Paper Number
1285
Paper Type
Complete
Description
Organizations have long been trying to assess job applicants' personality using self-reported psychometric tests, such as the Big Five personality test. However, these tests are not robust against incentives to pretend having certain desirable traits, for example, the disposition for being a good team player. We test whether machine learning classifiers trained on written self-descriptions, such as cover letters, predict people's true cooperativeness better than psychometric tests. Based on data from a controlled online experiment with 400 participants, we find that - when people have incentives to fake their personality - linguistic classifiers based on self-descriptions significantly outperform psychometric classifiers based on the Big Five. Moreover, we find that a fine-tuned, pre-trained natural language model can detect incentives to fake in people's self-descriptions. While further research is needed to achieve tamper-proof models, our findings illustrate the potential of automated personality tests based on job applicants' cover letters.
Recommended Citation
Kecht, Christoph; Kurschilgen, Michael; and Strobel, Magnus, "Revival of the Cover Letter? Experimental Evidence on the Performance of AI-driven Personality Assessments" (2022). ICIS 2022 Proceedings. 1.
https://aisel.aisnet.org/icis2022/data_analytics/data_analytics/1
Revival of the Cover Letter? Experimental Evidence on the Performance of AI-driven Personality Assessments
Organizations have long been trying to assess job applicants' personality using self-reported psychometric tests, such as the Big Five personality test. However, these tests are not robust against incentives to pretend having certain desirable traits, for example, the disposition for being a good team player. We test whether machine learning classifiers trained on written self-descriptions, such as cover letters, predict people's true cooperativeness better than psychometric tests. Based on data from a controlled online experiment with 400 participants, we find that - when people have incentives to fake their personality - linguistic classifiers based on self-descriptions significantly outperform psychometric classifiers based on the Big Five. Moreover, we find that a fine-tuned, pre-trained natural language model can detect incentives to fake in people's self-descriptions. While further research is needed to achieve tamper-proof models, our findings illustrate the potential of automated personality tests based on job applicants' cover letters.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.
Comments
13-DataAnalytics