The science behind the score

At SIOP 2026, Maki's Science team presented 'From Experts to LLMs: Validating Open-Ended Scoring in High-Volume Recruitment.' The study validated a GPT-4o scoring system against trained I-O psychologists across behavioral competencies including Positive Attitude and Teamwork. Deployed in a real recruitment context, the system maintained high completion rates and reliably filtered out low-quality responses, while preliminary fairness analyses showed minimal demographic effects across gender, age and ethnicity.

At the ITC Conference 2026, Maki presents 'Mitigating Validity Threats in Remote Pre-employment Testing: How Proctoring Improves Decision-Making,' research on how structured proctoring strengthens the validity of remote assessments. The conference theme is 'Testing for good: using tests to benefit individuals and society.

We've submitted technical evidence to the British Psychological Society to show our Problem Solving assessment meets the standards set by regulatory bodies, including the European Federation of Psychologists' Associations (EFPA). The review runs over the coming months, and we'll share the results as soon as we receive them.

The session outlines best practices for designing assessment strategies that are reliable, fair, valid, and relevant to client needs. It starts by challenging the assumption that pre-hire work experience strongly predicts job performance, training performance, or turnover, and argues for moving beyond CVs to assess skills and capabilities directly. The approach combines psychological science with client context, beginning with job analysis and then using a mix of assessment methods that work well together.