Case study · KMOP

Testing unconscious bias across Greece — and building the quality system to trust the results.

For KMOP, Headway designed, built and ran a nationwide gender–career bias study: a custom testing platform, Greek-language localisation, and a data quality framework that makes the findings defensible.

Sector: Gender Equality Client: KMOP Nationwide, Greece Wave 1 complete · Wave 2 in field
Discuss a study like this See published outputs
PARTICIPANT SCREEN · placeholder — pending asset
1,145
valid responses
0%
sessions dropped after QA
5
quality checks, built-in
2
published outputs
The ask

KMOP wanted to measure implicit gender–career bias across Greece, open to anyone online. That reach is also the risk: run a split-second reaction-time test on unsupervised devices, over months, through changing recruitment channels, and the sample can quietly become something you can no longer stand behind. KMOP needed the reach and the proof.

What we delivered

One project, four Headway competences at work.

{{ c.area }}
{{ c.title }}
{{ c.desc }}
Quality architecture

Five checks, running the whole time — not just at the end.

{{ m.name }}
{{ m.result }}

These five checks are the production deployment of Headway's Data Quality Framework — the same system now used across our online behavioural and survey research.

The result
D = 0.17

A slight average implicit bias, lower among women and younger participants — published with each figure's precision tier attached, so KMOP could say exactly how far the data would stretch.

Read the Wave 1 report ↗
D-SCORE DISTRIBUTION
placeholder — pending asset
Boundaries
Reported as a description of who chose to take the test, not a probability sample of Greece, not a measure of behaviour, and not a personal diagnostic — with individual scores read cautiously throughout.
What's next

Wave 1's evidence — solid instrument, thin subgroups, a sample skewed female and urban — set Wave 2's design: a preregistered, confirmatory study targeting 2,000+ responses, with recruitment aimed at the groups Wave 1 couldn't yet measure precisely.

Need behavioural research that will survive review?

Tell us the research question — we'll come back with a scoped approach.

Outputs
{{ o.tag }}
{{ o.title }}
{{ o.desc }}