Using Valid Cues to Predict Narcissism and Intelligence From LinkedIn Profiles
Journal
Academy of Management Proceedings
Type
conference paper
Date Issued
2023-08-01
Author(s)
Abstract
Recruiters routinely use LinkedIn profiles to infer applicants' key personality traits like narcissism and intelligence. However, little is known about LinkedIn profiles' predictive potential to accurately infer personality. According to Brunswik's lens model, accurate personality inferences depend on (a) the presence of valid cues in LinkedIn profiles containing information about users' personality and (b) the consistent utilization of valid cues. We assessed narcissism (self-report) and intelligence (aptitude tests) in a mixed sample of 406 students/professionals along with 64 deductively derived LinkedIn cues coded by 3 trained coders. Applying nested cross-validated elastic nets, we demonstrate that (a) LinkedIn profiles contain valid information about users' narcissism (e.g., uploading a background picture) and intelligence (e.g., listing many accomplishments). Furthermore, (b) mechanical perceivers like machine learning algorithms use these valid cues consistently so that the elastic nets attained substantial prediction accuracy (r = .28/.32 for narcissism/intelligence). This way, we uncover LinkedIn profiles' potential to accurately infer personality: Personality can be inferred accurately if (a) the valid cues contained in LinkedIn profiles are (b) used consistently like a mechanical perceiver does. The results have practical implications for improving recruiters' accuracy and foreshadow potentials of automated LinkedIn based personality assessments for recruitment purposes.
Keywords
Brunswikian lens model
cybervetting
machine learning
Volume
2023
Number
1
Division(s)
File(s)![Thumbnail Image]()
Name
12425.pdf
Size
428.52 KB
Format
Adobe PDF
Checksum (MD5)
f438eeedb8d70c874db17b6abfae5d6b