Sunday, January 20, 2008

"There is often value in obtaining another field's perspective on what one does. Some EEs have not been particularly reticent about providing BEs with such input (see, e.g., Smith, 1991). In this essay I attempt to return the favour in a small way...

[Footnote:] It is a curious fact that undergraduate economics majors, unlike psychology majors, rarely take courses in empirical research methods. As a result, while they may be well-tooled in regression techniques, they are typically at a complete loss when it comes to other aspects of empirical research...

Efficiency, however, is not the same as high external validity. Double oral auctions or even one-sided auctions, not to mention short-selling, are rare in daily life. I do not think I have ever encountered a double oral auction outside of participating in an experiment, and the last time I participated in any type of auction was as a teenager when I bought a broken washing machine motor for $0.25. [Ed: This was written in 1999.]

... Besides exaggerating the degree of learning that takes place in real-world settings, stationary replication can also affect people's preferences in a way that may or may not enhance external validity. Repetition tends to repress certain types of psychological motives, such as fairness, that may play a prominent role in early-period play. How many times can a subject get angry about someone splitting a pie unevenly? It must be acknowledged, however, that similar factors may be operative in daily life: how many trips to the wine store does it take before one forgets one's resolution to punish the French for violating the nuclear test ban treaty?...

[Footnote] Because context cannot be eliminated, experiments should never be used for the purpose of measuring individual propensities. It is tempting, but a big mistake, to think that behaviour in dictator games measures an individual's altruism, that responders' behaviour in ultimatum games measures their taste for fairness, or that behaviour in the trust game measures trustworthiness. I am aware of no evidence showing that people who give more money in dictator games contribute more to charities, are more likely to put themselves at risk to rescue a drowning stranger, or give their subway seat to the elderly or infirm. But, even if such data are collected and the correlations prove to be positive, the fact remains that it would be easy, through a suitable manipulation of context, to design a dictator study in which people would give none of their money to a stranger, or one in which most people would give all of their money away. Which of these contexts uncovers people's `true' level of altruism? Some EEs seem to believe that they know the answer: whatever context gives results that are closest to the standard economic model...

Some EEs seem to believe that some of these motives can be eliminated through procedures that assure anonymity... the idea that people directly `consume' (that is, care about) their reputations is not only eminently reasonable, but consistent with a myriad of studies (see Bodner and Prelec, 1996). For example, people behave very differently in the presence of a mirror, even when they believe that no one is observing them (Duval and Wicklund, 1972)...

Even if experimenters were able to eliminate motives other than profit maximisation from their experiments, it would not necessarily be a good thing insofar as external validity is concerned. Profit maximisation may be an important incentive in economic transactions, but it probably is not the most important. Anyone working in a business setting, for example, can attest to the power of social comparison. Most academics seem to be motivated more by ego than by the size of their salary, and academics probably are not exceptional in this regard. Even when people do care about money, their degree of concern is often remarkably unrelated to the amount of money involved. Small amounts of money can gain momentous significance under certain circumstances ± for example, if one gets an undeserved speeding or parking ticket, or if one is unfairly denied a small year-end bonus. Much larger changes in wealth, such as those which result from a change in stock prices, can leave one remarkably unmoved."

--- Experimental Economics from the vantagepoint of Behavioural Economics, George Loewenstein