Designing Classroom Experiments

Once you have the chance to conduct a classroom experiment and see how effective they are you will probably want to try designing one of your own! In addition to helping you with your teaching there are opportunities to publish your work so that you can share them with others in the economics community.

Elements of a Design a successful lab/experiment

One of the keys to successful classroom experiments is making sure that the experience matches up to the course content that students are trying to master. If this requirement is not met then at best the experiment might be fun, at worst the experiment may be confusing, and in any case the experiment will not help to achieve course goals.

The best classroom experiments share a number of other features:

• They focus on topics students have difficulty mastering
• Otherwise the experiment may not have a clear advantage over other teaching methods
  
• The instructor has selected measurable learning objectives related to the experiment
  
• They focus on one concept or a group of highly related concepts
  
• They motivate discussion or material from future classes and possibly further experiments
  
• Students should have a possibility of being surprised by the outcome, either because it confounds their expectations or because the experiment takes the student to an unfamiliar environment

Please Keep this in Mind...

Never design an experiment to MAKE something happen. Design an experiment where students can make decisions and watch what happens. Here is why:

• Standard economics theories often fail. That is one of the reasons why experimental research in economics has become such a useful research methodology. But often failures of theories are more interesting and can create even better learning opportunities for students. For example, if you let students play a Prisoner's Dilemma game it is unlikely that everyone will choose the equilibrium strategy. This produces interesting opportunities to discuss
• Assumptions of models and what happens when assumptions are unmet
• Strategic differences between one shot and repeated games
  
• Modifications of the game that make it more or less likely that one will observe equilibrium behavior
  

• In order to MAKE something happen you are going to have to constrain students so much it won't feel REAL to them and they won't get the experience we are going for when we right experiments. Then won't get it and that defeats the whole purpose.

Other Practical Tips!

It is a good strategy to write an experiment with another instructor or two. In workshops on classroom experiments a team can typically design a complete draft of an experiment from idea through instructions and teaching notes in about a day.

Test your new experiment with a small group of students (maybe buy them pizza or make it an extracurricular club activity) before you take an experiment to a whole class. It can be hard to tell what points in the instructions will be unclear and if students make unintended decisions because they don't understand the game then the experiment will be disappointing.

Steps in Designing an Experiment

1. Identify a topic that you want to illustrate and some learning objectives you hope that students will achieve.
2. Design an environment where students can make decisions related to the selected topic. It is often not necessary to start from scratch on this point - often the very best classroom experiments are modifications of published research experiments. This means all you need to do is make it practical to conduct the experiment in class!
3. Think about some variations on your decision making environment that you can implement during class. You might want to try one or two variations and leave the others for students to discover. For example, if students play a prisoner's dilemma game repeatedly with different partners you might let the class discover the idea that playing repeatedly with the same partner would yield different results.
4. Think about how you will collect student decisions and how and when you will communicate others' decisions back to them. Class size will be an important factor in determining whether students will make decisions in small groups as well as the form of communication. Some ideas:
• If there are only two possible decisions and there is no strategic advantage in seeing what others do before you make your decision it is often easiest to have students raise their hands.
• Asking students to fill out decision forms which you can collect helps keep student actions private. In some environments students will take different actions if they will become public.
• If calculations involving decisions are involved think about using a pre-formatted Excel spreadsheet in class.
• Sometimes it is helpful to prepare a blank table or graph that you can fill in after decisions are collected.
  
• Make sure that you record the data to preserve it for later discussion and analysis.
  
5. Plan to use easy to obtain and use materials when conducting the experiment. Playing cards, dice, kitchen timers, colored paper, marbles and poker chips are just some of the common items that you may be able to use. For example, many environments involve some form of risk - gaming stores sell dice with a variety of different numbers of sides and these can be used to determine a risky outcome.
6. Write instructions for students. If you are working from an existing research experiment and the instructions are published as part of the paper or available on the author's website that is a great starting point. Otherwise start with the instructions for a similar experiment and modify them to suit your needs.
7. It may help if they answer some "what would happen if" questions before they actually play the game for the first time. These can be used as a pre-class assignment. Sometimes it will be necessary to have parts of the instructions be the same for everyone in the class and other parts to be "private."
8. Write down the series of steps you will follow when conducting the experiment. Guiding student discussion is part of your role as facilitator in the experiment, so some of these steps should involve stimulating student discussion of the experiment. Plan out leading questions to help get the discussion going.
9. Design a post experiment assessment tool to help you assess whether your learning objectives were achieved. Assessment may include questions involving calculation or prediction as well as more open ended questions.