English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

We develop tools for utilizing correspondence experiments to detect illegaldiscrimination by individual employers. Employers violate US employment law iftheir propensity to contact applicants depends on protected characteristicssuch as race or sex. We establish identification of higher moments of thecausal effects of protected characteristics on callback rates as a function ofthe number of fictitious applications sent to each job ad. These moments areused to bound the fraction of jobs that illegally discriminate. Applying ourresults to three experimental datasets, we find evidence of significantemployer heterogeneity in discriminatory behavior, with the standard deviationof gaps in job-specific callback probabilities across protected groupsaveraging roughly twice the mean gap. In a recent experiment manipulatingracially distinctive names, we estimate that at least 85% of jobs that contactboth of two white applications and neither of two black applications areengaged in illegal discrimination. To assess the tradeoff between type I and IIerrors presented by these patterns, we consider the performance of a series ofdecision rules for investigating suspicious callback behavior under a simpletwo-type model that rationalizes the experimental data. Though, in ourpreferred specification, only 17% of employers are estimated to discriminate onthe basis of race, we find that an experiment sending 10 applications to eachjob would enable accurate detection of 7-10% of discriminators while falselyaccusing fewer than 0.2% of non-discriminators. A minimax decision ruleacknowledging partial identification of the joint distribution of callbackrates yields higher error rates but more investigations than our baselinetwo-type model. Our results suggest illegal labor market discrimination can bereliably monitored with relatively small modifications to existing auditdesigns.

Audits as Evidence: Experiments, Ensembles, and Enforcement