In contrast to what some economist today still say and believe, economics is an experimental science. At the latest, when the Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel was award to Daniel Kahemann (a psychologist) and Vernon Smith (an economist) in 2002 they should have acknowledged it.
Economic experiments have been proven useful in informing theory and testing (new) economic institutions before their implementation on a broader scale, e.g. the design of spectrum auctions that generated unprecedented revenues for the states running them. Unfortunately even within the community of experimental economist their use and purpose is not without controversy. Some, let’s call them experimental economists in a narrower sense see the main use of experiments in economics in showing that the theory works (well) and finding instances of when it works best. The other group, let’s call them behavioral economists see the economic experiment as one method to investigate the underlying assumptions of economic theory in order to inform theory building and inspire the revision of economic theories so that they may move more towards a positive than a normative model of the world.
With Experimental Economics a group of six British experimental economists now tried to critically assess the current state of the field that constitutes an invaluable tool for research in all areas of economics.
In a series of chapters they address the method and methodology of experimental economics, the domain of economic theory (where and when does it apply?) and the limits of experimental tests in terms of what can be said about the theory and the external validity of the experimental observations; and also how experiments are used as rhetorical devices, “exhibits” that reliably show some particular behavior of their participants illustrating a specific point. Two further chapters address the important issue of financial incentives in experiments (when are they needed, how should they be implemented?) and different sources of noise in the data that requires bespoke statistical treatment.
The last point, noise in the data and heterogeneity between subjects is in my opinion a very important one as this is still often a neglected topic in most experimental studies today. Of course, a well designed experiment may allow the authors to show their main point without any fancy statistics. On the other hand, in order to move to a positive theory of economic behavior the individual and not the aggregate behavior should be the focus of the analysis. This necessarily requires a more advanced statistical treatment of the data. As well as laboratory and field experiments (and happenstance data) are complements so are theory, experiments, and statistics complements.
In sum, even though I may not agree with some of the more specific points Bardsley, Cubitt, Loomes, Moffatt, Starmer, and Sugden make their book is an excellent text that will make it on the reading list for my courses in experimental economics.