Group Projects

Preamble

This page is to give some advice about the structure and format of a good experimental economics project. It should be noted at the outset that you do not actually have to implement an experiment – just design one.

Title

You are advised to frame your title in the form of a question that the experiment is intended to answer. This will focus your minds.

Purpose

A project is usually to design an experiment to investigate the validity of some hypothesis. The hypothesis could be purely descriptive (“Are women more risk averse than men”) but this is hardly exciting. I would prefer a test of some economic theory, or of economics combined with psychology or another discipline.

Format and length

I suggest a 25-page double space, font size 11 or 12, document as a guide, but quality is more important than quantity. You should include references to any relevant theory and to any previous experiments in closely related fields. You should include Instructions for the subjects, perhaps screen shots of a possible computer interface, and any relevant tables or figures. Number your pages, and in draft versions, the names of the members of the group.

Structure

I suggest you start with an introductory section briefly outlining your main hypothesis. Then you should have a section on any theory that you are intending to investigate. You do not need to provide a proof of any theoretical propositions, but you should detail and discuss the assumptions the theory makes, and list the main conclusions and hypotheses, emphasising the ones that you intend to test. You should also refer to, and discuss, any previous experiments related to your theory. Rather obviously, your proposed experiment should not simply be a replication of a previous experiment but should be different in some way. Then you need a section discussing, first in general terms, and then specifically, the way you intend to test the theory. Most theories are very general and need specific assumptions to make them implementable, and you should discuss the various possibilities, and your own implementation, saying why you have worked the way you have, and which alternatives there are. One section of the project should discuss various possible extensions and modifications that could be made. You should discuss the use of a pilot experiment to determine the structure of the main experiment. There should be a section detailing how you would use the data that you would get from the experiment in testing the question in your title. Any statistical or econometric tests should be discussed.

Detail

You need to give careful attention to detail about: the number of subjects; the number of different treatments and whether different treatments would be administered in the same session; the number of different parts of the experiment, and the order in which they appear; the number of repetitions of similar problems; the payoff mechanism, and, for example, whether you would pay subjects on all repetitions in the experiment or just on a randomly chosen one; what information you would give the subjects; what random device you would use to implement any random event; what interface (computerised or paper) you would use; how you would collect and process the data, and whether at an individual or an aggregate level; and so on. The more detail the better.