Ixly Practice Tests: Test-Toolkit, ACT and Cognitive Ability Pro

Ixly reports four numbers, not one. A stenscore from 1 to 10, a T-score around a mean of 50, a percentile, and an IQ score with a mean of 100 and a standard deviation of 15. Each comes with an 80% confidence interval, phrased on the report as being able to say with 80% certainty that your total score lies between two values.

A sten of 5 is an average score. It is not an onvoldoende. The scale has a mean of 5.5 and a standard deviation of 2, so 5 and 6 are the middle bands and roughly 38% of the population sits in them. Ixly's manual warns about this directly, because on a Dutch school report a 5 is a fail and candidates read it that way. It is also a sten and not a stanine: a stanine runs 1 to 9 with a mean of 5, and the two must not be conflated.

The comparison group is your education level, and it changes the answer completely. The short questionnaire before your first test collects age, gender and education level, and the last of those selects your norm group: VMBO, MBO, HBO or WO. The manual's own worked example takes one raw score and reports it as sten 8.3 and the 92nd percentile against the VMBO group, and sten 4.2 and the 25th percentile against the WO group. That is why a percentile with no named norm group means very little on this provider, and why our practice percentile is a ranking against people who practise here rather than a prediction of your Ixly sten.

There can be a real pass mark. Ixly's Sogeti case describes calibrated cut-off scores built for two role categories, engineering and consultancy, tuned over a year to balance rejecting unsuitable candidates against rejecting good ones. So depending on the employer, an Ixly result may be a hard cut rather than a ranking.

Ixly advises employers to decide on the overall g-score. Published reliabilities are .81 for Cijferreeksen, .77 for Figurenreeksen, .86 for Verbale Analogieën and .92 for the overall score, and Ixly notes that Figurenreeksen alone falls short of the usual .80 threshold. So one weak block matters less than a weak total.

An unanswered item is scored as wrong. Not skipped, not neutral: wrong. That single rule inverts the advice you may have learned on a negatively marked test like Aon's or the Thomas GIA. Here there is no such thing as a blank, so a guess can never cost you anything a timeout would not cost you anyway. At 45 seconds an item, the habit to build is committing your best answer at around 35 seconds and moving on.

You cannot go back.The manual states that an answer is saved to the database as soon as it is given and cannot be changed, and that the browser Back button has no effect. Leaving the test does not buy you thinking time either: the item's clock keeps running while you are away, and you resume at the next item rather than the one you left.

Every test is preceded by a worked example and three untimed practice items. The practice items tell you whether you were right and leave the message on screen as long as you want. Use them properly: they are the only untimed part of the sitting, and Ixly deliberately explains the adaptivity in plain language beforehand because its own research found that doing so improves the experience.

Figurenreeksen contains two different item shapes, and sources disagree about which. Ixly's Abstract Reasoning product page describes progressive series where you choose the right successor. The ACT manual's own worked example for the same bank is a 3x3 grid with eight cells filled, the bottom right empty and four options labelled A to D. A Dutch preparation vendor lists both Figuurreeksen and Matrixen. We build both, and you should expect either.

Verbale Analogieën asks for two answers, not one. Two words are shown in a square, and you choose from six words exactly the two that form the same relation. One mark for the pair. There are fifteen possible pairs among six options, so guessing here is worth far less than on an ordinary multiple choice, which makes working the relation out first the only viable approach.

Three numbers in our simulation are ours, not Ixly's. Ixly publishes no item count for Cognitive Ability Pro and no per-item time, so our 20 items at 40 seconds is inferred from the 15 minute duration. It publishes nothing at all about Stroomdiagrammen's internals, so our 14 items at 75 seconds is inferred from the 20 minute duration. And it names none of Cognitive Ability Pro's eleven question types: four are named by a Dutch preparation vendor, and those four are the ones we mix. We do not claim to know the other seven.

Our adaptation is a staircase, not Ixly's algorithm. The real ACT selects items by Maximum Expected Information, estimates ability by EAP, and stops when the standard error of measurement falls below 0.39. That needs a bank calibrated on real response data. Ours tracks your level up and down and stops when the estimate settles, which reproduces what you experience, the difficulty chasing you and the test ending at a different length for different people, without pretending to be the same engine.

What the countdown looks like on the real test is unknown. No source shows it. We show a visible clock, because practice that hides the clock teaches nothing about pacing, but the presentation is our design decision rather than a reproduction.

The TMA method, Talenten Motivatie Analyse, is a separate instrument published by a separate company, EhrmVision. Ixly's own product catalogue contains no TMA product, its about page never mentions TMA or EhrmVision and lists only university research collaborations, and its implementation partners page names no partners at all.

The confusion is understandable rather than careless. Both are Dutch publishers with a Utrecht address; several Ixly products use the word Talent in their names, which reads similarly to Talenten Motivatie Analyse; and Dutch practice-test aggregators list Ixly and TMA side by side as interchangeable assessment bureaus you might be sent to.

We assert no relationship in either direction because no source supports one: no partnership, no licence, no acquisition. What we can say is that they are different instruments, and that if you have been told you are sitting a TMA, practising Ixly's tests is not what you need.

Ixly splits its own catalogue into Selectie, Ontwikkeling, Leiderschap and Retentie, and we build from the first column only. Our product is preparation for a test you are about to sit as part of an application. Practising an instrument nobody is screened on would be selling you the wrong thing. So the Talent Development Profile, 360 Feedback, the Learning Agility Profile, Thinking Styles, Team Role, the Career Profile, Carrièrewaarden, the study and career interest questionnaires, the Employability Scan and every leadership instrument are all out, because Ixly lists them under development, leadership or career rather than under selection.

Four more are real selection tests that we cannot yet build faithfully. Integriteit Besluitvorming is explicitly used in selection and reports on four scales, but Ixly publishes nothing about its item format, response mechanics or clock. Ixly's four games, Rally, Connect, Blocks and Barista, come with a construct and a duration and nothing else. Talent Select Profiel and Competentie Select Profiel are composite report bundles assembled from underlying tests rather than single instruments. And Begrijpend Lezen reports CEFR language levels rather than a percentile or a sten, which is a fifth reporting model again. If you have been invited to one of these, the stage is real and we do not simulate it.

Carrièrewaarden is the closest call and is out for now. Ixly places Career Drives under retention while noting it supports selection, development and retention; a secondary source says the normative version is for advisory use while the ipsative version can be used for selection. Given that conflict we have left it out rather than describe it either way.