We will assign units to each of the experimental groups randomizing the assignment using a four dimensional blocking. The four dimensions of the blocking are class tier of the property (luxury, full-service, economy), past performance, location (airport or non-airport), and union status (unionized or not).
The performance periods will be three consecutive months (July, August, and September).
We will test our hypotheses by analyzing enrollment performance in absolute terms and as a function of the activity indicator.
We will use diff-in-diffs as the main econometric technique.