Field Experiments and Job Posting Sources: The Consequences of Job Database Selection for Estimates of Racial Discrimination

Abstract

Field experiments have proliferated throughout the social sciences and have become a mainstay for identifying racial discrimination during the hiring process. To date, field experiments of labor market discrimination have generally drawn their sample of job postings from limited sources, often from a single major online job posting website. While providing a large pool of job postings across labor markets, this narrow sampling procedure leaves open questions about the generalizability of the findings from field experiments of racial discrimination in the extant literature. In this paper, we present evidence from a field experiment examining racial discrimination in the hiring process that draws its sample from two sources: (1) a national online job posting website that aligns with previous research, and (2) a job aggregator service that scrapes the web daily in an effort to obtain all online job postings in the United States. While differing in the types of information they collect, we find the job postings drawn from the two sources result in similar estimates of discrimination against Black applicants. In other words, we do not find evidence that racial discrimination varies by the source of the job posting. We conclude by discussing the implications of these findings for studies of racial discrimination, discrimination along other axes of social difference, as well as field-experimental methods more broadly.

The (In)Flexibility of Racial Discrimination: Labor Market Context and the Racial Wage Gap in the United States, 2000 to 2021

“…Second, I advance the concept of wage discrimination flexibility to capture the ways in which racial discrimination changes in various economic environments. The rigidity of wage discrimination against black workers in the labor market is consistent with other findings documenting the stability, persistence, and diffuseness of antiblack racism in the U.S. labor market (Pedulla et al 2021; Quillian et al 2017). These findings also support Sears and Savalei’s (2006) “black exceptionalism” hypothesis, which posits that an “impermeable color line” continues to hinder the integration of black workers into the broader society, whereas the color line is more porous for other racialized groups (e.g., Hispanic and Asian workers).…”

“…Negative stereotypes about black workers, such as poor work ethic, lack of competence, being unmotivated, and lesser competence, persist and continue to shape employment decisions (Kirschenman and Neckerman 1991;Moss and Tilly 2001;Pedulla 2018;Waldinger and Lichter 2003). Showing the diffusiveness of antiblack discrimination in the labor market, Pedulla et al (2021) found that even when black job seekers provide additional market-related information on job applications (e.g., references, skills, short essays about their prior work experience and qualifications, background check information), they are still discriminated against at the same rate as when they do not provide such information. These findings suggest the important role of "deep-seated stereotypes and implicit bias" against black job applicants in hiring (Pedulla et al 2021:12).…”

“…Showing the diffusiveness of antiblack discrimination in the labor market, Pedulla et al (2021) found that even when black job seekers provide additional market-related information on job applications (e.g., references, skills, short essays about their prior work experience and qualifications, background check information), they are still discriminated against at the same rate as when they do not provide such information. These findings suggest the important role of “deep-seated stereotypes and implicit bias” against black job applicants in hiring (Pedulla et al 2021:12). Furthermore, black workers report experiencing inferiority-based discrimination (e.g., defined in terms of perceived intellectual, economic, and occupational prestige) compared with Hispanic, Asian, and white workers (Zou and Cheryan 2017).…”

A Novel Stream Mining Approach as Stream-Cluster Feature Tree Algorithm: A Case Study in Turkish Job Postings

“…It can be found that clustering algorithms are used predominantly for text mining analysis [24,25], even if classification algorithms are used to create general prediction models [26] or specific models, such as those for detecting errors in job postings [27]. Studies have utilized clustering methods for job postings associated with various occupational groups, such as those in the accounting [28], health [29], art [30], engineering [31], and education [32] fields.…”

Addressing methodological assumptions of correspondence tests when measuring discrimination

“…Some methodological caveats of correspondence testing have already been addressed in previous research. These include the use of names to signal characteristics of the candidate (Crabtree & Chykina, 2018;Gaddis, 2017;Martiniello & Verhaeghe, 2022), the experimenter bias in coding the responses to the applications (Lahey & Beasley, 2018), and the robustness of correspondence tests when using different sampling methods (Auspurg et al, 2020;Pedulla et al, 2021). However, whereas these studies focused predominantly on the correspondence tests as such, we focus on the robustness of the tests in relation to different methodological choices made by the researcher.…”