that we should not, since it will increase bias. 
# The goal of this code is to test that through simulation. 

# Setup ---- 

#... Libraries ----