How to Solve Number Series Reasoning Questions

What should you test first?

Start with the simplest relationship between neighboring terms: calculate what changes from one number to the next. A term is one entry in the ordered list. The first difference is the amount added to a term to get the next one; if that amount stays the same, it is the common difference. For 8, 13, 18, 23, write the terms on one row and the changes beneath them: +5, +5, +5. The repeated addition accounts for every visible move, so the next candidate is 28. This is more reliable than noticing only that the numbers rise: many unrelated rules can rise, while a repeated operation makes a specific prediction. If the differences are not constant, check whether each term is a fixed multiple of the one before it. The ratio is the result of dividing a term by its predecessor; when it repeats, it is called the common ratio. For 3, 6, 12, 24, the operation row is ×2, ×2, ×2, so the next candidate is 48. “Sequences” on Math Is Fun defines an arithmetic sequence by a constant difference and a geometric sequence by multiplication by a constant, which supports these as useful first checks. They are starting points, not a complete catalogue of possible patterns. A consistent subtraction or a fractional multiplier works by the same test. For example, 20, 15, 10, 5 has a difference of −5 at each step. And 16, 8, 4, 2 has a ratio of 1/2, because each term is half the preceding term. Negative changes and fractions do not make a rule irregular; the question is whether the same operation explains every visible transition. When dividing is undefined, as it is after a zero term, do not force a ratio calculation. Check differences or move to another simple structure. Keep the test falsifiable: write each operation, apply it across the whole list, then use it to predict the missing or next term. If even one visible transition disagrees, the candidate rule has failed. If neither constant differences nor a constant ratio explains all the steps, move on to regular changing differences or a simple repeated structure. Do not add exceptions just to preserve a first guess. A concise rule that covers the evidence is a better-supported continuation for a reasoning item; the short list alone may not prove it is the only possible continuation.

Sources: Sequences; Sequences FAQ

When do changing gaps form a usable rule?

When gaps between consecutive terms change regularly, calculate the gaps between those gaps. A repeated second difference supports a quadratic-pattern interpretation: extend that row, predict the next first difference, then add it to the last term. Treat this as a candidate rule, not proof that a short list has only one continuation.

Consider this original illustration: 4, 9, 16, 25. Subtract each term from the next to make the first-difference row: 5, 7, 9. Since those differences are not constant, repeated addition does not explain the sequence. Compare adjacent entries in that row: 7 − 5 = 2 and 9 − 7 = 2. The second-difference row is 2, 2. If it continues, the next first difference is 9 + 2 = 11. Add 11 to 25, and the predicted next term is 36.

First differences show how much the sequence changes at each step. Second differences show how much those first differences change. Here, terms rise by increasingly large amounts, and those increases themselves rise steadily. Growing gaps alone are not enough: they might increase by one, double, alternate, or follow another pattern.

The mathematical assumption is a quadratic sequence, whose terms follow a degree-two polynomial in their position. For such a sequence, second differences are constant. The open textbook chapter “Polynomial Fitting” explains finite differences and their relation to polynomial degree. In this illustration, second differences of 2 give a simple, coherent prediction. The chapter supports this mathematical relationship; it does not show that every reasoning item with growing gaps was generated by a quadratic rule.

Two observed second differences show consistency in the available row, but cannot establish the author's rule or rule out alternatives. More terms might confirm the pattern or break it. If answer choices are present, compare them with the prediction; a matching option can support an interpretation, but should not replace the calculation.

Finite differences can extend beyond the second row. “Polynomial Fitting” explains how higher-degree polynomial sequences can have constant differences at a higher level. That is valid mathematics, but it creates a trap: solvers can keep fitting a short list with more elaborate rules. A higher-order fit may reproduce every visible term yet give little reason to trust its next prediction. Without a clear cue or enough terms, do not escalate the table to force a pattern.

Use the lowest difference row that becomes consistently simple, then check its predicted term against the whole sequence. Here, first differences are not constant, while second differences repeat, so the second row earns a trial extension. If second differences are irregular, mark this candidate unsupported and return to other simple possibilities, such as a ratio, repeating operation, or interleaved tracks. Do not call increasing gaps a solved rule until you check how the gaps themselves change.

The opened guide “How to Solve Number Series Problems” presents a familiar progression from first differences to ratios, second differences, and alternation. It is a menu of checks, not evidence that one type is most common or that matching gaps prove uniqueness. Check second differences when first differences change regularly; their value is a testable prediction with few assumptions.

For this sequence, the constant second difference of 2 leads to a next gap of 11 and candidate term 36. Under a quadratic rule, not a guarantee that no other rule fits the same four terms. Keeping those claims separate makes the calculation useful without treating a convenient fit as certainty.

Sources: Polynomial Fitting; How to Solve Number Series Problems

How should you handle ratios, zero, and negative terms?

Use a ratio rule when each term is produced by multiplying the previous term by the same number. Divide each term by the one before it when defined, then see whether the quotients repeat. For example, 64, 32, 16, 8 has the same ratio at every step: multiply by one-half, or divide by 2. The next term under that rule is 4. This is a quick test of a different relationship from subtraction. In Sequences, the mathematical reference defines a geometric sequence by multiplying each term by a constant and calls that factor the common ratio. Check that the factor accounts for every adjacent pair, not just the first two.

Negative terms can still follow a constant ratio. Consider 2, −6, 18, −54. Dividing each term by its predecessor gives −3 each time: 2 × (−3) = −6, −6 × (−3) = 18, and 18 × (−3) = −54. The signs alternate because the repeated multiplier is negative; the rule itself does not change. Under that same rule, the next term would be 162. A negative value is therefore not evidence by itself that a sequence has become irregular. Check whether one operation continues to explain every transition. If it does, the sign changes belong to the pattern rather than requiring a second rule.

Zero needs more care because division by zero is undefined. If a proposed series includes a zero as the term you would divide by, you cannot calculate that adjacent ratio. Do not write down a quotient or treat the failed calculation as proof that no pattern exists. Switch to a check that remains defined, such as subtracting consecutive terms, or test a simple operation rule directly. A zero can also occur in a perfectly regular multiplication sequence: for a starting value a, repeatedly multiplying by zero gives a, 0, 0, 0, and so on. Sequences describes this special case of a geometric pattern. But a run of zeros by itself does not tell you that multiplication by zero was the intended explanation; other rules could produce the same visible terms. Treat it as a candidate and check the whole list.

Compare the two descriptions of 64, 32, 16, 8. The differences are −32, −16, and −8, so they are not constant. The ratios are one-half at each step, so repeated multiplication by one-half gives a shorter, consistent account. For a series such as 14, 9, 4, −1, the reverse comparison favors differences: subtracting 5 each time works, while the quotients do not repeat. These examples show why a solver should test both relationships instead of deciding from whether the numbers rise, fall, or cross below zero. The useful question is whether the same operation explains all the visible steps and predicts the blank without an extra exception.

When ratios fail because of zero, or because the quotients vary, return to differences or another simple candidate that fits the terms. A failed ratio check rules out that particular constant-ratio description for the visible transitions; it does not prove that the sequence has no rule. Write the operations beside every pair and compare them. If both ratio and difference rules seem to fit a short list, use the full item and its choices as checks. Choose the simplest rule that covers the evidence, and regard it as the best-supported intended answer rather than a mathematical certainty.

Sources: Sequences; Sequences FAQ

How can one rule mix operations or split into tracks?

When constant differences and ratios do not explain every step, try a repeating pair of operations or separate the odd- and even-position terms into two tracks. Keep either candidate only if it uses the same rule throughout and predicts the missing term. A compound rule links each term to the one before it through a repeated sequence of operations. An interleaved rule instead treats alternating positions as separate sequences. Both can explain a list, but each adds assumptions, so check them after simpler one-operation rules. Consider this original illustration: 3, 8, 18, 38, 78, __. From 3 to 8, multiply by 2 and add 2. The same pair takes 8 to 18, 18 to 38, and 38 to 78. Applying it once more gives 78 × 2 + 2 = 158. The useful evidence is not that “multiply and add” can be made to fit the first jump; many operations can do that. It is that the exact same ordered pair accounts for every displayed transition and makes one clear prediction. Write the operation beside each arrow: ×2, +2; ×2, +2; ×2, +2; ×2, +2. If one arrow needs a different addition or a reversed order, the proposed rule has failed. The commercial guide “Number Series Questions In Numerical Reasoning Tests” describes mixed sequences as using more than one rule and demonstrates the same kind of multiply-then-add procedure with its own example. That supports treating a short operation pair as a recognizable item format, not assuming it is the answer whenever first differences vary. The SkillVeris page “How to Solve Number Series Problems” also lists alternating rules among its suggested checks. These are instructional examples of conventions, not evidence that mixed or alternating items occur at any particular rate. For an interleaving check, label positions before calculating: 1st, 2nd, 3rd, and so on. Consider 3, 12, 5, 22, 7, 32, __. The odd-position track is 3, 5, 7, so it increases by 2; the even-position track is 12, 22, 32, so it increases by 10. The next blank is position seven, which belongs to the first track, giving 9. Positional labeling prevents treating the blank as if it must continue the most recent jump. Check both tracks separately: each has a consistent change, and the requested position is clear. Looking only at neighboring terms would show alternating jumps of +9, −7, +17, −15, +25, which obscures the simpler pattern. The tradeoff is straightforward. A compound rule keeps one continuous chain but requires repeating an operation sequence at every step. Interleaving avoids a changing chain of jumps by using two simpler tracks, but each track has fewer visible terms to check. Do not accept either merely because it produces a plausible next number. Ask whether the rule recurs in the same form, whether it explains all the terms, and whether it predicts the requested position without a special exception. If only one track fits cleanly while the other needs a new rule at each point, the split has not simplified the series. If both approaches fit, note the tie and use the item’s full wording or choices as a check, not as permission to invent extra operations.

Sources: Number Series Questions In Numerical Reasoning Tests; How to Solve Number Series Problems

What changes when a term is missing from the middle?

A missing middle term gives you evidence on both sides of the blank. Apply the same candidate rule into the blank from the left and out of it toward the right. If either neighboring step fails, reject the value or reconsider the rule. A final blank has no later term to check, so the proposed pattern must fit all earlier transitions. A fitting rule is a best-supported continuation, not proof that no other rule could fit. Write the visible terms in order and mark the blank as unknown. In 10, 14, __, 22, the changes suggest adding 4 each time. Continuing from the left gives 14 + 4 = 18; checking from the right gives 22 − 4 = 18. The same value satisfies all three links. This catches a guess that works for one pair but breaks the other. State the operation at each step rather than relying on a feeling that the terms “look” consistent. Now consider 3, 6, __, 24. Repeated doubling gives 12: double 6 to reach the blank, then double 12 to reach 24. Backward checking agrees, because half of 24 is 12 and half of 12 is 6. “How To Solve Number Series In Numerical Ability Tests” demonstrates a middle blank in a division-by-two series and applies the relationship across the sequence. “Sequences” defines a geometric sequence by multiplying each term by a constant. These sources support checking the same operation on both sides. The numbers here are fresh illustrations, not assessment items. Two-sided checking does not mean every rule relates only adjacent terms. A recurrence may depend on more than the immediately preceding value. If each term is the sum of the two before it, for example, test the missing position using the relevant earlier pair and then check the later terms. “Sequences” describes this recursive form and notes that it needs starting terms. When a simple difference or ratio fails to explain the list, do not force it merely because a blank appears; consider another simple relationship and require it to account for every visible position. Answer choices can test a candidate, but should not supply the rule. First identify a plausible operation from the terms, then substitute each option and check every transition the rule governs. If two choices fit different simple rules, the options have exposed ambiguity rather than proved which rule the writer intended. “How To Solve Number Series In Numerical Ability Tests” notes that more than one logical relationship can fit a pattern. Compare coverage and simplicity instead of inventing a complicated explanation after choosing a preferred option. Before moving on, state the operation, calculate into and out of the blank, and confirm the remaining terms. For a rule using earlier terms, name those inputs and verify each applicable step. A value that works on only one side has not passed. A value that works on both sides is best supported under that rule, while another rule may still produce a different continuation.

Sources: Number Series Questions In Numerical Reasoning Tests; Sequences

What if two rules fit the visible terms?

A short list of terms can fit more than one rule, so no method can always prove which continuation its author had in mind. For a reasoning item, choose the rule that accounts for every visible step, uses a small number of repeated operations, and predicts the requested term without a special exception. That makes it the best-supported intended answer under the item’s conventions; it does not make it the only continuation mathematics permits. Imagine the small sequence 1, 2, 4, __. Repeated doubling gives 8. But another rule can match the same three terms and give a different next value: add 1, then add 2, then add 3, which gives 7. Both descriptions cover what is shown, and each uses a simple operation. With only these terms, neither is ruled out by the arithmetic. More generally, a finite list does not specify its own generating rule: a person must choose a model or convention for extending it. The chapter “Polynomial Fitting” explains how polynomial rules can be fitted to finite data and why the chosen model matters. That point sets a boundary; it does not mean every aptitude item is defective or that every possible rule deserves equal consideration. Use three practical checks to compare candidates. First, coverage: does one rule explain each transition, or does it only fit the beginning and then require a patch? Second, economy: how many distinct operations and exceptions must be remembered? Third, prediction: does the rule produce a clear next term before you look for a convenient answer? These checks are a solving heuristic, not a theorem about all sequences or a validated measure of reasoning. They help rank interpretations when an item supplies limited evidence. In the example, doubling is compact and predicts 8; the increasing-addition rule is also compact and predicts 7. Because both survive the checks, the sequence alone cannot settle the answer. An additional term, an explicit instruction about the permitted pattern, or a design constraint could favor one. Answer choices can help test a candidate, but they cannot create evidence that is absent from the sequence. If 8 appears and 7 does not, that may reveal the item writer’s intended convention. It does not logically prove that doubling was the only possible rule. “Number Series Questions In Numerical Reasoning Tests” shows that preparation materials use several pattern families, including mixed and alternating operations; this supports checking plausible alternatives rather than assuming a single universal template. Still, do not generate elaborate rules just because they can be invented. If one economical rule fits all transitions and uniquely matches an option, use it. If two simple rules remain viable, mark the item as underdetermined in your reasoning, follow the test’s instructions, and move on if time or format requires. Recognizing that evidence has run out is different from making an arithmetic error.

Sources: Polynomial Fitting; Number Series Questions In Numerical Reasoning Tests

How should you pace a timed item?

The right pace depends on the test’s rules. If a timed format lets you skip and return, make a brief first pass: check each series for a repeated difference, then a repeated ratio, then regular changing gaps, and only after those consider a repeating pair of operations or two interleaved tracks. If no compact rule emerges, mark the item and continue when the interface allows it. If answers lock immediately or the assessment is untimed, follow those conditions instead of imposing a generic deadline. This order is a search procedure, not a claim that one family appears most often. First differences show the change from one term to the next. A constant difference supports an addition or subtraction rule; a constant ratio supports multiplication or division. If these vary, compare the differences with one another. A regular change there may support a second-difference pattern. Only then test whether a short operation pair repeats or whether odd- and even-position terms form simpler tracks. At each stage ask whether the candidate explains every visible step and predicts the blank. A rule needing an exception has not earned more time merely because the item feels difficult. A series may be difficult because the rule is unclear, arithmetic takes time, or the format restricts what you can do next. When allowed, jot an operation row beneath the terms to make checks visible. Use rough paper, a calculator, or other aids only when permitted. If answers lock, compare the simplest candidates and verify the proposed term instead of planning to revisit the screen. Instructions can change the advice materially. The UK government’s “Preparing for the Civil Service Verbal and Numerical tests” says its recruitment tests are untimed, adapt question difficulty to performance, and do not allow candidates to skip or go back. It recommends reading the instructions and answering every question. This shows why “skip and return” is not universal advice; it does not establish how other tests work. It says candidates may have rough paper, a pen, and a calculator, subject to that test’s directions. Read the directions for the assessment in front of you. “Number Series Questions In Numerical Reasoning Tests” from Practice Aptitude Tests describes timed numerical assessments and suggests estimating or making an educated guess when an answer is not apparent. Treat that as format-specific preparation guidance, not a guarantee that guessing is harmless. Check the instructions for unanswered items, penalties, and answer changes. Spend effort according to the interface. On a revisitable test, defer a stubborn item; on a locked response, make the best supported choice you can; on an untimed test, work carefully within the stated deadline. Pacing should protect accurate reasoning, not replace it.

Sources: Preparing for the Civil Service Verbal and Numerical tests; Number Series Questions In Numerical Reasoning Tests

What can your answer record tell you?

A practice log shows which operations you applied correctly under that session’s conditions. A count of correct answers is a raw score: performance on those items, without comparison to a population. A norm-referenced comparison interprets a score against a defined reference group. The Standards for Educational and Psychological Testing says normative data should describe the group and how and when it was gathered. Without an appropriate reference population, a raw count cannot establish a percentile or IQ. Neither measure describes your worth, identity, diagnosis, or fixed potential.

Keep the log small: note each item’s outcome, the rule you tried, and the error if you missed it, such as an arithmetic slip, overlooked alternation, premature guess, or failed verification. Record whether the format was familiar and relevant conditions, including timing or interruptions. This can guide the next practice task: check calculations after arithmetic slips; practise separating odd and even positions after missed alternation. That is an inference from your own work, not a population comparison.

Retesting in Selection: A Meta-Analysis of Coaching and Practice Effects for Tests of Cognitive Ability combined 50 studies, 107 samples, and 134,436 participants. It reported an adjusted overall retest effect size of .26; effects were larger with coaching and identical forms. The authors called for more research on implications for score-validity inferences. This evidence concerns cognitive-ability tests in selection contexts. It does not show that number-series practice raises general intelligence. Treat familiarity as context for your log, not proof of broader change.

Test IQ Free’s free 50-question activity offers reasoning practice and a provisional estimated IQ range from an unvalidated conversion with no validated adult reference population. Treat the range as provisional, not a normed comparison. The optional paid report adds explanation, not validity.

Sources: Standards for Educational and Psychological Testing; Retesting in Selection: A Meta-Analysis of Coaching and Practice Effects for Tests of Cognitive Ability

Practise one operation, then choose your next step

Choose one fresh series and solve it without looking at an answer key. Write the operation between each pair of terms, then check that the same rule accounts for every visible transition and predicts the blank. If it fails, record the error precisely: arithmetic slip, missed alternation, an untested gap, or a guess made before checking. Compare your reasoning with the worked explanation, then try another fresh example from the operation family that caused difficulty. This cycle gives practice a specific target; repeating an already familiar item mainly checks whether you remember that item’s route. Familiarity can affect performance on similar test formats, but the retesting meta-analysis indexed by PubMed examined cognitive ability tests across repeated occasions, not number-series lessons or gains in general intelligence. Use the record to decide what to practise next, not to label yourself. For low-stakes practice, you can take the free [50-question reasoning test](/test) and review its worked answers. Its estimated IQ range is provisional, generated by a conversion with placeholder parameters and no validated adult reference population. It is not a normed result or an established comparison with other adults. The optional $9 report adds explanation, not validity. If you use a score for a consequential decision, seek an appropriate assessment rather than relying on a browser exercise.

Sources: Retesting in Selection: A Meta-Analysis of Coaching and Practice Effects for Tests of Cognitive Ability; Preparing for the Civil Service Verbal and Numerical tests

Questions readers ask