From the world to the lab and back again is a phrase the LSE's Department of Psychological and Behavioural Science (PBS) proudly uses to describe our approach to research and its application in the real world. In this “lab-field” research, Matteo Galizzi and Daniel Navarro-Martinez have questioned an approach used widely in behavioural and experimental economics in understanding social preferences - games conducted in the lab, presented as predictors of real-world behaviours. They found that behaviours in these lab games poorly explain our behaviours in life.
This research draws on the extensive literature based on experiments conducted in lab settings that use popular “social preference” games such as the Dictator Game, the Ultimatum Game, the Trust Game, and the Public Good Game to understand social behaviours. Galizzi and Navarro-Martinez link and compare these lab experiments with participants’ behaviour out into the field and with their own self-perception. First, they conducted several social preference games in a lab setting (detailed below*); second, they observed the same participants in real-life pro-social behaviour situations, such as asking participants to help carry boxes, or to give money to charities; and third, they asked participants to report on their own past pro-social behaviours, for instance “I have helped a stranger” or “I have given money to a charity”. A distinguishing feature of the study is that it systematically and transparently reports all the associations between the pro-social situations in the field and the social preferences games in the lab (rather than reporting just a “cherry-picked” subset of the correlations).
Galizzi and Navarro-Martinez found that social preference identified from the lab games do not predict pro-social behaviour occurring in natural life situations. Only one lab-field correlation out of forty is statistically significant – none is when the analysis corrects for multiple hypothesis testing. Galizzi and Navarro-Martinez also conducted a systematic review and meta-analysis of all the published and unpublished studies that have previously reported data on the “external validity” of the same social preferences games: only about one-third of all the reported lab-field correlations is statistically significant.
One notable conclusion is the need of a systematic approach to analysing the external validity of social preferences games. Another dilemma is that, in order to measure social preferences, those games will always lack context and will have a large element of artificiality. So while social preference games in the lab do have some validity and help researchers, for example, create a controlled environment and study how subjects play those games, they do not necessarily translate easily and accurately outside lab settings where pro-social behaviour occurs within a context. Therefore, Galizzi and Navarro-Martinez suggest that this approach needs reexamining, perhaps bringing more context to social preferences lab games. As pointed out by Harrison and List (2008), in fact, “it is not the case that abstract, context-free experiments provide more general findings if the context itself is relevant to the performance of the subjects” (p. 840).
This summary is based on the article:
Galizzi, Matteo M. and Navarro-Martínez, Daniel (2017) On the external validity of social preference games: a systematic lab-field study. Management Science. ISSN 0025-1909.
*What are social preference games?
Participants played each of the seven games below, each only once to more accurately reflect the real-life scenarios:
Two-player game in which Player 1 decides how to divide £10 between the self and Player 2. Player 2 simply receives the allocation established by Player 1. Half of the participants were Player 1 and the other half Player 2.
Like Dictator Game 1, but switching the roles (and matching people with different partners).
Two-player game in which Player 1 decides how to divide £10 between him/her and Player 2. Player 2 decides whether to accept the allocation or not. If the allocation is rejected, both players get nothing. Half of the participants were Player 1 and the other half Player 2.
Like Ultimatum Game 1, but all the participants were player 2 and all of them had to respond to the same allocation of £5 for player 2, which was determined by a participant wjo was player 1 in a preliminary pilot session.
Two-player game in which player 1 has an endowment of £10 and decides how much of it to send over to player 2. The amount sent over is multiplied by three and given to player 2, who has to decide how much of it to send back to player 1. Half of the participants were player 1 and the other half player 2.
Like Trust Game 1, but all the participants were player 2 and all of them had to repond to the same amount of £5 sent over by player 1, which was determined by a participant who was player 1 in a preliminary session.
Four-player game in which all the players have an endowment of £10 and have to decide simultaneously how much of it to contribute to a common group fund. The overall money in the group fund is then multiplied by two and split between four players.