DataTerminology.comNEW
Also known as: Statistical Parity, Group Fairness (Demographic Parity)
A fairness goal where outcomes are independent of protected attributes.
DESCRIPTION
Demographic parity requires that the rate of positive outcomes is equal across protected groups. It focuses on group-level equity regardless of differences in base rates. While simple and auditable, it can conflict with predictive performance and may be inappropriate when qualification rates differ meaningfully. It is useful for screening and outreach contexts prioritizing equal access.
ALTERNATIVE DEFINITIONS
Fairness and Machine Learning (Barocas, Hardt, Narayanan)
Statistical parity, also called demographic parity, requires that the rate of positive classification be the same across groups.
Concise, canonical statement in a widely cited open book.
Fairlearn
Demographic parity requires that the selection rate (the fraction of examples with positive outcome) be identical across the groups defined by the sensitive feature.
Operationalizes parity via the selection rate across sensitive groups.
IBM AI Fairness 360
Statistical parity difference measures the difference in selection rates between the unprivileged and privileged groups; zero indicates demographic parity.
Defines a concrete metric operationalizing demographic parity.
RELATED TERMS
Equal Opportunity Predictive Parity Fairness Metrics Calibration Base Rate Sensitive Attribute Bias Mitigation Selection Rate Disparate Impact
COMPETING TERMS
KNOWLEDGE AREAS
AI Data Science AI Governance Testing & Validation
TARGET AUDIENCE
Data Scientist AI Engineer MLOps Engineer Data Governance Manager Data Steward
EXTERNAL SOURCES
- Wikipedia - Statistical parity
- Fairlearn - Glossary and metric definition
- IBM - Operational metric references
- Fairness and Machine Learning - Authoritative open textbook
TOP QUESTIONS
When should Demographic Parity be used?
Use it when equal access or exposure is the primary objective, such as outreach or non-selective screening, and differences in base rates are not intended to drive outcomes.
Category: Guidance
What is a common mistake with Demographic Parity?
Applying it where base rates differ legitimately can increase error rates or reduce utility for everyone. Context and policy objectives must guide metric choice.
Category: Mistakes
How do you measure Demographic Parity in practice?
Compute selection rates by group and compare differences or ratios, such as statistical parity difference or disparate impact ratio.
Category: How-To
How does Demographic Parity relate to legal compliance?
It aligns with disparate impact analysis in some jurisdictions but is not a legal standard itself. Consult legal counsel for sector-specific requirements.
Category: Compliance
How does Demographic Parity compare to Equalized Odds?
Demographic parity ignores labels and seeks equal positive rates, while equalized odds aligns error rates across groups conditioned on true labels.
Category: Comparison
What is the risk of optimizing only for Demographic Parity?
You may admit more false positives in some groups or reject qualified candidates in others, harming utility and trust.
Category: Risks
Can Demographic Parity be partially satisfied?
Yes, many teams set bounds on acceptable disparities and optimize within those constraints, balancing utility and fairness.
Category: Practice
How stable is Demographic Parity over time?
It can drift with population changes and seasonality; monitor selection rates and confidence intervals, retraining or re-thresholding as needed.
Category: Monitoring
What tools support Demographic Parity evaluation?
Fairlearn, AIF360, and Fairness Indicators provide group-wise reports and metrics for parity evaluation.
Category: Tooling