Author(s)

Yueting Wu, Guanghua Ren

Corresponding Author

Yueting Wu

ABSTRACT

Causal inference is the process of determining the actual, independent effects of a given phenomenon (cause) within a larger system, which is getting more and more attention in the area of sociology, economics and medicine. Judea Pearl said that causal inference is the foundation of scientific research. However, under the framework of counterfactual causality, it is extremely difficult to make causal inferences based on quantitative analysis of survey data on account of endogeneity. Confounders affect both cause and effect, which leads us to draw false conclusions. Fortunately, instrumental variables provide us with a solution. When we find an instrumental variable that only affects the cause but is independent of confounders, we can use this variable to make causal inferences. The choice of instrumental variables determines the validity of causal inference, so wise choice is very important. In this review, we summarize three widely used instrumental variables. Their applications have been analyzed and explained in detail. Their use reveals the most appealing characteristic of instrumental variables: all of them seem to be irrelevant to what we try to explore, but they play a crucial role in finding out the causality among objects of the study. So far, few articles have summarized the three instrumental variables, which has made this review contributory.

KEYWORDS

causal inference, instrumental variable, prescribing preference, gender, SNP