Deduction with numbers: proving an answer instead of guessing a pattern
Most number puzzles on the internet are induction dressed up as logic. You are shown four terms of a series and asked for the fifth, and you are expected to guess which rule the author had in mind. That is a real skill, but it is not deduction, and it has an awkward property: for any finite run of numbers there are infinitely many rules that fit, so the "correct" answer is really the most conventional one. This test does something different. Each item hands you a complete set of constraints — the totals of every row and column, the fact that two pans of a balance are level, a chain of comparisons between five objects — and asks for the one value that can satisfy all of them at once. Nothing is hidden, nothing is implied, and the answer can be proved rather than argued for.
The three families of puzzle you will meet
Grids are the most familiar. A small array of numbers has some cells filled and one blank, and the row totals, column totals or products are given. Solving one is a matter of finding the line where only a single value is missing, filling it, and letting that new value unlock the next line. Some grids in this test add a diagonal constraint, and one of them asks you to respect a Latin-square rule instead — every value appearing exactly once in each row and each column — which changes the method entirely, because you are reasoning about what a cell cannot be rather than adding anything up.
Balances present relationships between quantities rather than the quantities themselves. You might learn that four red weights balance three blue ones, that two blue balance three green, and that a green weight is eight kilograms. Nothing tells you what a red weight is directly; you have to walk the chain from the one known value back through each relation. The common mistake is to start at the unknown you were asked about and work forwards, which leaves you carrying two unknowns at once. Start where the ground is solid.
Orderings replace arithmetic with comparison. Five items have to be arranged from heaviest to lightest, or five runners placed from first to last, given a handful of statements like "B finished before D". These items look easy and trip a surprising number of people, because a partial ordering is not the same thing as a full one: you need to check that the clues force a single arrangement before you read a position off it. In this test they always do, but confirming that is part of the work.
What the test actually measures
The arithmetic here is deliberately small. Nothing goes beyond addition, subtraction, a little multiplication, and the occasional division that comes out whole. If the test measured calculation it would be a poor one, because calculation is exactly the thing a phone does better than any human. What it measures instead is the choice of first move. Given six facts, which one do you use? A strong deductive reasoner scans for the most constrained element — the row missing only one entry, the balance with a known weight on it, the person named as finishing last — and starts there. A weaker one starts at the question and works backwards into a tangle of unknowns. That difference costs far more time than any arithmetic ever will, and it is the single most transferable habit you can take away from this page.
There is a second, quieter thing being measured: your willingness to stop when the proof is complete. Constraint puzzles produce a peculiar temptation to keep checking, and on a timed test that costs you items you would otherwise have solved. Once a value satisfies every stated constraint, it is the answer. There is no second candidate hiding behind it.
Reading your score
Your raw total out of sixteen is mapped onto the familiar scale where 100 is average and fifteen points is one standard deviation. With sixteen items the resolution is coarse, so read the band rather than the number, and expect a genuine spread of several points between two attempts taken on different days.
| Score | Band | What it usually reflects |
|---|---|---|
| 130 and above | Very superior | Almost every item solved, and quickly |
| 120–129 | Superior | Comfortable with all three families |
| 110–119 | High average | Solid on grids, slower on multi-step balances |
| 90–109 | Average | Correct method, not enough time to finish |
| 80–89 | Low average | Often an ordering or arithmetic slip, not reasoning |
| Below 80 | Below average on this format | Usually means items were left blank |
How to get a better result honestly
- Write down what you know. A three-line sketch of a balance chain is faster than re-reading the question twice.
- Look for the line with one unknown before you look at the question. The blank cell is rarely where you should start.
- Check your answer against one constraint you did not use to derive it. If a grid gave you both row and column totals, you derived the value from one and can verify it with the other.
- Answer everything. Nothing is deducted for a wrong answer, so a blank is strictly worse than a guess between two survivors.
- Do not fight a single item. Twenty-five seconds of no progress means the item is costing you two easier ones somewhere further down.
Where this test fits among the others
If you want the same kind of strict logic without any arithmetic at all, the figure-based version of this test uses abstract shapes and is the fairer choice when numbers themselves are the source of your anxiety. If you want a broader estimate across several kinds of thinking, the quick IQ test mixes series, analogies and matrices into five minutes, and the culture fair test removes language from the picture entirely. Numerical deduction is the closest of the group to what employers actually assess in graduate selection, which is why the format shows up so often in aptitude batteries: it is hard to fake, hard to coach, and easy to mark.
What this test does not tell you
A sixteen-item screen taken alone at a computer is not a clinical assessment and cannot be one. There is no supervision, no standardised environment, no observation of how you approached each problem, and no properly stratified norm sample behind the numbers — only the people who happened to visit this page. Treat the score as feedback on one specific skill, useful for practice before an aptitude test and interesting as a comparison against your own results on other formats. It is not a basis for any decision about education, employment or diagnosis, and a large gap between your score here and on a visual test says more about the formats than about you.
Frequently asked questions
What is a deductive reasoning test with numbers?
It is a test where every item gives you a small set of numerical constraints — row and column totals, a balance that must stay level, or a chain of comparisons — and exactly one value satisfies all of them. You are not guessing at a pattern; you are proving a number.
How is this different from a number series test?
A number series asks you to infer a rule that was never stated, so more than one rule can often fit the visible terms. Here the rules are stated in full and the answer follows from them by arithmetic, which means every item has one defensible answer rather than a most likely one.
Do I need to be good at mathematics to score well?
No. The arithmetic never goes beyond addition, subtraction, small multiplication and the occasional division. What the test measures is whether you can pick the constraint that unlocks the puzzle first, which is a reasoning skill rather than a calculation skill.
Can I use a pen and paper?
Yes, and you should. Writing down what you already know is part of how deductive puzzles are solved, and nothing about the test assumes you are holding six numbers in your head at once. A calculator is unnecessary but not forbidden.
Why is there a fifteen minute limit?
Sixteen constraint puzzles take a well practised solver about eight minutes, so fifteen leaves room to think without letting anyone brute force every item. If the clock expires the test scores whatever you have answered so far.
Is this deductive reasoning test free?
Yes. It is free online with no account, no email address and no payment step, and you can retake it as often as you like — though a first attempt gives you the most honest number.