Our methodology
How every iPrepGenius question gets made
Human judgement decides what a good question is. Technology makes sure your child never runs out of them. Nothing reaches a child unverified.
Most free practice material is a fixed set of questions someone typed up years ago. Children memorise it, parents mistake that for readiness, and test day tells the truth. We built the opposite: hand-designed patterns, scaled into practice that never repeats, verified every time, and sharpened by how real children perform on it.
We start from the blueprint, not from a worksheet
Before writing a single question we work out what a test actually measures — battery by battery, question type by question type, at each grade level. The CogAT's verbal, quantitative and nonverbal batteries test different reasoning; the NNAT is entirely figural; the SSAT penalises wrong answers while the ISEE doesn't. Those differences decide everything downstream.
This is also why we publish our composition research openly — we had to measure it to build against it.
A human curates the core patterns for every grade and test
For each skill we hand-build the underlying question patterns: the relationship families behind verbal analogies (opposites, part-to-whole, worker-to-workplace), the rules behind number series (constant difference, growing difference, squares), the transformation logic behind figure matrices (rotation, fill, count, position, AB and AAB repeats).
This is the part that can't be automated. Knowing that a kindergartner needs an AB colour pattern while a sixth grader needs two attributes changing at once is a judgement call — made by someone who has sat with children doing these questions.
Technology scales those patterns into endless variation
Once a pattern is right, our systems generate thousands of distinct questions from it — varying the numbers, shapes, vocabulary, distractors and difficulty, while the underlying reasoning stays faithful to the pattern a human designed. Each question is built for a specific grade band, so the numbers stay inside a kindergartner's range and the relationships stay inside a sixth grader's.
This is why practice never repeats. A child can't memorise their way through the set — improvement has to come from the reasoning improving.
Every question is checked before a child sees it
Answer keys are verified against the generating rule, so the stated answer is provably correct. Distractors are checked to be plausible but unambiguous — exactly one option can be right. Duplicates are removed at the database level, and question wording is reviewed for age-appropriate readability.
Anything that fails is rejected rather than published. We would rather have a smaller bank than a wrong answer key.
Real practice data keeps it honest
Once questions are live, we watch how children actually perform on them. A question that nearly everyone gets right isn't measuring anything; one that nearly everyone misses is usually flawed rather than hard. Both get recalibrated or retired automatically.
The bank gets better the more it's used — which is the opposite of a static PDF that was already dated when it was scanned.
Why this matters for your child
- Practice that never repeats means progress is real, not memorised.
- Grade-banded questions keep the work in the zone where learning happens — not so easy it's pointless, not so hard it's discouraging.
- Building to the real blueprint means time is spent on the reasoning the test actually measures, including the nonverbal skills classrooms rarely teach.
- Every question carries a full explanation, so a wrong answer becomes a lesson instead of a dead end.
What we never do
- We never reproduce real test items. Every question is original — anyone selling you "actual questions" is either mistaken or breaking the law.
- We never scrape or recycle other sites' content.
- We never publish a score cutoff we can't verify — we point you to your district instead.
- We never ship a question whose answer key hasn't been checked.
Common questions
Are iPrepGenius questions written by humans or by AI?
Both, deliberately. The question patterns — what a skill is, how it should be tested, and what makes it appropriate for a given grade — are designed by people. Technology then scales those patterns into thousands of unique variations, and every generated question is verified against its rule before publication. Humans decide what a good question is; technology handles the volume that makes non-repeating practice possible.
How can practice questions never repeat?
Because we generate from patterns rather than storing a fixed worksheet. There are over 130,000 verified questions in the bank and more can be produced from the same rules, so a child working daily won't see the same item twice — which means score improvements reflect real skill gains, not recognition.
How do you know the questions match the real test?
Each test's published blueprint tells us its batteries, question types and their proportions. We build to that composition, then verify our own bank against it — the analysis is published openly in our research section.
Are the questions right for my child's grade?
Yes. Every question is generated inside a grade band that controls both content and difficulty — number ranges, vocabulary, and how many attributes change at once. A kindergarten pattern question and a sixth-grade one are built by different rules, not the same question with bigger numbers.
This approach came out of one parent's frustration with everything else on offer — read the story behind it, or see the open research we publish from the bank itself.
See the questions for yourself
Free, grade-matched, with a full explanation on every answer. No signup.
A question built exactly this way, plus one practical tip — free, every week.
Double opt-in. We never sell your email. One-click unsubscribe.
Test names are trademarks of their respective owners. iPrepGenius is an independent study resource and is not affiliated with or endorsed by any test publisher. Our practice materials are original and are not reproductions of any live or retired test form.