7 Domain-Specific Weight Matrices (UK 2025)

BiasClean v2.0 uses seven evidence-based weight matrices, each aligned with structural inequality patterns in the United Kingdom. Every domain uses the same seven fairness features, but each feature is weighted differently depending on the evidence for that sector.

1. The Seven UK Domains

BiasClean operates across the following seven public and economic domains:

  • Justice

  • Health

  • Finance

  • Education

  • Hiring

  • Business

  • Governance

Each domain has a distinct weight vector based on UK evidence and regulatory priorities.

2. Universal Fairness Features

All domains use the same seven demographic fairness features:

  • Ethnicity

  • Socioeconomic Status (SES)

  • Region

  • Age

  • Gender

  • Disability Status

  • Migration Status

These features reflect key determinants of inequality highlighted across UK regulatory and statistical bodies.

3. Why Weights Differ Across Domains

Structural disadvantage varies by sector.
A domain-specific matrix ensures the bias score reflects:

  • Real UK demographic disparities

  • Policy-relevant protected characteristics

  • Sector-specific risk patterns

  • Evidence strength for each demographic feature

Without domain-specific weighting, fairness analysis would under-represent critical areas (e.g., ethnicity in justice, SES in finance).

4. Final Weight Matrix (UK, 2025)

Ethnicity
Justice 0.25 | Health 0.25 | Finance 0.20 | Education 0.20 | Hiring 0.25 | Business 0.25 | Governance 0.25

Socioeconomic Status
Justice 0.20 | Health 0.20 | Finance 0.30 | Education 0.25 | Hiring 0.15 | Business 0.15 | Governance 0.15

Region
Justice 0.15 | Health 0.10 | Finance 0.20 | Education 0.15 | Hiring 0.10 | Business 0.15 | Governance 0.15

Age
Justice 0.15 | Health 0.10 | Finance 0.10 | Education 0.10 | Hiring 0.10 | Business 0.10 | Governance 0.05

Migration Status
Justice 0.10 | Health 0.05 | Finance 0.05 | Education 0.05 | Hiring 0.05 | Business 0.05 | Governance 0.10

Disability Status
Justice 0.10 | Health 0.15 | Finance 0.05 | Education 0.15 | Hiring 0.15 | Business 0.10 | Governance 0.10

Gender
Justice 0.05 | Health 0.15 | Finance 0.10 | Education 0.10 | Hiring 0.20 | Business 0.20 | Governance 0.20

All columns sum to 1.00 for strict comparability.

5. Rationale for Each Domain

Justice

Ethnicity and SES dominate due to disproportionality across policing, prosecution and sentencing. Region and Age contribute to geographic and life-course variation.

Health

Ethnicity, SES, Disability, and Gender have strong correlations with chronic illness, waiting times and health outcomes.

Finance

SES and Region shape access to credit, lending outcomes and financial inclusion. Ethnicity remains a key axis of disparity.

Education

Attainment, progression and exclusion rates follow SES, Ethnicity, Region and Disability Status.

Hiring

Audit studies show major disparities linked to Ethnicity, Gender and Disability. SES influences hiring access and progression.

Business

Investment fairness and entrepreneurship success differ by Ethnicity, Gender, SES and Region.

Governance

Voting, participation and representation vary across Ethnicity, Gender, SES and Migration groups.

6. How BiasClean Applies These Weights

When the user selects a domain, the BiasClean engine:

  1. Loads that domain’s weight matrix

  2. Measures disparity for each demographic feature

  3. Multiplies each disparity by the corresponding domain weight

  4. Aggregates into a final weighted bias score

  5. Applies SMOTE mitigation with domain-aligned priority

This creates a transparent, UK-specific fairness pipeline.

7. Evidence Base and Official Sources (Clickable URLs)

The weight matrices are grounded in the BiasClean v2.0 Methodology Appendix, Step 1–2 Report, and the following official UK data sources:

Office for National Statistics (ONS)

Census 2021 demographics:
https://www.ons.gov.uk/census

Regional inequality & IMD:
https://www.gov.uk/government/statistics/english-indices-of-deprivation-2019

Migration statistics:
https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration

Ministry of Justice (MoJ)

Race and the Criminal Justice System:
https://www.gov.uk/government/collections/race-and-the-criminal-justice-system

NHS England / UK Health Security Agency

Health inequalities & outcomes:
https://www.england.nhs.uk/about/equality/equality-hub/health-inequalities/

Financial Conduct Authority (FCA)

Financial inclusion & consumer fairness:
https://www.fca.org.uk/firms/consumer-duty

Bank of England

Household finance and credit fairness:
https://www.bankofengland.co.uk/statistics

Department for Education (DfE)

Attainment gaps & exclusion:
https://explore-education-statistics.service.gov.uk/find-statistics

Education Policy Institute / Sutton Trust

Social mobility & education inequality:
https://epi.org.uk
https://www.suttontrust.com

British Business Bank

Diversity & entrepreneurship finance:
https://www.british-business-bank.co.uk

Electoral Commission

Civic participation statistics:
https://www.electoralcommission.org.uk

Equality and Human Rights Commission (EHRC)

Inequality and fairness reviews:
https://www.equalityhumanrights.com

8. Key UK Indicators

Justice

  • Black defendants are disproportionately represented across arrests and sentencing (MoJ).

  • Regional custody rates differ significantly between the North West, West Midlands and London.

Health

  • Chronic illness prevalence is 2–3× higher in deprived regions (ONS, NHS England).

  • Ethnicity strongly predicts major conditions such as diabetes and cardiovascular disease.

Finance

  • People in the lowest SES groups are 4× more likely to be denied credit (FCA).

  • Regional financial inclusion gaps persist between London/South East and North East/Wales.

Education

  • Disadvantaged pupils are ~18 months behind by GCSE stage (DfE/Sutton Trust).

  • Exclusion rates are substantially higher among SEND and minority groups.

Hiring

  • UK field experiments show 60–80% reduced callback rates for equivalent minority applicants.

  • Gender pay gap and leadership representation remain persistent across industries.

Business

  • Women-led and minority-led firms receive disproportionately less investment capital (British Business Bank).

Governance

  • Registration and turnout gaps exist across ethnicity, age and migration status (Electoral Commission).