Game-based assessments in recruitment: what the research shows

Game-based assessments measure candidates through play — short interactive tasks that record not just answers but behaviour. The marketing claims run ahead of the science, but the science is now substantial enough to say what these tools genuinely add to a selection process, and where the cautions lie.

What they are

The research distinguishes two families. Gamified assessments add game elements — storylines, points, progress — to a conventional test. Game-based assessments (GBAs) are actual games designed so that play itself measures a construct: working memory through a sequence game, risk tolerance through a balloon-pumping task, learning through how performance changes across rounds.^[1] The distinction matters because the evidence differs: gamified versions of established tests inherit much of their psychometric pedigree; purpose-built games must earn theirs construct by construct.

The evidence for them

Validity is achievable. A large applicant study of game-based cognitive measures reported convergent validity of r = .50 against conventional ability tests and test–retest reliability of .68^[2] — respectable, if below the best traditional instruments. Gamified personality measures show significant convergent correlations with the self-report traits they target.^[3]

Candidate experience is the clearest win. Feedback from 4,778 real applicants produced a strongly positive endorsement of the format,^[2] and experimental work finds GBAs induce flow — absorbed engagement — which in turn raises perceived job-relatedness, fairness perceptions, organisational attractiveness and intention to accept an offer.^[4]

Test anxiety is lower. The same applicant research found no significant gender differences in test-taking anxiety from the games and no direct anxiety–performance link, supporting the claim that games soften the anxiety that depresses some candidates' scores on formal tests.^[2] For early-career candidates with thin CVs and little test experience, that is a real measurement improvement, not just a nicer experience.

Behavioural data is richer. Games record process — exploration, error recovery, adaptation across rounds — that a right-or-wrong test discards. This is the signal that makes GBAs distinctive in a battery: they observe how a candidate approaches an unfamiliar problem rather than asking them to report it.^[1]

The honest cautions

A systematic review of the field concluded that GBAs can be used in selection, but that their supposed advantages over conventional tests are "fewer than imagined", with construct validity the area most needing work.^[1] Applicants themselves raise face-validity concerns about abstract games — a balloon task does not obviously resemble the job.^[2] Recent applicant research also finds reactions vary with gaming experience: candidates who play games recreationally react more positively and may hold an edge, a fairness consideration test designers must check rather than assume away.^[5] And where machine-learning scoring is used, fairness must be engineered and audited — the same study demonstrating fairness-optimised scoring did so precisely because unaudited models can inherit bias.^[2]

None of these cautions is unique to games — every method on the market has its versions — but they set the bar: a defensible GBA publishes what each game measures, evidences it against established tests, and monitors subgroup differences and gaming-experience effects in live data.

Where GBAs fit

The research supports GBAs as a complement, not a replacement: use them at the top of the funnel, where volume is high, candidate experience shapes employer brand, and test anxiety costs the most signal — and pair them with skills tests, structured interviews or SJTs deeper in the process, where face validity and role fidelity matter most. Used that way, they widen the funnel's fairness and appeal without asking the games to carry evidential weight the literature says they cannot yet bear.^[1,2]