A CEO is unsure whether to allow workers to work from home,…
A CEO is unsure whether to allow workers to work from home, or require them to work form home, so she runs an experiment. People who participate are randomly assigned to treatment or control. In the treatment arm, people will be allowed to work from home, and in the control arm, people will not to be allowed to work from home. Participation in the experiment is voluntary. After the experiment, we have the following data. Group Sample size Average performance score Treatment (optional work from home) 324 4.5 Control (no optional work from home) 325 4.3 Did not participate in experiment 200 4.6 The “did not participate” also does not work from home. Should we use them in our analysis of the effects of working from home?