Numerical Reasoning Test

What is numerical reasoning?

Numerical reasoning is the ability to think logically about numbers. It is not raw arithmetic speed (how fast you can compute 47 times 8) nor stored mathematical knowledge (whether you remember the quadratic formula). It is the cognitive skill of seeing patterns, ratios, and relationships in quantitative information - spotting that a sequence is doubling rather than adding, recognizing that a 20% increase followed by a 20% decrease leaves you below the start, inferring a trend from a table of figures.

It is one of the most heavily tested cognitive skills in the modern workplace. Graduate employers use SHL, Kenexa, and Cubiks numerical reasoning batteries to screen candidates. Banks, consultancies, and most analytical roles require a passing score. The format is largely standardized: data is presented in tables or charts, and you compute an answer from multiple choices under time pressure.

Numerical reasoning vs math ability

The two are related but distinguishable. Math ability includes computational fluency, procedural knowledge (algebra, geometry, calculus), and the ability to construct proofs. Numerical reasoning is narrower: applying logic to quantitative information, without necessarily requiring advanced math.

In practice:

• A trained mathematician will likely score very high on numerical reasoning - the skills overlap heavily
• A person who is "bad at math" because they failed algebra in school may still score well on numerical reasoning, which depends more on pattern recognition than on procedural knowledge
• A person with excellent computational arithmetic but weak logical reasoning may score middling - the test rewards interpretation, not just computation

The number sequence families you need to know

Most numerical reasoning items reduce to one of these pattern families. Recognizing them on sight is most of the skill:

The hardest items combine families - for example, an alternating sequence where each sub-sequence is itself a geometric progression.

Strategies that genuinely help

• Check the differences first. Subtract consecutive terms. If the differences are constant, it is arithmetic. If they themselves form a pattern, it is second-difference.
• Check the ratios second. If consecutive terms divide cleanly, it is geometric.
• Try splitting alternating positions. Many seemingly random sequences are two interleaved simple sequences.
• Watch for hidden formulas. n²+1, n²-1, n(n+1)/2 (triangular numbers) all appear frequently.
• Read the chart caption first. On data-table items, the units (percentages? thousands? cumulative?) matter more than the numbers.

Why numerical reasoning improves with practice

Of all cognitive skills, numerical reasoning is among the most trainable. The reason is that pattern families repeat across tests - once you have seen 50 number sequences with full explanations, you have implicitly learned a classifier for the common patterns. Studies of test-prep effectiveness consistently show double-digit percentile gains on graduate numerical batteries from 4 to 8 hours of focused practice.

This is partly why employers should not over-weight numerical reasoning scores: a high score reflects both ability and preparation, and the preparation component is large. Two candidates with the same underlying ability will produce very different scores if one has practiced and the other has not.

Try numerical puzzles now

How numerical reasoning fits in your IQ profile

On the WAIS-IV, numerical items appear in Arithmetic (under Working Memory Index) and across the Perceptual Reasoning Index. On our full IQ test, numerical reasoning loads onto the logical/working-memory subscores depending on item type. A profile with strong numerical and weaker verbal is common in mathematics, engineering, and quantitative-finance backgrounds - it is one of the most cleanly trainable cognitive styles.

Important caveats

Numerical reasoning scores are sensitive to recent practice. A high score from a well-prepared candidate and the same score from an unprepared one represent different underlying ability. Use the result as a starting estimate and re-test after 4 to 6 weeks of practice if you want a more stable measure.