Using machine-learning approach to predict Big-five personality traits based on subconscious behaviors

Date

2024

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

Hiring the right people is the first step for hospitality organizations to enhance service quality and customer satisfaction by obtaining competitive advantages. To find the right people, numerous hospitality companies pay attention to job candidates’ personality traits to evaluate the fit between the companies and job candidates during job interviews. However, evaluations of personality traits can be biased and misunderstood as interviewees often disguise their personalities to create desired impressions. Previous studies in psychology have suggested that people’s unique characteristics can be inferred by their subconscious communication types, that is, verbal, para-verbal, and non-verbal behaviors. Nevertheless, it is yet to be tested how three subconscious behaviors are associated with personality traits, thus influencing hiring decisions made by interviewers. Thus, the primary purpose of this study is to investigate the relationships between interviewees’ three communication types, Big-five personality traits, and interviewers’ hiring decisions in the job interview context. Moreover, Big-five personality traits are measured by three types of Big-five personality traits assessments: interviewer-reported, self-reported, and machine learning-inferred Big-five personality traits, to capture various aspects of personality traits assessment. To test the proposed hypotheses, 10 mock-up interviewers and 106 mock-up interviewees were recruited from Prolific and one of the universities in mid-west areas of US, using four screening conditions: (1) 18 years or older, (2) currently living in the United States, (3) seeking supervisory job opportunities in the hospitality industry, and (4) having LinkedIn personal page, for mock-up interviews. Four machine learning-based techniques were employed to measure interviewees’ verbal, para-verbal, and non-verbal behaviors. Confirmatory factor analysis was performed to verify the validity and reliability of interviewer-reported and self-reported Big-five personality traits. Regression analyses were employed to test the proposed hypotheses. Regarding the relationships between three communication types, interviewer-reported Big-five personality traits, and hiring decisions, verbal behaviors had significant effects on extraversion (β = .364, p < .01), openness (β = .227, p < .01), and agreeableness (β = .169, p < .01) traits, while para-verbal behaviors were significantly associated with openness (β = .108, p < .01) and consciousness (β = .090, p < .01) traits. Non-verbal behaviors had significant impacts on openness (β = .079, p < .05), agreeableness (β = .083, p < .01), and conscientiousness (β = .104, p < .01) traits. Among five dimensions of interviewer-reported Big-five personality traits, neuroticism (β = -.406, p < .01), extraversion (β = .270, p < .01), agreeableness (β = .327, p < .01) were significantly associated with hiring decisions. In terms of the associations between three communication types, self-reported Big-five personality traits, and hiring decisions. both para-verbal behaviors (β = -.096, p < .01) and nonverbal behaviors (β = .045, p < .05) were significantly related to an openness trait. Extraversion (β = .288, p < .05) and conscientiousness (β = .486, p < .01) traits, among five dimensions of self-reported Big-five personality traits, were found to be significant in explaining hiring decisions. When it comes to the effects of three communication types on machine learning-inferred Big-five personality traits, verbal (β = .650, p < .05) and non-verbal (β = .322, p < .05) behaviors were significantly related to an extraversion trait. Among five dimensions of machine learning inferred Big-five personality traits, neuroticism (β = -.072, p < .05), extraversion (β = .063, p < .05), openness (β = .082, p < .05) traits had significant effects on interviewers’ hiring decisions. Finally, regarding the relationships between interviewees’ three communication types and hiring decisions, verbal (β = .304, p < .01) and non-verbal (β = .104, p < .05) were significantly associated with hiring decisions. Based on the results, this study’s theoretical and practical contributions, and limitations with recommendations for future research are discussed.

Description

Keywords

Job interviews, Communication types, Big five personality traits, Machine learning, Hiring decisions, Signaling theory

Graduation Month

August

Degree

Doctor of Philosophy

Department

Department of Hospitality Management

Major Professor

Jichul Jang

Date

Type

Dissertation

Citation