While we assumed that the above experiment is flawless to answer Q1-3, the experiment is in fact subject to a critical error. If the shoppers of store A are innately heavier milk consumers than the shoppers of store B (for example, if the shoppers of store A purchase two liters per shopping trip while the shoppers of store B purchase 1.5 liter per shopping trip, regardless of the bottle design), the experimental results in Q1 loses internal validity because the difference in sales results not from the difference in the design of the bottle. You can resolve this issue by applying the two-group before-after design. Refer to the Week 3 lecture about the schema of a two-group before-after design (i.e., EG, CG, O1, O2, O3, O4, and X) to answer the following questions.

2-1) What should be EG, CG, O1, O2, O3, O4, and X (10 points)

2-2) Suppose that O2 and O4 are 2 liters and 1.5 liters, respectively (same as in Q1). Suppose further than O1 is 1.8 liters and O3 is 1.2 liters. Does it make sense to replace the current package with the new one? Show your computation (20 points) (Hint: In a two-group before-after design, the effect of new package is computed as (O2 - O1) - (O4 - O3)).