Assessing the validity of the Self versus other interest implicit association test

Abstract

There is great variability in the ways that humans treat one another, ranging from extreme compassion (e.g., philanthropy, organ donation) to self-interested cruelty (e.g., theft, murder). What underlies and explains this variability? Past research has primarily examined human prosociality using explicit self-report scales, which are susceptible to self-presentation biases. However, these concerns can be alleviated with the use of implicit attitude tests that assess automatic associations. Here, we introduce and assess the validity of a new test of implicit prosociality–the Self versus Other Interest Implicit Association Test (SOI-IAT)–administered to two samples in pre-registered studies: regular blood donors (Study 1; N = 153) and a nationally representative sample of Americans (Study 2; N = 467). To assess validity, we investigated whether SOI-IAT scores were correlated with explicit measures of prosociality within each sample and compared SOI-IAT scores of the control sample (representative sample of Americans) with the prosocial sample (blood donors). While SOI-IAT scores were higher in the prosocial blood donor sample, SOI-IAT scores were generally uncorrelated with explicit measures and actual prosocial behaviour. Thus, the SOI-IAT may be able to detect group differences in everyday prosociality, but future testing is needed for a more robust validation of the SOI-IAT. These unexpected findings underscore the importance of sharing null and mixed results to fill gaps in the scientific record and highlight the challenges of conducting research on implicit processes.