Association vs. Prediction vs. Causation#

Association#

  • Two variables are associated means they are correlated in some way, they are not independent. But we don’t know how exactly they affect each other

  • Simply conducted multiple regression may only contribute to association

Prediction#

  • What the outcome will be given the predictor(s). The goal of predictive model is to find the model that minimize the prediction error for future data.

Causation#

  • What would happen to the outcome when we change the predictor(s). The goal of causal inference is to find the best unbiased estimator of the model parameter.

  • Causality subsumes prediction and is more than prediction

From Prediction vs. Causation in Regression Analysis | Statistical Horizons

There are two main uses of multiple regression: prediction and causal analysis. In a prediction study, the goal is to develop a formula for making predictions about the dependent variable, based on the observed values of the independent variables….In a causal analysis, the independent variables are regarded as causes of the dependent variable. The aim of the study is to determine whether a particular independent variable really affects the dependent variable, and to estimate the magnitude of that effect, if any.