Implicit Associations in Computer-Mediated Communication

Abstract

Abstract: The inclusion of skin-tone modifiers into the standard emoji set in 2015 marked a shift toward increased skin-tone representation in emoji characters. We investigated whether implicit skin-tone bias, as typically found for human faces, also exists for emojis – a topic that is becoming increasingly relevant in computer-mediated communication (CMC) as more of our communication, education, and social exchange takes place in digital spaces. We systematically adapted Harvard University’s most recent skin-tone IAT (Implicit Associations Test) to assess implicit skin-tone bias for emojis. The reliability of our novel skin-tone IAT was good (internal consistency = .87). Data from a racially and ethnically diverse sample of 248 participants revealed that, on average, participants held more positive implicit associations for light than for dark skin-tone emojis: MDscore = 0.39; SDDscore = 0.42; t(247) = 14.76, p < .001; Cohen’s d = 0.93. Participants’ own skin tone predicted the strength of skin-tone bias, R2 = .02, F(1, 247) = 5.41, p = .02. The darker the skin tone of the participants, the less likely they were to exhibit bias in favor of light skin-tone emojis (β = −.14). These results align with the patterns of skin-tone bias that are typically observed for human faces, and extend those insights to emojis that are frequently used in CMC. The results also provide the first evidence that our novel emoji skin-tone IAT can be a useful tool for assessing emoji skin-tone bias.