A recent study led by Lancaster University in the UK raised concerns about the unintended consequences of using equality, diversity, and inclusion (EDI) in job advertisements. This language, intended to create a more gender-balanced workplace, is counterproductive. The study indicates that such language, while designed to attract racial minorities, does not effectively change the racial makeup of the workforce.
The research shows that workplaces with more female employees often include job ads featuring language related to family-friendly policies and flexible work arrangements. This type of language attracts more female than male applicants, potentially increasing gender segregation within these organisations. Similarly, workplaces with a significant number of racial minority employees frequently use EDI policies and language that signals a commitment to a diverse and inclusive workplace culture. However, these messages have little effect on the actual racial composition of the workforce.
The study’s findings are published in PNAS Nexus, an official journal of the National Academy of Sciences of the United States of America. The paper, titled ‘Language in Job Advertisements and the Reproduction of labour force gender and racial segregation,’ offers an extensive audit of how gender and EDI language in job advertisements influence the gender and racial composition of the workforce. It also examines how the existing composition of the labour force affects the use of gender and EDI language in job ads.
This research was a collaborative effort involving universities from the UK, Canada, and the USA, supported by funding from the UKRI Economic and Social Research Council (ESRC) in the UK and the Social Sciences and Humanities Research Council (SSHRC) of Canada. The project is part of a more extensive study focusing on AI and labour market equality. The research team, comprising sociologists, management scholars, and data scientists, analysed 28.6 million job advertisements in the UK from 2018 to 2023 using advanced natural language processing techniques. Combined with ONS labour force statistics, this approach makes it the most comprehensive study.
Professor Yang Hu from Lancaster University, the study’s lead author, emphasised the importance of understanding and addressing persistent gender and racial segregation in the labour force to enhance equality and diversity in the job market. Associate Professor Nicole Denier from the University of Alberta highlighted how job advertisements are the initial contact point between job seekers and employers, shaping the labour force by influencing who applies for jobs and how candidates are evaluated.
The research developed a novel inventory to categorise the language used in job ads across six dimensions related to gender and EDI. These include explicit gender references, gendered psychological cues, gendered work roles and skills, family-friendly versus family-unfriendly policies, EDI policy pledges, and EDI cultural references. This inventory helped the researchers map the use of gender and EDI language across different occupations and industries and correlate it with the gender and racial demographics of the respective workforces.
The study identifies three distinct patterns in the interaction between job ad language and workforce composition. Firstly, it suggests that language in job ads can reinforce workforce segregation by using cues that predominantly attract female applicants in sectors already dominated by women. Secondly, it finds that certain language elements in job ads could disrupt workforce segregation by balancing gender representation. Lastly, it reveals that despite the inclusion of EDI language in job ads, there is little impact on the representation of racial minorities in the workforce.
These insights underline the complexities of using language in job ads to influence workforce composition and point to the limitations of such interventions. Professor Hu called for a significant reevaluation of how employers frame their job ads, suggesting more meaningful approaches to communicate and implement EDI strategies.
Professor Monideepa Tarafdar from the University of Massachusetts Amherst, a co-principal Investigator of the project, and Professor Karen Hughes from the University of Alberta highlighted the importance of this cross-disciplinary, international collaboration. They noted that integrating large language models and AI-driven text analysis tools in drafting job ads offers a promising path towards building labour market equality, which this research helps pave the way.
More information: Yang Hu et al, Language in job advertisements and the reproduction of labor force gender and racial segregation, PNAS Nexus. DOI: 10.1093/pnasnexus/pgae526
Journal information: PNAS Nexus Provided by Lancaster University