YoVDO

Validity and Bias in Epidemiology

Offered By: Imperial College London via Coursera

Tags

Epidemiology Courses Causal Inference Courses

Course Description

Overview

Epidemiological studies can provide valuable insights about the frequency of a disease, its potential causes and the effectiveness of available treatments. Selecting an appropriate study design can take you a long way when trying to answer such a question. However, this is by no means enough. A study can yield biased results for many different reasons. This course offers an introduction to some of these factors and provides guidance on how to deal with bias in epidemiological research. In this course you will learn about the main types of bias and what effect they might have on your study findings. You will then focus on the concept of confounding and you will explore various methods to identify and control for confounding in different study designs. In the last module of this course we will discuss the phenomenon of effect modification, which is key to understanding and interpreting study results. We will finish the course with a broader discussion of causality in epidemiology and we will highlight how you can utilise all the tools that you have learnt to decide whether your findings indicate a true association and if this can be considered causal.

Syllabus

  • Module 1: Introduction to Validity and Bias
    • Every time you conduct a study, the most important questions to ask are whether your results are an accurate reflection of the truth both within your sample and in the broader population of interest. This is called validity of the study and more or less determines if your study is of any value. In this module we will discuss what validity actually means and we will describe the different types of systematic error, or bias that may undermine the validity of a study. You will learn how to identify and prevent selection bias and information bias and their variations.
  • MODULE 2: Confounding
    • Studies often focus on the association between two variables; for instance, between a risk factor and a disease. However, reality is usually complex and there are many other variables that may influence this association. Sometimes, the presence of a third variable can either exaggerate the association between the two variables we study or mask an underlying true association. This is called confounding and is any researcher’s nightmare. In this module, you will learn multiple methods to detect confounding in a study, so that you can prepare to deal with it. By the end of the module, you will be able to apply these methods to actual data and conclude whether there is confounding.
  • MODULE 3: Dealing with Confounding
    • This module is dedicated to dealing with confounding. Confounding can be addressed either at the design stage, before data is collected, or at the analysis stage. You will learn the main approaches to dealing with confounding and you will see practical examples on how to do this in your own studies. We will also briefly discuss about the Directed Acyclic Graphs, which is a novel way to detect bias and confounding and control for them.
  • MODULE 4: Effect Modification
    • This is the final module of the course. We start by discussing what happens when the effect of an exposure on an outcome differs across levels of another variable. This is called effect modification. We will discuss how to approach effect modification and we will highlight the distinction between confounding and effect modification. We will close the course by revisiting causal inference in epidemiology, discussing how we can go through all potential explanations of an association before deciding whether it is of causal nature.

Taught by

Filippos Filippidis

Tags

Related Courses

Knowledge and Uncertainty
Brilliant
Causal Inference
Columbia University via Coursera
Causal Inference 2
Columbia University via Coursera
Regression Discontinuity Design and Instrumental Variables
Codecademy
Potential Outcomes Framework for Causal Inference
Codecademy