Causality in statistics and econometrics

Causal problems are an important part of decision and measure analysis. Besides philosophers, econometricians are getting interested in causal effects. Heterogeneity and nonexperimental features of data unfortunately make causal estimates more complicated. Cause-consequence relationships are usually depicted in causal diagrams, which lead to some theoretical criteria. This thesis is focused on causality in cross-sectional data and the most applied methods such as conditioning, instrumental variables and mechanisms of causal exposure are demonstrated. The estimators based on these approaches are unbiased under a quite restrictive set of conditions. In the last chapter of this thesis, some examples of incorrect use of causal techniques are shown including their reasons and impacts on the biasedness of calculated estimators.