Being nice with the experimenter?

Abstract

This article investigates the determinants of reciprocity towards the experimenters in the lab under a flat-wage scheme. We find that personality attributes – such as agreeableness – help predict the behaviour of the subjects. We additionally propose and assess a general measure of reciprocity.

Keywords:

• flat-wage
• reciprocity
• agreeableness

JEL Classification:

• C91
• J33

Acknowledgements

This research has benefited from discussions with a number of people. We would like to thank Maria Bigoni, Elisa Ciaramelli, Francesco Feri, Fabio Galeotti, Alexia Gaudeul, Alessio Moneta, as well as participants at the SONIC meetings and the 10th Young Economists Workshop on Social Economy at University of Bologna.

Notes

¹ The most widely accepted taxonomy of personality traits is the Big-Five taxonomy, which comprises five factors: openness to experience, conscientiousness, extraversion, agreeableness and neuroticism (also called emotional stability). See Borghans et al. (2008) for a complete discussion.

² In this case, similarly to Englmaier and Leider (2012), we would have needed to vary the fixed wage across treatments.

³ The standard procedure to prime subjects consists of two questions to be answered briefly that ask participants to describe the emotions that the thought of their own death arouses in them and what would happen to them as they physically die. The exact questions to be answered are: • ‘Please describe as specifically as you can the emotions that the thought of your own death arouses in you’  • ‘Please describe as specifically as you can what do you think will happen to you as you physically die’.

⁴ Indeed in most gift-exchange experiments there is a minority of subjects who continuously provide the minimal effort, regardless of the wage rate (Cooper and Kagel, 2009, forthcoming).

⁵ We also checked for other thresholds (such as 4, 3, ..., 0). Our results do not change. We could have also collapsed all positive observations (i.e. Reciprocity Absence > 0) and treat this as a probit (or logit) estimation problem, but doing so will discard information on the behaviour of subjects for the entire experiment with no additional benefits (as no difference emerged).

⁶ In several specifications, these traits turned out to be individually and jointly not significant.

⁷ From column d on, the number of observations reduces to 120 because in the pilot (made with only six subjects) we did not include the general question on reciprocity.

⁸ The question is: ‘How willing are you to take risks, in general?’.

⁹ We only rely on two questions to derive this indicator as our questionnaires were already quite long.

¹⁰ Since subjects answered two similar questions on motivation, we construct an indicator out of these two questions relying on principal factor analysis.

¹¹ Individual math ability has been measured by asking subjects to rate – from 1 (i.e. definitely no) to 5 (i.e. definitely yes) – their agreement with the following statement: In the past I have proved to have good mathematical skills.