Game-based assessments: what we know

Game-based assessments are sold on two arguments. The first is candidate experience: a task with movement, feedback and an element of play feels less like an exam than a questionnaire with two hundred statements. The second is that candidates find it harder to work out what is being measured, and therefore harder to answer strategically.

Both arguments are reasonable. Drop-off during a recruitment process is a real problem, particularly in volume hiring and among younger candidates, and anything that shortens perceived wasted time has value. And unlike a personality questionnaire, where the desirable answer is often obvious, a game-based format generally measures behaviour rather than self-report.

What is said less plainly is that both promises concern the format, not what is being measured. A game can be a pleasant wrapper around a weakly evidenced measurement.

The question is whether game-based assessments measure the same thing as established instruments, and whether they predict job performance equally well.

Here the picture is uneven. Some vendors build their games on well-established constructs, typically cognitive ability or specific executive functions, and can produce studies of the relationship with traditional measures. Others measure a range of behavioural signals generated by the task without making clear which psychological construct those signals represent. The difference matters, because documentation on the second track often consists of internal analyses rather than independent studies.

Bias is the second area to watch. A game-based format does not remove skew by itself, and it can introduce new kinds. Gaming experience, motor skill, reaction speed and equipment quality all affect the result without having anything to do with the job. A vendor that has taken the question seriously can tell you which groups were tested for differences, and what was found.

Three questions cover most of it. Which construct does the game measure, and who determined that? Do studies exist comparing the game's scores against an established instrument or against job performance, and were they conducted independently? And which bias analyses have been run, on which groups?

A useful addition is to ask what happens to a candidate who cannot complete the task because of equipment, motor skill or a disability. If an alternative exists, the format has been thought through. If not, that tells you how far the vendor has got.

Source: Based on publicly available vendor manuals and peer reviewed literature on assessment methods.