JasonSills
commented
2 years ago @CynthiaLteif
Selection bias occurs when participants in the program are different from those who do not participate in the program. We can think of a program attempting to improve volunteer commitment to the mission. This could suffer from selection into the program because those that choose to participate will likely have high levels of commitment. Selection out refers to attrition in the evaluation. In this case suppose we were able to get a non-biased sample to start, but those with high levels of commitment stayed in the program and those with low levels selected out of the program.
To illustrate this point continue on to slides 47 through 55. In slide 47 the tests for selection-into tests for background characteristics between treatment and control. This would let us know if our volunteers that chose to enter the program are different from those in the control group. Slide 49 shows how to test for selection out. Here we are looking to see if the volunteers who remained in the program are in fact the same as the control group.
CynthiaLteif
May you please explain the difference between selection-in and selection out? I'm not sure I fully understand the explanation in the "Testing the validity of the counterfactual" lecture notes slides 21 and 26. @JasonSills