Hypotheses in applied microeconomics take roughly two forms. Sometimes we hypothesize that some event, policy, situation, etc. (e.g., being assigned to the quad) changes the costs and benefits associated with a decision (participating in distant activities). Empirically, we test these hypotheses by regressing our measure of the decision (the Y -- participation) on the metric for the event, policy, or situation (the X -- being assigned to the quad) and whatever other controls we deem necessary. Other times, we hypothesize that some policy or decision (like attending Harvard) affects some interesting outcome (like earnings), and we want to understand if this effect is present (and its magnitude). We test these types of hypotheses by regressing the outcome (higher earnings) on the decision (attending a selective college) and whatever other controls we deem necessary.
These are the most typical formats for empirical work in applied microeconomics. As noted in class and in the "My First Paper" documents, these simple forms don't produce fully credible estimates unless the X variables are randomly assigned (and other basic criteria like random sampling, accurate measurement of variables, and appropriate functional form are met).
For more details re-read sections 1, 2, and 6 of the "My First Paper"