This course will introduce a variety of scientific/statistical issues in causal inference. The concept of causal inference is not new. In fact, we are using the concept of causal inference daily perhaps without realizing it. Without the use of causal inference, we would not have survived today. However, causal inference has not been extensively discussed in standard statistics textbooks. One possible reason is that foundation of statistics is based on probability theory, which focuses on parameters of a joint distribution from a sample drawn from a population. These parameters permit researchers to estimate probabilities of past and future events and update those probabilities in light of new information under the assumption that experimental conditions remain the same during the duration of experiment. However, causal inference goes one step further before these association probabilities to infer probabilities under changing experimental conditions。
In this course, I will first discuss the foundation for causal inference, based on potential outcomes and give mathematical definitions for causal parameters of interest. Then, I will discuss statistical issues in causal inference in randomized clinical trials, and then move on discussing statistical issues in causal inference in observational studies and how to make credible causal inference in absence of randomization.
