Our online profiles are more than just representations of our professional lives — they also appear to be windows into our personalities. A new study, published in the Journal of Occupational and Organizational Psychology, reveals that LinkedIn profiles might offer significant insights into individual traits like narcissism and intelligence. This research suggests that machine learning algorithms can accurately infer these traits from the information available on LinkedIn, potentially offering a powerful tool for recruiters and organizations.
LinkedIn has grown into the most popular online professional network, with nearly a billion users globally. Recruiters and employers routinely use LinkedIn profiles to gauge potential hires, attempting to infer qualities such as professionalism, competence, and even personality traits. Despite its widespread use, there has been considerable debate over how accurately LinkedIn profiles can reflect a person’s traits. Previous research has produced mixed results, leaving it unclear whether LinkedIn provides reliable signals about individual traits.
This study sought to clarify LinkedIn’s predictive potential by using machine learning to analyze LinkedIn profiles. The researchers aimed to determine whether LinkedIn contains valid cues that could accurately reflect traits like narcissism and intelligence. They also wanted to see if machine learning algorithms, acting as “automated perceivers,” could outperform human recruiters in assessing these traits consistently and accurately.
“I’ve long been intrigued by how much our social media presence can reveal about us, especially when it comes to stable traits like personality,” said study author Tobias M. Härtel, a tandem professor at the Osnabrück University of Applied Sciences and a people analytics specialist at BASF.
“It’s fascinating how platforms like Facebook can be used to predict things like personality traits from likes and status updates. But using that kind of personal data in recruitment processes raises a lot of ethical and legal questions. LinkedIn, on the other hand, is specifically designed for professional networking, so it seemed like a better fit for exploring these ideas. Plus, while recruiters often use LinkedIn to gauge a person’s traits, they don’t always get it right. So, I thought — why not see if machine learning can do better?”
To explore LinkedIn’s ability to signal traits, the researchers conducted a detailed analysis of LinkedIn profiles belonging to 406 German-speaking users. The participants, who were recruited online, provided their LinkedIn profiles and completed surveys measuring narcissism and intelligence. Narcissism was assessed using the Narcissistic Admiration and Rivalry Questionnaire, which captures both the self-promotional and defensive aspects of narcissism. Intelligence was measured through tests of fluid intelligence (reasoning and problem-solving) and crystallized intelligence (accumulated knowledge).
The researchers developed a set of 64 LinkedIn profile cues, based on theory and empirical research, that could potentially signal narcissism and intelligence. These cues included straightforward, objective information like the number of listed skills, as well as more subjective elements such as the presence of a smiling profile picture or the use of a background photo.
The study employed machine learning algorithms, specifically elastic nets, to analyze the relationship between these LinkedIn cues and the participants’ narcissism and intelligence scores. Elastic nets are a type of regression model that is particularly effective at handling large sets of interrelated data, making them ideal for this kind of analysis. The algorithms were trained to identify which LinkedIn cues were most predictive of narcissism and intelligence, and their performance was rigorously tested using a method called nested cross-validation to ensure the results were robust and reliable.
The machine learning algorithms demonstrated moderate to strong accuracy in predicting narcissism and intelligence based on LinkedIn profiles, with correlation coefficients of .35 and .41, respectively. These figures suggest that the algorithms were able to identify individuals with above-average levels of narcissism or intelligence about 67.5% to 70.5% of the time — a level of accuracy comparable to that achieved by previous studies using social media data from platforms like Facebook.
The analysis identified several LinkedIn cues that were particularly predictive of these traits. For narcissism, key indicators included having a background picture, listing public speaking skills, and showing fewer smiles in profile pictures. These findings align with the idea that narcissistic individuals often seek to project a grandiose image and are less inclined to display warmth or friendliness in professional settings.
For intelligence, important cues included listing schools with many followers, having a detailed description of educational and professional experiences, and posting accomplishments related to honors and awards. These cues reflect a focus on academic and professional achievements, which are strong indicators of intelligence.
“One big takeaway is that machine learning algorithms are surprisingly good at predicting traits like narcissism and intelligence just from LinkedIn profiles,” Härtel told PsyPost. “They can even be more accurate than coworkers reporting on personality or recruiters assessing applicants’ traits from resumés, LinkedIn profiles, or other social media information. This is because the algorithms consistently pick up on certain patterns, like how narcissists might be more likely to upload background pictures or list public speaking skills, and how intelligent individuals often showcase awards or follow prestigious schools. The study is a nice demonstration of how much information is actually packed into our LinkedIn profiles.”
While this study provides valuable insights, it is not without limitations. One limitation is the static nature of the data used in the study. LinkedIn profiles are dynamic, and users may update their profiles as they gain experience or as they become more familiar with how to present themselves on the platform. This means that the cues identified in this study might evolve over time, requiring continuous updates to the algorithms.
While the machine learning algorithms performed well in this study, the researchers caution against over-reliance on automated systems for making decisions about hiring or promotion. LinkedIn profiles offer only a snapshot of an individual’s professional identity, and while they can provide useful insights, they cannot fully capture the complexity of a person’s traits. For this reason, the researchers suggest that a hybrid approach, combining automated assessments with human judgment, might be the most effective strategy for organizations.
“We do need to be careful about how we use this technology,” Härtel said. “There are definitely ethical implications to consider, especially when it comes to using these predictions in recruitment or other professional settings. Plus, LinkedIn profiles only show a part of who we are, so it’s important to remember that these algorithms aren’t capturing the full picture of someone’s personality. The algorithms are powerful, but they should be seen as a tool to complement human judgment, not replace it. And transparency is key — we need to make sure people understand how these tools are being used and how the decisions are being made.”
The study, “‘LinkedIn, LinkedIn on the screen, who is the greatest and smartest ever seen?’: A machine learning approach using valid LinkedIn cues to predict narcissism and intelligence,” was authored by Tobias M. Härtel, Benedikt A. Schuler, and Mitja D. Back.
