This project can be of several types, instructions for that below.
You can choose to do a field experiment, a lab experiment, work on a natural experiment, run simulations or even do a thought experiment. First thing to remember, almost everything is a game. The goal is to find one and analyze it. The standard case would be a) an interesting story b) translate story to game c) solve game using standard theory d) let people play and describe what they do e) try to explain why they deviated from theory, if they did f) include a calculation of a rational benchmark (e.g. Nash equilibrium if it’s a static game). Not everyone has to stick with the standard case. Example: Some people feel comfortable with complicated game theory but don't feel they have access to many subjects. For them it could be best to devise a complicated game, solve it and explain how they think people would behave. Then run a small pilot with the subjects they can find. Example 2: Some people love statistics but are not deeply in love with game theory. They can find that a very simple game is actually being played in real life. Analyze briefly the (very simple game). Get data from field/natural experiments and work heavily on the stats side.
OR b) A replication of an existing experiment You should have a reason for doing the replication!