Improve Numerical Reasoning

Numerical reasoning improves with two complementary forms of practice: building automaticity in basic arithmetic (so that working memory is freed up for the rule-search) and developing pattern-detection fluency on number sequences. Both respond to targeted practice within weeks.

For arithmetic automaticity, the goal is to make multiplication tables, common percentages (10%, 25%, 50%, 75%), and basic factoring of small integers as automatic as recognizing your own name. Drill cards, timed worksheets, and apps like Mental Math Master all accomplish this for adults who put in 10 to 20 minutes per day for several weeks. Once arithmetic becomes automatic, more cognitive resources are available for rule-search on series items.

For pattern-detection fluency, work through varied series families: arithmetic (constant difference), geometric (constant ratio), polynomial (squares, cubes, factorials), prime numbers, Fibonacci-like sums, alternating sub-sequences, and modular arithmetic patterns. The ICAR Letter and Number Series subtest items provide a public-domain corpus; competitive math-olympiad problems from the Art of Problem Solving website offer harder material.

Common errors on numerical series fall into a few categories: jumping to a multiplicative rule when an additive one fits, missing alternating sub-sequences, and failing to check candidate rules against multiple terms. Studying your own error patterns is more useful than just doing more practice.

Realistic expectations: practice can produce 0.5 to 1.0 standard deviation improvement on numerical-reasoning subtests over 30 to 50 hours of focused work. The improvement is durable for several months and partially transfers to closely related quantitative-reasoning tasks. Transfer to fluid reasoning measured by other formats (matrix items, rotation items) is small.