Does Practising Psychometric Tests Actually Work? What the Evidence Says
Practice effects on aptitude tests are real, well documented and much smaller than prep sites imply. Where the gains actually come from, when they stop, and how to spend your preparation time.
We sell assessment practice, so treat what follows with appropriate suspicion and check it against your own experience. The claim we are not going to make is that practice raises your cognitive ability. It does not, in any meaningful sense, in the timescale of a job application. What it does is remove a set of costs that have nothing to do with ability and that are quietly eating your score.
What the research consistently finds
Retest effects on cognitive ability tests are one of the better-replicated findings in assessment research. Sit the same kind of test twice and the second score is typically higher. The effect is reliable enough that test publishers design around it: they maintain parallel forms, they generate items algorithmically rather than reusing a fixed bank, and employers restrict retakes partly for this reason.
The two features of the effect that matter to you are its shape and its source. The shape is front-loaded: the jump from the first sitting to the second is much larger than the jump from the second to the third, and by the fourth or fifth exposure it has largely flattened. The source is mostly not ability. It is familiarity.
Where the gain actually comes from
Where practice does almost nothing
Being straight about this is the point of the article.
Personality questionnaires
There is nothing to improve. Practising a personality inventory tells you what the format feels like, which is worth one run and no more. Attempting to optimise a profile is actively counterproductive on any instrument with an ipsative format or a self-presentation measure, and most of the major ones have one or the other.
Raw processing speed
Speed batteries like the Thomas GIA or the checking sections in Saville and Sova measure how fast you handle simple information. The first run removes the friction. The tenth run is training a trait that moves slowly, and the transfer from a practice version to the real instrument is imperfect.
Behavioural game-based assessments
Games that record how you behave rather than what you answer, such as the balloon-inflation risk task, are measuring your pattern of decisions. There is no correct pattern to converge on, and a candidate who has decided in advance to look bold is producing a profile of someone else. The differences between game-based and traditional tests are worth understanding for exactly this reason.
The way practice can backfire
This is underrated and genuinely costly. Aptitude tests differ between vendors on rules that invert, and the most important one is whether wrong answers are penalised.
Spend a week training yourself to answer every item because blanks score zero, then sit an Aon cut-e scales test where wrong answers subtract marks, and your preparation has actively damaged your result. The reverse is equally true: arriving at an SHL test with a carefully trained skipping habit leaves marks on the table for no reason.
Which means the first useful thing to establish is not how to answer but which provider sent your test. Practising the wrong format is the only version of preparation with a negative expected value.
How much practice is enough
A defensible allocation for a single upcoming assessment, assuming you know the vendor:
- One full-length run of each section under real conditions. This buys the largest part of the effect. Do it timed, on the device you will use, without pausing.
- Targeted work on your worst section. Usually numerical for humanities graduates and verbal for engineers, and usually worth two or three focused sessions rather than twenty.
- One more full run a day or two before. Enough to keep the format familiar, not so much that you arrive stale.
Beyond roughly that, you are into diminishing returns, and time spent on the interview or the application itself has a higher marginal value. The week-before plan sets that out day by day.
The uncomfortable conclusion
If two candidates have genuinely different reasoning ability, practice will not close the gap and nobody should promise you it will. What practice reliably does is stop you scoring below your own ability because you spent the first ninety seconds working out where the data was hidden and whether you were allowed to guess.
That is a smaller claim than the one prep marketing usually makes. It is also, for most candidates, the difference between a result that represents them and one that does not.