Main content
Psychometric Evaluation of the Dictator Game and Trust Game
Date created: | Last Updated:
: DOI | ARK
Creating DOI. Please wait...
Category: Project
Description: This study assesses the psychometric properties of the Dictator Game and the Trust Game, two experimental games that are widely used in experimental economics and other areas of social science. The Dictator Game is used as a behavioral measure of generosity toward strangers, and the Trust Game is used as a behavioral measure of trust and trustworthiness toward strangers. Despite their wide use, the validity of these games has never been systematically evaluated. We will conduct an experiment to assess their internal consistency, test-retest reliability, external validity, and convergent validity. Participants (N = 200) will be recruited through the University of Miami psychology student subject pool, through flyers posted around University of Miami, and through Craigslist and flyers posted around South Miami. Participants will play six Dictator Games And twelve Trust Games (six as investor; six as trustee) anonymously with other participants over a computer network. Afterward, they will have three opportunities to donate their earnings from the games to different charities. Participants will return to the laboratory at least two weeks later to complete the same tasks. We will also collect self-reports and up to five peer-reports per participant. Both participants and peers will complete the self-report altruism scale, the trust facet of the NEO-PI-R, and a modified version of the Propensity to Trust scale. Participants will also complete life history measures and a shortened version of the Big-Five inventory. We will derive a measure of response acquiescence by positively loading each item on the Big-Five Inventory on one factor. We will conduct an Exploratory Factor analysis, internal consistency analyses, test-retest correlations, correlations of aggregate game measures of generosity, trust, and trustworthiness with non-game generosity toward strangers, statistically controlling for the amounts in participants’ accounts at the times of the three donation decisions. We will also correlate aggregate game measures with self-reports of generosity, trust, trustworthiness, life history, and response acquiescence. Finally, we will correlate aggregate game measures with peer-reports of generosity, trust, and trustworthiness. We will use non-parametric tests if the observed distributions violate normality. We achieve 82% power (assuming a two-tailed test, p < .05) for detecting correlations of r = 0.20 when N reaches 200. To date (10/29), we have run 23 participants.