Do not commit to the first pattern that looks familiar. Write the changes between neighboring terms, check whether those changes repeat or change regularly, and test any proposed rule against every number shown. Then use it once to predict the next term. This makes your answer explainable, though no finite list can always prove that only one continuation is possible.
What are you actually trying to find?
A number-sequence question asks you to infer a rule from the terms provided and extend it. In reasoning research, this is called rule induction: the solver identifies a relation among examples and applies it to a new position. A study of number- and letter-series completion describes both as inductive tasks, while noting that people may use different strategies even when the underlying rules match. Number problems can invite quick retrieval of familiar arithmetic relations. That speed is useful, but the first relation you notice still needs checking. [Liang and colleagues’ study](https://pubmed.ncbi.nlm.nih.gov/24841995/) supports this description of the task and strategies; it does not test the checklist in this article.
Start with adjacent differences: subtract each term from the one after it. In the original illustration 4, 7, 10, 13, the differences are +3, +3, +3. If that constant change continues, add 3 to get 16. The useful observation is not simply that the sequence rises; it is that one stated operation explains every transition and predicts the next term.
This first pass keeps the work observable. If the differences are not constant, do not jump straight to a complicated formula. Check whether the differences themselves have a simple pattern, or whether a short cycle of changes could account for the list. You are gathering evidence for a rule, not proving that the author could not have meant anything else.
How can you test a rule before choosing an answer?
Use this sequence of checks: calculate adjacent differences; if they vary, inspect the differences for a simple progression; then consider ratios when division is meaningful, or a repeating alternation. State the candidate rule in plain language, replay it across every supplied term, and apply it one more time to produce the answer. A ratio is the result of dividing one term by the previous term; it can expose repeated multiplication, but it is unhelpful when division is undefined or the values make the ratio pattern needlessly awkward.
For example, take 5, 9, 7, 11, 9. Its changes are +4, −2, +4, −2. One compact explanation is a repeating two-step cycle: add 4, then subtract 2. The next change would be +4, giving 13. Check all four transitions against the cycle before accepting that continuation. This is a fresh teaching example, not an assessment item.
A number-series item-generation study offers a more limited lesson: researchers generated 92 items and administered them to 466 students, then examined item difficulty empirically. That is evidence about developing and evaluating items, not a description of the mental stages every solver uses. It also does not show that this checklist beats guessing in a controlled comparison. The study is useful here as a reminder that item difficulty should be checked with actual responses, and that a pattern that seems obvious to its writer may not work as intended. [The series-item study](https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2019.00884/full) reports its generated items, student sample, and difficulty analysis.
Under time pressure, stop when one simple rule fits every transition and yields a consistent next step. More searching can become counterproductive if each new candidate adds exceptions without explaining more of the data.
Sources: Automatic Generation of Number Series Reasoning Items of High Difficulty; Usefulness of Test Items That Involve Finding a Pattern in Data: A Reply
Can a short sequence prove there is only one answer?
No. A finite list does not specify its generating rule by itself. Many rules can agree on the displayed terms and then diverge at the next position. In a multiple-choice test, the choices and instructions add context about the intended convention, but they do not turn every plausible rule into a mathematical proof of uniqueness.
This ambiguity has also appeared in discussion of test items. The accessible abstract of an Educational Testing Service memorandum from 1955 addresses the criticism that more than one solution may fit a pattern-finding or number-series item. As the abstract presents it, the author argues that a defensible best solution may still be identified by applying reasonable rules, while conceding that eliminating plural solutions is desirable. The memorandum itself is not available in the accessible record, and the abstract describes an old argument about test-item usefulness; neither should be treated as a current universal standard for puzzle design. [The ETS record and its abstract](https://www.ets.org/research/policy_research_reports/publications/report/1955/ickg.html) support only the narrower point that ambiguity was recognized and that a best intended answer might still be sought.
So treat simplicity as a practical selection rule, not a theorem. Prefer the clearest rule that explains all the terms and follows the apparent conventions of the question. If two comparably simple rules survive the check, answer choices may indicate the expected one. If the sequence is open-ended and context does not break the tie, say that more than one continuation fits instead of presenting your choice as forced.
Sources: General fluid/inductive reasoning battery for a high-ability population
What can this method tell you about your reasoning?
It can make one answer traceable: you can identify the operation you used, see whether it covers the data, and review where a mistaken rule stopped fitting. It cannot turn one puzzle, or a few practice items, into an IQ score or a conclusion about fixed ability. An intelligence-test interpretation depends on the instrument, its administration, and evidence supporting the intended use.
The distinction is visible in formal test development. A peer-reviewed study designed a multi-item fluid and inductive reasoning battery for a high-ability population, including number series among several item types. Its three studies had samples of 3,067, 2,511, and 801; the researchers used item response theory, a statistical approach for calibrating how items perform across ability levels, to evaluate the battery and assemble parallel forms. Those samples and target population matter: this work does not establish how a single sequence performs for a general population. It also does not validate this site’s conversion or any other online quiz. [The accessible PubMed abstract](https://pubmed.ncbi.nlm.nih.gov/30478802/) reports the battery’s population, samples, and calibration; a lone sequence is not equivalent to that assessed battery.
For low-stakes practice, the free Test IQ Free assessment offers 50 questions across pattern, number, and spatial reasoning, with worked answers. Its displayed estimated IQ range is provisional: it comes from an unvalidated conversion with no validated adult reference population, so it is neither a confidence interval nor an established comparison with the adult population. Treat it as performance on this particular browser test under its conditions, not as identity, diagnosis, or a basis for consequential decisions. The optional paid report adds explanation, not validity.
The practical answer to ‘How do I solve it without guessing?’ is to make your candidate rule earn the next term: show the operation, check every transition, then acknowledge any ambiguity that remains. If you want more low-stakes practice, take the [free 50-question test](/test) and use its worked answers to review the reasoning steps. Practice can build familiarity with these operations; this article makes no claim that it raises general intelligence.
Sources: General fluid/inductive reasoning battery for a high-ability population
Questions readers ask
What should I check first in a number sequence?
Subtract each term from the next to see whether the adjacent differences are constant. If not, inspect how the differences change, then consider a simple ratio or repeating alternation. Verify any candidate rule against every term.
Can a number sequence have more than one correct next term?
Yes. A finite list can fit multiple rules that predict different next terms. In a test, use the simplest rule that explains all the terms and fits the stated conventions, while recognizing that this identifies a likely intended answer rather than proving uniqueness.
Does solving number sequences measure my IQ?
One sequence cannot establish an IQ score. A test score needs an instrument and supporting evidence for the interpretation. Test IQ Free’s estimated range is provisional and uses a conversion without a validated adult reference population; it is not a normed score or confidence interval.
Sources
- Different strategies in solving series completion inductive reasoning problems: an fMRI and computational study
Supports describing number-series completion as rule induction and noting that solvers use different strategies; it does not validate this article’s checklist.
- Automatic Generation of Number Series Reasoning Items of High Difficulty
Reports 92 generated number-series items administered to 466 students, with item difficulty examined empirically; it does not establish a sequence of cognitive stages or validate this article’s checklist.
- Usefulness of Test Items That Involve Finding a Pattern in Data: A Reply
The accessible ETS abstract—not the full 1955 memorandum—describes multiple solutions as a criticism of number-series items and argues that a defensible best answer may still be selected.
- General fluid/inductive reasoning battery for a high-ability population
The accessible abstract describes a multi-item fluid/inductive reasoning battery for a high-ability population, evaluated and calibrated using item response theory across three studies with samples of 3,067, 2,511, and 801.
Try the difference yourself.
Work through 50 adaptive reasoning questions, then see your estimated IQ band. Unlock a detailed personalized report after purchase.
Take the free test