Employment discrimination based on applicants’ gender, race, age, or other traits is well documented, but the extent to which discrimination depends on specific job demands—the bundle of qualifications, skills, and expectations—remains unclear. Sociologists Katherine Weisshaar and Koji Chavez will examine the extent to which gender and racial discrimination vary across job demands and whether employers are more likely to discriminate against women or Black applicants when they do not meet a job’s required demands. Weisshaar and Chavez will study hiring in four occupations—accountants, human resource professionals, sales professionals, and software engineers—that differ in female representation (high, low) and task orientation (interpersonal, technical). They will field a large audit correspondence study in which applicants vary by perceived race (Black and White) and gender (male and female), signaled by name. They will submit about 16,000 job applications to real online job openings across each occupation, recording callbacks across experimental conditions. To identify the specific demands for each opening, the PIs will integrate their correspondence study data with data from Burning Glass Technologies (BGT), which captures most online job postings and uses machine learning techniques to identify required skills, education levels, qualifications, and credentials for each listing. BGT also classifies job demands by local labor market characteristics (e.g., supply, demand, level of specialization, salary associations, and job vacancy duration). By merging the BGT data with the correspondence study data, the PIs can identify demands for each job opening along with local labor market outcomes associated with each job.