ABSTRACT
If we talk about the effect of a covariate Xj on an outcome variable Y in common language, we mean something like that a difference between two subjects in this covariate is or may be responsible for some difference in the outcome variable. We often express this in the way that differences in Xj have caused or are causal for differences in Y . For example, if we observe-as in Section 4.1-that with increasing amount of smoking the blood pressure increases, we would like to conclude that smoking a lot causes high blood pressure.
