Title

Abstract

Representative Design

Representative design is a term, first used by Egon Brunswik, used to describe experimental or survey methods in which both the samples and the situations tested are representative of the ecology (or wider-world circumstances) to which the results are to be generalized.  I will first give a rigorous definition of representative design and describe the research context in which it was first used.  I shall then consider a wider range of examples of what representative design means in practice, drawing on my own research and that of others.  As I will observe, although representative design is an ideal toward which we should strive, those ideals often have to be traded off against other research requirements.  There are circumstances in which representative design should be considered as essential, when it can be useful, and when it is unnecessary and perhaps even a hindrance to good research.  I will conclude by arguing that representative design should be considered as part of a triumvirate approach, in which a representative sample of situations, persons and methods are used to test phenomena.  Moreover, this approach should be seen as the goal of the entire scientific process, and not merely of individual studies. 

Title

Abstract

The Experimental Method and Rational Economic Man: A perspective from Economic Psychology

Behavioural economists have enthusiastically embraced experimental methods but they have done so in ways that would surprise most psychologists. An example is the practice of specifically asking participants to maximise their income or wealth and when results suggest less instrumental motives, findings are viewed with suspicion. Results from a questionnaire study of tax compliance are reported which reveal important cultural and individual differences in willingness to act instrumentally.

Title

Abstract

Lessons from Social Psychology for New Experimental Disciplines

Experimental social psychology has had the benefit of over 50 years of designing experiments to capture often elusive aspects of human thought and behaviour.  As disciplines such as economics, sociology, and anthropology start to seek experimental answers to behavioural questions, it is important to learn from both advances and pending problems in experimental social psychology.  I will discuss internal validity (e.g., confounding factors in manipulations); external validity (e.g., reliance on university student populations), and ethics (e.g., justifiability of experimental deception).  I also hope to open a dialogue with members of other disciplines about their concerns and needs regarding experimental methods.

Title

Abstract

On Testing Game Theory

In recent years there has been considerable growth in the use of experimental methods in economics research. A common strategy in such research is to use economic theory to derive ‘predictions’ and to report the behaviour of experimental subjects as tests of them. Amongst consumers of this research,  however,  there is disagreement about whether relevant propositions in economics are testable at all; and amongst those who think they are, there is disagreement about what the relevant tests show. In this paper, I explore what is involved, from a methodological point of view, in alleged tests of economic theory. Using examples from game theory, I aim to shed light on the source of disputes about what experiments test, and to contribute to broader debates about the appropriate role of experiments in the development of economic theory.

Title

Abstract

How Close Should Experiments be to Theoretical Models and Why?

The stock defence of experimental economists to scepticism over external validity argues from a premise that similarity between the experiments and models is appropriate for theory-testing, to a conclusion that external validity considerations do not arise. I argue that this defence fails for applied economics experiments because model-implementing designs are not, in fact, appropriate for theory-testing given an improved understanding of the relationship between an economic theory and a model. The immediate implication is not that the designs lack external validity, but that the question cannot be dismissed. However, the reductionism involved in implementing applied economics models makes this question awkward for many designs, in part for reasons embraced by constructivists. These considerations call for a less model-centred approach to laboratory experiments than is often taken by economists. To test a theory, for example, requires principally that the theory makes a prediction in the lab. But this means that the phenomena picked out by the theory have to be implemented, not an economic model. This also implies that, for some experiments but by no means all, monetary incentives are desirable.

Available downloads