Potential bias when using social media for selection: Differential effects of candidate demographic characteristics, race match, perceived similarity, and profile detail

Abstract

Organizations are using social media as part of their selection processes. However, little is known about whether bias or discrimination is problematic when using these sources. Therefore, we examined whether manipulating the name and photograph of two otherwise equivalent LinkedIn‐like profiles would influence evaluations of candidate qualifications and hireability as well as perceived similarity using an experimental design. To test our hypotheses based on bias/discrimination research and the similarity‐attraction paradigm, a total of 401 working adults were recruited through Mechanical Turk. No evidence was found for bias or discrimination against women or people of color. However, female candidates were viewed as more hireable than male candidates, and Black men were viewed as less qualified than Black women and White men. Furthermore, we found that perceived similarity increased when the participant's gender or race matched the candidate's gender or race, respectively, and also that perceived similarity was related to candidate ratings; however, neither gender nor race match was directly related to candidate ratings. When profiles were more detailed, participants rated candidates of the same race higher than candidates of other races, and perceived similarity fully indirectly mediated this relationship. Conversely, when less detail was provided, participants rated candidates of the same race lower. Thus, while bias/discrimination toward women and people of color is not inherent when using LinkedIn for selection, having a racially diverse set of selectors is important to ensure fairness. This reveals a nuanced view of diversity issues when using social media for selection.

Employer and employee perceptions of cybervetting as a selection method

“…Finally, the current study is descriptive and correlational in nature, and relies on self‐report measures. In this regard, experimental studies manipulating the online profiles of dummy candidates may be helpful in understanding the predictors of selection decisions, as recent work shows that demographic characteristics may be influential in selection decisions (Henderson & Welsh, 2023).…”

“…Similarly, gender has been shown to be a notable factor in hiring decisions (Koch et al, 2015), with even artificial intelligence‐based assessments showing differential scoring between genders (Merritt et al, 2023). Despite this, the impact of gender bias has been considered negligible in Western samples (Henderson & Welsh, 2023), especially when highly structured hiring processes are employed (Fisher et al, 2022). In this context, we considered both gender and organizational position as plausible background variables given cultural differences.…”