The Mensa Norway IQ test: what it measures, how it scores, and what it leaves out

0 Quick Answer

The Mensa Norway IQ test is a 35 item figural matrix test with a 25 minute limit, scored from 85 to 145 within four age bands, and it is a reasonable indication of one ability, inductive reasoning on visual patterns under time pressure, with no published norms or reliability to tell you how precise the indication is. The test page, read on September 19, 2026, states all of that in its own words: the test gives an indication of general cognitive abilities, represented by an IQ score of between 85 and 145 where 100 is the population average, it is not a substitute for professional intelligence tests such as those administered by Mensa and licensed psychologists, the puzzles are weighted equally with one point per correct answer, there is no penalty for wrong answers, and the taker chooses one of four age ranges before starting. The copyright line credits Olav Hoel Dørum and Mensa Norway, 2023.

Mensa International hosts a version of the same engine as the Mensa IQ Challenge and describes it on its own page as provided for entertainment purposes only, with a score that will not qualify you for Mensa. Every national Mensa requires a supervised administration for admission, and the page on official IQ tests sets out what those administrations cost and produce.

What the test leaves out is everything that is not figural induction: verbal knowledge, quantitative reasoning, working memory, processing speed and visual spatial manipulation, which together with reasoning make up the six domains of a modern battery. It also leaves out the error band, because it publishes no reliability. A score of 130 on it is best read as strong performance on this kind of puzzle in this session, which is a real finding and a narrow one.

1 What the Test Is, in Its Own Words

Almost everything worth knowing about the Mensa Norway test is printed on its front page, and the claims it makes for itself are modest. The page describes the test as an online test that gives an indication of general cognitive abilities, represented by an IQ score of between 85 and 145, where 100 is the population average. It then disclaims itself in the next sentence: the test is not a substitute for professional intelligence tests, such as those administered by Mensa and licensed psychologists. The format follows: 35 puzzles in the form of visual patterns, a 25 minute time limit, no specialised knowledge or mathematical skill required, puzzles that get progressively harder and are weighted equally, one point per correct answer, no bonus for speed, no penalty for wrong answers. The page advises a ventilated room free from distractions and asks the taker to select an age range, with four choices: 16 to 17, 18 to 50, 51 to 60, and 61 and over.

The engine behind it is shared. Mensa International hosts the Mensa IQ Challenge on its own site with the same visual format and a language selector covering English, two forms of Chinese, French, German, Hungarian, Icelandic, Italian, Macedonian, Norwegian, Polish and Spanish, per the page read on September 19, 2026. Mensa International's framing is blunter than Norway's: the challenge is provided for entertainment purposes only, the score will not qualify you for Mensa, and a taker pleased with the result is invited to take a properly administered and supervised test through a local Mensa. The page on high IQ society requirements lists what the societies actually accept.

The item format is the one used by the Figure Reasoning Test, an instrument that several national Mensas, including Mensa Norway, use for supervised admission testing. The online test resembles that format without being the admission instrument, which is the whole reason it exists: a taker can find out whether the supervised session is worth booking. The page on what Raven's 2 is describes the modern commercial version of the same family of tasks, and the page on culture fair IQ tests explains why figural formats were adopted in the first place.

2 What a Timed Matrix Test Measures

A matrix test measures inductive reasoning with visual material, which is a central part of fluid intelligence and a strong indicator of the general factor, and it measures nothing else on purpose. Each item presents a pattern with a missing piece and asks the taker to infer the rule and choose the piece that completes it. The ability that does the work is induction: noticing regularities and extending them. In the Cattell Horn Carroll model, which the page on cognitive domains explains, induction sits inside fluid reasoning, Gf, alongside general sequential reasoning and quantitative reasoning. The page on fluid versus crystallized intelligence sets Gf against the knowledge based abilities that a matrix test deliberately excludes.

Matrix tests have a long reputation as the purest measures of the general factor, and the reputation is partly earned and partly overstated. Gilles Gignac examined the question directly in Intelligence in 2015, volume 52, pages 71 to 79, under the title Raven's is not a pure measure of general intelligence, and found that the Raven's matrices load on the general factor at a level comparable to other well constructed subtests rather than uniquely high, and that a single matrix test is neither a pure nor a sufficient measure of g. The practical reading is that a matrix score is a good single indicator and a poor whole assessment, which is why every professional battery includes one and none consists of one alone.

The ACIS technical manual makes the same point about its own Matrix Reasoning subtest. It describes the task as a central fluid reasoning task that is not the whole of Gf, and it strengthens the Fluid Reasoning Index by combining Matrix Reasoning with Figure Weights, Visual Number Series, Logic Grid and Complex Relations, because agreement across formats supports broad reasoning and disagreement reveals that format or content is influencing the profile. In the manual's confirmatory model, Matrix Reasoning loads .840 on the fluid reasoning factor and .774 on the general factor, with a subtest reliability of .930 and a standard error of 0.79 scaled score points. The five subtest Fluid Reasoning Index has a composite reliability of .9727 and a standard error of 2.48 points, and it loads .922 on the higher order general factor. Those numbers describe what a normed matrix task contributes to a battery, and they are the numbers the Mensa Norway test does not publish for itself.

3 How the Score Is Produced, and What the Range Means

The score is a raw count of correct answers out of 35, mapped to a number between 85 and 145 within an age band, and the mapping is the part the test does not show you. Equal weighting means the raw score is the number right. The absence of a guessing penalty means every item should be answered, and the page says so. The age band adjusts the mapping in some way for the four groups the taker can choose, which is a coarse form of the age norming that a professional instrument does in narrow bands; the page on how IQ scores are normed explains why band width matters, and the ACIS adult frame, for comparison, is built in age bands across 16 to 90 from 3,243 records.

The range is the most informative published fact. A scale that stops at 145 cannot distinguish a person who would score 145 from one who would score 160 on an instrument with a higher ceiling, and a scale that starts at 85 cannot distinguish 85 from 70. The page on the IQ score chart shows where those points fall: 145 is three standard deviations above the mean on a scale with a standard deviation of 15, the top 0.13 percent, and the IQ rarity calculator converts any score to its frequency. With 35 items, the number of distinct raw scores is 36, and the top of the scale is reached by a perfect or near perfect performance, which means the last few points of the range are separated by single items. A one item difference at the top of a 35 item test moves the reported score by several points, and the test page's own advice to manage time and guess is a recognition that single items decide the result.

The page does not state the standard deviation of its scale, its reference sample, or the reliability of its score. Without the first, a 130 cannot be placed against a Wechsler 130 with confidence, and the page on the 15 point standard deviation explains why the same rank maps to different numbers on different scales. Without the second, a percentile cannot be attached to the score. Without the third, no error band can be drawn around it. A normed instrument publishes all three because they are what turn a count into a measurement, and the section below on accuracy is written around their absence.

4 How Accurate Is It? What Can Be Said Without Published Figures

Accuracy has a technical meaning, the reliability of the score and the validity of its interpretation, and the Mensa Norway test publishes neither, so the honest answer is built from what is known about tests of its kind. The page on reliability and validity defines the terms. Reliability is the consistency of the score across items or sittings and sets the width of the error band. Validity is the evidence that the score means what it is used to mean. A 35 item figural test taken once under uncontrolled conditions has a reliability that can be estimated only by analogy, and the analogy points to a wider error band than a professional instrument.

The first source of error is the item count. Precision rises with the number of items, and 35 items with a single format is fewer than any full battery uses even for one index; the ACIS Fluid Reasoning Index, for comparison, draws on five subtests with different formats. The second is the conditions. The International Test Commission's guidelines on internet delivered testing, published in the International Journal of Testing in 2006, volume 6, issue 2, pages 143 to 171, classify an unproctored, unidentified administration as open mode and state that open mode scores support weaker interpretations than supervised ones. The test page's request for a quiet room is the only control the format allows. The third is the scale itself: with a floor of 85 and a ceiling of 145, error is compressed at both ends, and a repeat sitting by a high scorer tends to move toward the ceiling or away from it by the width of a single item.

What can be said in the test's favour is also specific. The format is a recognized one, the items get progressively harder as a well built figural test should, the scoring rule is transparent, and the disclaimer on the front page is more honest than most commercial competitors manage. The page on whether online IQ tests are accurate applies the same reading to the wider market, and the Mensa Norway test sits near the top of the free tier on that reading, for the reason that it does not pretend to be more than it is.

The comparison that answers the question is with a normed matrix task. ACIS's Matrix Reasoning subtest reports a reliability of .930 and a standard error of 0.79 scaled score points, which on the subtest scale means a score of 12 is best read as 10 to 14 at 95 percent confidence, our arithmetic on the manual's figure. A single format test with fewer items and open conditions should be assumed to be less precise than that, and the Mensa Norway page, by declining to print a figure, leaves the reader to make exactly that assumption.

5 Practice, Leaked Answers, and What They Do to a Score

Matrix tests are learnable, the Mensa Norway items have been explained item by item on video, and a score obtained after either exposure no longer means what the scale intends. Bors and Vigneau tested the effect of practice on Raven's Advanced Progressive Matrices across three sittings and published the result in Learning and Individual Differences in 2003, volume 13, issue 4, pages 291 to 312, documenting reliable score gains on repeated administration of the same figural format. The gain is not intelligence. It is familiarity with the rule types, which are finite: progression, distribution of elements, rotation, addition and subtraction of figures, and a handful of others. A person who has seen the rule types finds them faster, and a 25 minute limit rewards speed.

The Mensa Norway test has a specific exposure problem that a supervised instrument does not. Because it is free and popular, its 35 items have been worked through publicly in walkthrough videos and forum threads, and a search for the test returns pages promising the answers. A taker who has watched any of them is no longer measuring reasoning. Mensa International's testing FAQ, on its own page, gives the same reason for the society's move to computerised adaptive testing: paper tests may be reaching the end of their useful life because of the internet and the increasing likelihood that items are leaked online. The society is replacing its supervised instruments for this reason, and the free test is exposed to the problem in a way the supervised ones never were.

The consequence for a reader is a rule: take the test once, cold, or do not treat the result as a measurement. A second attempt is a practice measurement, and a third is a measurement of memory for the items. The page on how to prepare for an IQ test separates the preparation that preserves a score's meaning, sleep, a quiet room and understanding the instructions, from the preparation that removes it. Every normed battery limits retakes for the same reason, and the ACIS technical manual states that repeated practice attempts are not treated as equivalent to first valid completions.

6 Why the Score Does Not Qualify You for Mensa, Anywhere

No national Mensa accepts the Mensa Norway online score, including Mensa Norway, because admission requires a supervised administration and the online test is not one. Mensa International's FAQ states the constitutional rule: membership is open to persons who have attained a score within the upper two percent of the general population on an approved intelligence test that has been properly administered and supervised, and online tests cannot be used for admission. The rule is about conditions, not about the items. The same figural format, administered in a room with a proctor and an identity check, is an admission test in several countries; on a laptop at home it is practice.

What the score can do is tell you whether to book the session. A person who scores near the top of the online range under honest conditions has a reasonable prospect on a supervised figural test, and the sessions are cheap: American Mensa charged 60 dollars for a local group session and 99 dollars for private electronic testing on its testing page read on September 19, 2026, and British Mensa charged 49 pounds for a supervised test in person or online on its page the same day. The page on Mensa's IQ test describes what happens at a session, and the page on official IQ tests sets the Mensa route against the clinical and online routes.

The thing to know before booking is that the United States session is not the figural test. American Mensa's page describes an admission test of one to two hours, and its session uses different instruments from the Figure Reasoning Test that Mensa Norway administers; a strong figural score transfers to the extent that reasoning transfers across formats, which is substantial but not complete. A person whose strength is specifically visual pattern induction may do better on a figural admission test than on a battery with verbal and quantitative content, and the reverse is equally possible.

7 What a 130 on Mensa Norway Probably Means

A 130 on this test means that you solved enough of the 35 patterns within 25 minutes to be placed near the top of the scale for your age band, and everything beyond that sentence is inference. The inference runs in two directions. Toward the general factor, a strong matrix score predicts a strong score on other reasoning tasks, because induction is central to fluid reasoning and fluid reasoning is the domain that loads highest on g; in the ACIS model the Fluid Reasoning Index loads .922 on the general factor, the highest of the six domains. Toward the profile, a strong matrix score predicts nothing specific about verbal knowledge, working memory or processing speed, because those are different abilities with their own scores, and the page on what an IQ test measures explains why a battery reports them separately.

Regression toward the mean applies to every extreme score and applies more to a less reliable one. A score near a scale's ceiling contains more positive error than a score near its middle, because the people who reach the ceiling include those who had a good day. A retest on a different instrument tends to come in lower, not because the first test lied but because the error does not repeat. The page on the g factor explains why a general factor is estimated better from many indicators than from one, and the arithmetic is the same reason a Full Scale IQ is more stable than any single subtest. On ACIS, the Full Scale IQ composite has a reliability of .9886 and a standard error of 1.60 points, against 2.48 for the Fluid Reasoning Index and 0.79 scaled points for Matrix Reasoning alone, which is the precision gained by adding indicators.

The number to hold onto from a 130 is therefore not 130. It is the rank it implies on figural induction, the top few percent for the age band under time pressure, and the question it raises, which is whether the other domains sit at the same level. The page on the gifted IQ range explains what a full battery score in that region supports and what it does not, and the honest next step after a strong figural score is a measurement that covers the rest.

8 What the Four Age Bands Do to the Score

The test adjusts for age in four bands, and the width of the middle band, 18 to 50, is wide enough to change what a score means at either end of it. Reasoning of the kind a matrix test measures is not flat across adulthood. Timothy Salthouse asked when age related cognitive decline begins, in Neurobiology of Aging in 2009, volume 30, issue 4, pages 507 to 514, and concluded from cross sectional data on thousands of adults that measures of reasoning, speed and memory begin to decline in the third decade of life, well before the age at which most people notice anything. Park and colleagues had found the same shape for speed and working memory in Psychology and Aging in 2002, volume 17, issue 2, pages 299 to 320, with nearly linear declines from the twenties, while verbal knowledge held into the seventies.

A single band from 18 to 50 therefore compares a 48 year old with 19 year olds on a task where the 19 year olds hold the advantage. The page on whether IQ changes with age works through how a professional norm table handles this, with bands narrow enough that each person is ranked against contemporaries; the ACIS adult frame uses age bands across 16 to 90 for exactly this reason, and its technical manual states that processing speed, working memory and some visual spatial operations do not age at the same rate, so the norming is done within age. The effect of a wide band is systematic rather than random: a score near the top of the band's age range is likely an underestimate of that person's rank among their own contemporaries, and a score near the bottom of the range is likely an overestimate.

The 61 and over band has the opposite problem, because it has no upper edge. A 62 year old and an 85 year old are ranked against the same distribution, and the difference in expected performance across those 23 years is larger than the difference across any of the other bands. The 16 to 17 band is the narrowest and, for that reason, the best matched; it is also the band that Mensa International's FAQ warns about when it notes that many tests are not valid for people under sixteen.

None of this makes the four bands wrong for the test's stated purpose, which is an indication. It makes them a reason to read the score with the band in mind. A 130 at 49 in the 18 to 50 band and a 130 at 19 in the same band are not the same performance, and the first is probably the stronger one. A normed battery removes the question by ranking each person against a band a few years wide, and reports the band it used.

9 Speed and Power: What 25 Minutes Does to the Result

The Mensa Norway test is a speeded test whether or not it intends to be, and speed is a separate ability from reasoning that the format cannot separate out. Thirty five items in 25 minutes is about 43 seconds per item, with the later items designed to be harder. A taker who reasons well but slowly runs out of time on items they could have solved, and their score records speed as much as induction. The page on why IQ tests are timed explains the design choice, and the page on processing speed explains the ability that time limits import into any timed reasoning score.

Professional instruments handle the problem by separating the two. Pearson's WAIS-5 page, read on September 19, 2026, lists among its frequently asked questions why Visual Puzzles and Figure Weights have strict time limits and Matrix Reasoning does not, which reflects the design decision to measure matrix reasoning as a power task, where the question is whether you can solve the item, rather than a speeded one. The scale then measures speed on its own subtests and reports it as a separate index. ACIS follows the same architecture: its Processing Speed Index is built from Symbol Search and Coding and correlates .475 with the Working Memory Index and less with the reasoning domains, per the technical manual, and its reasoning subtests are timed generously enough to function as power measures within a session limit.

The practical reading for a Mensa Norway taker is that a score obtained by running out of time on the last ten items is a lower bound on reasoning and an upper bound on nothing. A taker who finishes with time to spare has a score that reflects reasoning more cleanly. The test does not report which of the two happened, and the taker usually knows.

10 The Mensa Norway Test Against a Normed Battery

Set the two side by side and the difference is not that one is right and the other wrong; it is that one measures one ability in one format with no published error, and the other measures six domains with twenty formats and prints the error next to every score. The table below uses the test page's own statements for the Mensa Norway column and the technical manual and order page for the ACIS column, read on September 19, 2026.

The last row is the one to read twice. Neither instrument is accepted for admission, and the difference between them is not in acceptance but in measurement. A reader who wants admission needs a supervised session. A reader who wants to know their standing needs a measurement with a published error, and the page on the best online IQ tests ranks the options on that criterion. A reader who wants to know whether they are good at matrix puzzles already has their answer from the free test.

11 How to Use the Free Test Well

Used once, cold, and read as an indication of figural reasoning, the Mensa Norway test is a good free tool, and every misuse of it comes from reading the number as more than that. Take it in the conditions the page asks for, a quiet room and 25 uninterrupted minutes, without having watched any walkthrough, and answer every item because there is no penalty. Record the score with its age band and the date. Then treat it as one indicator of one ability and ask what the rest of the profile looks like.

Do not retake it to improve the number, because the second number measures practice. Do not compare it across scales without knowing the standard deviation, which the page does not state. Do not report it as an IQ to anyone who will act on it, because the page itself says it is an indication and not a substitute. And do not take a strong score as a reason to skip a measurement of the other domains, because the domain a matrix test measures is the one most people already suspect they are good at when they seek out a matrix test.

The page on free IQ tests and the page on free versus validated IQ tests give the general checklist for reading any unnormed score, and the page on nonverbal IQ tests explains what figural formats gain and lose relative to batteries with verbal content. The page on IQ test questions shows the other item types a full battery uses, which is the fastest way to see what 35 matrices leave out.

The Cognitive Metrics catalog review offers a comparison with broader batteries while preserving differences in norms, purpose and supervision.

12 Where ACIS Sits, and What a Buyer Should Do

We sell a paid assessment that includes a matrix reasoning subtest, so treat this section as a disclosure and check it against the rest of the page. ACIS is not accepted for Mensa admission either, and it does not claim to be. What it offers relative to the free test is breadth and error reporting: 20 subtests across six domains, an adult reference frame of 3,243 records in age bands from 16 to 90, a confirmatory factor model fitted on 2,750 complete records with a comparative fit index of .9761, a Full Scale IQ with a composite reliability of .9886 and a standard error of 1.60 points, and a reliability and standard error printed for every index and subtest in the technical manual, version 1.4, updated August 3, 2026. Its Matrix Reasoning subtest has a reliability of .930 and a standard error of 0.79 scaled score points and sits inside a five subtest Fluid Reasoning Index with a standard error of 2.48.

The prices are one time: 15 dollars for the Quick form of six subtests, 30 for the Optimized form of thirteen, 50 for the Full Scale form of twenty, and a per subtest price for the Custom form shown on the order page. The trial includes two subtests for Quick, four for Optimized and five for Full, without a card, and lasts seven days. Paid access stays open for 30 days, and the quality guarantee is a full refund within five days if the report does not deliver the features described at checkout or a technical issue prevents access; it does not cover disliking the score.

The limits are the ones this page has applied to the free test, stated for ACIS. It is self administered, which the International Test Commission would call controlled mode, and its scores rest on integrity controls, retake limits and completion requirements rather than a proctor. It is normed on English speaking adults. One subtest, Spatial Navigation, was retired on September 7, 2026 and replaced by Layer Rotation in the Visual Spatial Index, and the manual's notice says the affected tables will be reissued with the next edition, which a buyer should read before interpreting that index.

What a buyer should do is the same in every case. If you have taken the Mensa Norway test once and want to know what the rest of your profile looks like, the Full trial offers five subtests in working memory and processing speed, with scores, percentiles and eligible composites. Matrix Reasoning is part of the paid assessment, so the trial cannot provide a matrix score to compare with your Mensa Norway result. If the question is admission, book a supervised session with your national Mensa, which costs less than most people assume. If the question is whether you are good at matrix puzzles, the free test has answered it, and no purchase is needed.

13 Sources Behind This Page

Every figure above is traceable to one of the following, and each is linked at the point where it is used. The test format, scoring rules, age bands and disclaimers were read on the Mensa Norway and Mensa International pages on September 19, 2026 and may change. The 95 percent intervals quoted for ACIS subtest scores are our arithmetic on the reliabilities in the technical manual, using the standard formula, and are labelled as such.

• Mensa Norway. IQ Test Made by Mensa Norway, test description, scoring rules and age ranges. test.mensa.no, read September 19, 2026.
• Mensa International. Mensa IQ Challenge, entertainment disclaimer and supported languages. mensa.org, read September 19, 2026.
• Mensa International. Getting your IQ tested: FAQs, admission rule, online tests and computerised adaptive testing. mensa.org, read September 19, 2026.
• American Mensa. Take the Mensa Admission Test, fees and session format. us.mensa.org, read September 19, 2026.
• British Mensa. IQ testing and puzzles, supervised test fee. mensa.org.uk, read September 19, 2026.
• Gignac G E. Raven's is not a pure measure of general intelligence: Implications for g factor theory and the brief measurement of g. Intelligence, 2015, volume 52, pages 71 to 79.
• Bors D A and Vigneau F. The effect of practice on Raven's Advanced Progressive Matrices. Learning and Individual Differences, 2003, volume 13, issue 4, pages 291 to 312.
• Salthouse T A. When does age-related cognitive decline begin? Neurobiology of Aging, 2009, volume 30, issue 4, pages 507 to 514.
• Park D C and colleagues. Models of visuospatial and verbal memory across the adult life span. Psychology and Aging, 2002, volume 17, issue 2, pages 299 to 320.
• International Test Commission. International Guidelines on Computer-Based and Internet-Delivered Testing. International Journal of Testing, 2006, volume 6, issue 2, pages 143 to 171.
• Pearson. Wechsler Adult Intelligence Scale, Fifth Edition (WAIS-5), product page including the FAQ on time limits for Matrix Reasoning. pearsonassessments.com, read September 19, 2026.
• ACIS. Technical manual, version 1.4, updated August 3, 2026: subtest technical table, reliability tables, confirmatory factor analysis output and retake policy. acisiq.com/technical-manual.

14 Frequently Asked Questions