Advanced Topics in Causal Research

At the end of the course, you'll should be able to:

• understand the concept of confounding
• understand the different methods to determine whether there is confounding and to adjust for confounding, and is able to apply these methods in a computer practice
• understand more advanced methods to adjust for confounding namely propensity score method and instrumental variable method, and is able to apply these methods in a computer practice
• understand the usefulness of sensitivity analysis to estimate the impact of unmeasured confounding, and is able to apply these methods in a computer practice
• understand the concept of effect modification
• understand the difference between effect modification on an additive and a multiplicative scale and is able to calculate effect modification, and its confidence interval, on an additive and a multiplicative scale by hand and by computer
• understand the different ways to present effect modification in a paper and can derive information on effect modification from published studies
• understand the difference between effect modification and confounding

One particular problem in causal research is that the observed relation between a determinant and an outcome is attributed to that determinant, while in fact the observed relation should be attributed to another variable. This form of bias, often referred to as confounding, is central in this course. Other sources of bias, notably missing data and measurement error are not addressed in detail.
Epidemiological research often focuses on quantifying the relationship between a treatment, risk factor, or exposure and health outcomes. The treatment, exposure or risk factor is often referred to as the determinant. The motive of a researcher is often to find out whether the observed relation between the determinant and the outcome is causal. This course deals with this type of research.