8.3 Overlap/common support

• Q: Inspect Figure 8.1 below: Are there areas in the distribution below where we don’t have common support across treatment/control focusing on the covariate education?
  • Find a unit (ID = ?) in the control group that has no exact equivalent in the treatment group.
  • Q: Assume you increase the number of variables (covariates) on which you match. Does that make it more or less difficult to find matches? Exact vs. inexact matching?
• Q: Do you have experiences with matching? What problems did you encounter?

Figure 8.1: Joint distribution: Threat (D), Education (X1) and Age (X2)