Infrequent Identity Signals, Multiple Correspondence, and Detection Risks in Audit Correspondence Studies.

Area of Science:

• Social Sciences
• Economics
• Sociology

Background:

• Audit correspondence studies are field experiments used to detect discrimination in markets.
• These studies compare responses to individuals with different characteristics.

Purpose of the Study:

• To analyze the trade-offs in audit correspondence studies concerning the number of correspondences sent.
• To investigate how detection risk impacts discrimination estimates, particularly for infrequently targeted groups.

Main Methods:

• The study discusses theoretical trade-offs between single and multiple correspondences.
• It examines the impact of including less common identity signals.
• Presents results from an audit correspondence study demonstrating detection risks.

Main Results:

• Using multiple correspondences can increase the likelihood of audited units detecting the study.
• This detection can occur when studying discrimination against individuals with infrequent interactions.
• Detection leads to attenuated (biased towards zero) discrimination estimates.

Conclusions:

• Researchers must carefully consider the number of correspondences to balance detection risk and data quality.
• The methodology of audit correspondence studies can be compromised by participant awareness.
• Findings highlight potential biases in measuring discrimination, especially for marginalized groups.