Analogies IQ test: what it measures and where to take one

0 The short answer

An analogies test measures relational reasoning, the ability to work out how two things are related and apply that relation to a new pair, which is induction (fluid reasoning, Gf) when the content is unfamiliar and also draws on word knowledge (Gc) when the terms are words. The best known analogy tests have faded: the College Board announced in 2002 that the SAT would drop analogies from 2005, the GRE removed them in 2011, and Pearson retired the Miller Analogies Test on November 15, 2023. ACIS offers two relevant subtests, Complex Relations, which asks for verbal analogies and sits in the Full Scale form (50 dollars, about 175 minutes), and Similarities, which is in every form from Quick (15 dollars, about 45 minutes); a buyer can also pick both in the Custom Test at 5 dollars each (prices read October 29, 2026). The decisive limit is that a word analogy score mixes reasoning with vocabulary and language exposure, so it should be read inside a profile and never as a stand alone verdict.

1 What does an analogies test measure?

An analogies test measures how well a person sees a relation between two things and applies it to a different pair, and in the Cattell-Horn-Carroll (CHC) model that skill is induction, one narrow ability inside fluid reasoning. The classic form is four terms, written A is to B as C is to D. In a research paper on the brain activity behind four term analogies, Krawczyk and colleagues describe the format as A:B::C:D and compare it with control tasks that required matching without analogical mapping. The point of the format is that the answer cannot be read off one pair. The solver has to decide what links the first two terms and then test that link against a second pair.

An original example, written for this page and not taken from any test: doctor is to hospital as teacher is to ___. Most adults see the link at once (a person who works in a place) and supply the missing term. An item at this level would not discriminate between adults. What a test builder controls is how hidden the relation is, how many relations have to be held together, and how rare the words are.

The CHC definitions help here. Kevin McGrew's CHC theory codebook from the Institute for Applied Psychometrics, read October 29, 2026, defines induction as the ability to observe a phenomenon and discover the underlying principles or rules that govern it. It defines lexical knowledge, a narrow ability inside comprehension-knowledge, as knowing the definitions of words and the concepts behind them. A word analogy asks for both at once: knowledge of the words, and induction of the rule that joins them. The page on the CHC model places both abilities among the broad abilities of the model.

The idea has a long history. Sternberg's 2022 review of basic processes in intelligence, in the Journal of Intelligence, recalls that Charles Spearman identified three processes he believed were the keys to individual differences, and that they can be seen in an analogy. Apprehension of experience is what we would now call encoding, reading the terms and understanding what they mean. Eduction of relations is an inferential process, recognizing the relation between the first two terms. Eduction of correlates is applying the inferred relation to a new term. Sternberg adds that these three processes may have been useful for analogies and other inductive reasoning problems but fell far short of a comprehensive model of intellectual processes.

Cognitive science supplies a second description of what is being measured. Gentner's 1983 paper on structure mapping, in Cognitive Science, proposes that in an analogy relations between objects, rather than attributes of objects, are mapped from a base to a target, and that the relations that get mapped are chosen by systematicity, meaning their membership in a higher order relational structure. Her account explains why a good analogy feels different from a similarity. The two terms of a good analogy do not need to look alike. They need to stand in the same relation.

Two practical consequences follow for the person choosing a test. First, an analogy test is not a vocabulary test, although vocabulary can limit the score. The ACIS technical manual makes the same separation for its own task: with two open slots rather than one, the rule linking the given pair must first be induced and then applied twice, so the item cannot be answered from how readily the words associate. Second, the format is one of several that sample induction. The ACIS manual describes Visual Number Series, Matrix Reasoning and Complex Relations together as induction tasks, with different content. The page on fluid intelligence tests treats analogies as one family among several, and a battery that includes several formats is more informative than a single quiz.

2 How does the mind solve an analogy?

Research describes analogy solving as a sequence of components, encoding the terms, inferring a relation, mapping it and applying it, and the best evidence is that better reasoners spend more time encoding and less time on the rest, not that they simply go faster. The framework most often cited comes from Sternberg's componential analysis. His original paper is Component processes in analogical reasoning, published in Psychological Review in 1977. His 2022 retrospective lists the components of inductive reasoning as encoding, inference, mapping of higher order relations, application, comparison between stimuli, justification of an answer that is better than the alternatives but imperfect, and preparation and response, the last estimated mathematically as a regression constant.

Three findings from that research bear directly on how a score should be read. The 2022 retrospective reports that in the analogy studies of 1977 and 1979 the encoding component took longer for better reasoners, who spent more time encoding so that they could operate more quickly on the encodings. It also reports that the information processing score with the highest correlation with fluid intelligence tests was the regression constant that Sternberg called preparation and response, a component that had been expected to be a throw away term. And it concludes that what mattered most was not overall speed but the allocation of resources, so that components needing more time obtained more time. A timed analogy test that rewards the fastest answer on every item would therefore miss part of what the research says separates strong reasoners.

The same review is candid about limits. Its abstract describes the search for basic processes as having had, at best, mixed success, and says that intelligence cannot be fully understood through any one conceptual or methodological approach. The component list is a model of what happens, fitted to response times in particular samples. It is not a validated checklist of what an individual did on an item.

Brain imaging gives a third description. Krawczyk, McClelland, Donovan, Tillman and Maguire published an fMRI study of the cognitive stages of reasoning by analogy in Brain Research in 2010. They divided solution into encoding, mapping and inference, and response phases. Activation for analogies was greater than for the matching controls at the encoding phase in several predominantly left lateralized and medial prefrontal areas, and a similar pattern held at mapping and inference, limited to the left and rostral prefrontal cortex. At the response phase the analogy condition produced the fastest and most accurate responses, which the authors read as most of the processing having occurred in earlier phases. It is a laboratory finding about a four term task and does not describe any commercial test.

Working memory enters through relational integration. Cho, Holyoak and Cannon ran experiments with a pictorial analogy task in which the number of relations to integrate, the need to resolve interference and the duration of maintenance were manipulated, and reported in Memory & Cognition in 2007 that performance fell when several relations had to be integrated rather than one, and fell further when integration was combined with interference resolution. They concluded that relational integration and interference resolution depend on shared resources in working memory. This is one reason a strong score on a hard analogy item is not purely a reasoning score. Processing speed is a separate domain, and the research above says speed alone is not what the format rewards.

The Standards for Educational and Psychological Testing ask for exactly this kind of evidence. Standard 1.12 says that if the rationale for a score interpretation depends on premises about the psychological processes of test takers, theoretical or empirical evidence in support of those premises should be provided. A test publisher who says an analogy subtest measures induction owes the reader that evidence, and the process literature above is where it comes from.

3 How is an analogy different from a logic puzzle or a pattern test?

An analogy asks you to discover a relation and carry it across, a logic puzzle asks you to derive what must follow from stated rules, and a pattern test asks you to find the rule in an arrangement, so the three overlap in fluid reasoning but are not interchangeable. The page on the logic IQ test owns deduction, reasoning from given premises to a conclusion that has to be true. The page on pattern recognition tests owns induction in general, the family that includes matrices and series. An analogy belongs to the second family, because the relation is not stated and has to be inferred, and it differs from a matrix in that the content is usually words rather than shapes.

The ACIS technical manual treats the same distinction as a reason to measure more than one format. It says that agreement across Matrix Reasoning, Figure Weights, Visual Number Series, Logic Grid and Complex Relations supports broad fluid reasoning, while disagreement suggests that format, quantitative content, general sequential reasoning or relation abstraction may be influencing the profile. In other words, a person can be stronger on word relations than on shape rules, or the other way round, and a single analogy score cannot say which. The page on abstract reasoning tests explains why an abstract reasoning label covers formats that overlap with IQ only in part.

The manual draws a second line, between analogies and the neighbouring verbal task. Complex Relations and Similarities both involve relations and abstraction, but the manual says the modalities differ: Similarities is verbal conceptual, and Complex Relations is broader induction over relations. Similarities asks how two things are alike and grades the answer by how general the shared category is. Complex Relations asks the solver to infer a rule from one pair and apply it to produce two missing terms. The first is a measure of how organized verbal concepts are. The second is a measure of rule induction that happens to use words.

That difference is also the difference between crystallized and fluid ability. Crystallized ability reflects what has been learned and stored, fluid ability reflects reasoning with novel material, and a word analogy sits on the border because it needs the first to deliver the material for the second. The page on fluid and crystallized intelligence compared sets out that border, and the sibling guide to the crystallized intelligence test covers the knowledge side in full.

4 Do verbal, numeric and picture analogies measure the same thing?

They share the relational core but not the load around it: word analogies lean on vocabulary, picture analogies lean on visual processing, and each format needs its own evidence before it is treated as a measure of reasoning. The CogAT shows the split inside one product. Riverside Insights' help pages, read October 29, 2026, list Picture Analogies at Levels 5/6 to 8 of the Verbal Battery and Verbal Analogies at Level 9 and Levels 10 to 17/18. The page describes the word version as examining a pair of words for how they are related and applying that relationship to a third word to form a new pair.

The knowledge confound is the central problem of the word format. The Educational Testing Service, in the compendium of research behind the revised GRE, says analogies and antonyms rely heavily on test takers' vocabulary knowledge, are short and are easily memorized. That is a statement by a test publisher about its own retired item types, and it matches what the Standards require of any test. Standard 3.2 makes test developers responsible for minimizing the potential for tests to be affected by construct-irrelevant characteristics, and lists linguistic and cultural characteristics among them.

Culture can change the keyed answer, not only the difficulty. A comment on Standard 11.11 in the Standards illustrates the point with an analogy whose correct answer depends on which seasoning people ordinarily use in a given culture. The lesson for a test taker is practical. A word analogy built for one population can mean something different to another, which is why the ACIS technical manual says its verbal and knowledge tasks have to be interpreted within the English speaking frame it was built for, and why the page on the verbal IQ test treats a verbal score as a description of performance in English.

Picture and number versions reduce the language load and add their own. A picture analogy asks for a visual transformation to be recognized, so it overlaps with visual processing as well as induction. Number analogies and series bring in quantity, and the page on the mental math test shows how much of an apparently numeric task is memory and calculation rather than relation finding. A task with no verbal and no numeric content at all, such as arranging blocks to match a design, measures a different ability again, and the sibling page on the block design test explains what that task adds.

The conclusion for buyers is simple. If the question is whether a person can see relations, a word analogy is a reasonable probe for a native reader of English and a poor one for a new learner of the language. If the question is broader, a battery that mixes formats gives a better answer than any one of them, and a task with no words is the cleaner choice where language has to be minimized.

5 Where did analogies appear in famous tests, and why did several drop them?

The SAT, the GRE and the Miller Analogies Test all used analogy items and have all left the format behind; the SAT and GRE publishers explained the change in terms of classroom fit, vocabulary and memorization, and Pearson gave no reason for the Miller Analogies Test. The record is short and checkable.

The SAT announced the end of verbal analogies in 2002. Education Week reported on July 10, 2002, in Overhauled SAT Could Shake Up School Curricula, that beginning in 2005 the test would drop two sections, quantitative comparisons and verbal analogies, following the College Board decision of June 27, 2002. A spokeswoman said the new test would be more closely aligned with classroom practices, and officials said eliminating analogies would put more emphasis on critical reading in subjects such as science, history and the humanities. The College Board's research report on the history of SAT content, by Lawrence, Rigol, Van Essen and Jackson (2003), adds that analogies are not covered in most high school English classes, that they would be replaced by more reading passage questions, and that the first administration of the new test was set for March 2005. The same report's table shows analogies at 19 to 20 questions of the verbal test in each period from 1958 to 2002. The page on how IQ relates to SAT scores covers the correlation between the two, and nothing in the change implies the earlier version was a poor test of general ability.

The GRE followed in 2011. The ETS compendium of research behind the revised test, edited by Wendler and Bridgeman, says that in August 2011 the program launched the GRE revised General Test, and that the antonyms and analogies of the prior test were removed from the Verbal Reasoning measure to reduce the effects of memorization and replaced with new question types. It describes the revision as placing more emphasis on complex reasoning with text based materials and less dependence on vocabulary out of context. It also reports a 2003 pilot study with a factor analysis whose results provided evidence that replacing analogy and antonym questions did not appear to change the construct being measured. The account is the publisher's own, and the compendium reports no post-launch comparison of the two designs. The page on IQ against GRE scores has the conversion question.

The Miller Analogies Test is the cleanest case of a test whose name was its format. Pearson's page on the test, read October 29, 2026, states that the Miller Analogies Test was officially retired on November 15, 2023, and is no longer available, and that official transcripts and personal score reports are no longer available as of November 15, 2024. The page gives no reason for the retirement, and this page does not supply one.

The format survives in a school age test, as the CogAT Verbal Analogies subtest described above shows. It is absent from the adult clinical battery whose subtest list was read for this page. The WAIS-5, according to the list of subtests in Pearson's overview brochure, has 20 subtests in five domains and none of them is named as an analogy subtest. Its Verbal Comprehension index uses Similarities and Vocabulary, and its Fluid Reasoning index uses Matrix Reasoning and Figure Weights. The page on every WAIS-5 subtest explains which of them build the Full Scale IQ.

6 What do analogy scores predict?

The strongest evidence on what analogy scores predict comes from one meta-analysis of one test, and it supports general cognitive ability as the predictor, not the analogy format. Kuncel, Hezlett and Ones published Academic Performance, Career Potential, Creativity, and Job Performance in the Journal of Personality and Social Psychology in 2004. According to its abstract, they examined the validity of the Miller Analogies Test for predicting 18 academic and work related criteria, and meta-analyzed its correlations with other cognitive tests such as Raven's Matrices and the Graduate Record Examinations. The results, they report, indicate that the abilities the MAT measures are shared with other cognitive ability instruments and that these abilities are generalizably valid predictors of academic and vocational criteria, of evaluations of career potential and of creativity.

Read carefully, that is a statement about general ability. The MAT was an analogy test, but the finding is that what it measured overlapped with what other cognitive tests measure, and that the shared ability predicts outcomes. It does not say that the analogy format adds something beyond g.

The limits are the usual ones. The abstract describes group level validity, so it says nothing about what an individual's analogy score means for that person's career. The abstract does not describe the samples, so how far the result transfers to the general adult population is not something this page can state. And the test at the center of the meta-analysis no longer exists, so no one can take it today to find out where they stand.

Readers who ask the question for a particular career, for example whether verbal relational skill matters for technical work, will find the occupational angle on the page about the IQ test for engineers. Readers asking whether a high analogy result says anything about giftedness will find the wider question on the page about a gifted test for adults. In both cases the honest answer starts with the same caution: one subtest is a small sample of ability.

7 What are the real options for taking an analogies test?

Today an adult can take an analogy based test from a psychologist who gives a battery without a dedicated analogy subtest, from a school ability test that the pages read do not describe for adult use, from free practice pages, from an employer's screening vendor, or from a normed online battery such as ACIS, and the options differ most in what they report. The table sets them side by side. Prices and times were read on October 29, 2026 and can change. The ACIS rows use the forms and prices on the ACIS home page of that date.

The clinician route is the reference. Pearson's WAIS-5 product page, read October 29, 2026, lists the test as published in 2024, with a qualification level of C and the administration times in the table. It does not publish what a clinician charges, so the table does not either. A clinician can also place a score in a history and a context that no online test can. The page on where to take an IQ test compares the full set of routes, supervised and unsupervised.

Free pages answer a different question. A free analogy page is useful as practice with the format, and it can be fun, but a count of correct answers is not a score on a scale. The page on free IQ tests sets out what a free result can and cannot support. The same checks apply to a paid product: does it name its norm group, does it report an interval, does it say what it does not measure.

ACIS is a seller in this market, so treat this section as a disclosure. Other online batteries exist, and the page on the best online IQ tests compares them without an analogy focus. The page on the IQ test for adults covers what a normed adult test reports at all. What the table shows is narrow: only the clinician route is supervised and clinically interpreted, and the dedicated analogy tests that carried the name in admissions have been retired or reshaped.

8 What do ACIS Complex Relations and Similarities measure?

Complex Relations asks for the missing terms of a verbal analogy and is classed as fluid reasoning, while Similarities asks how two things are alike and is classed as comprehension-knowledge, so together they cover the two sides of an analogy, the relation and the words. The descriptions below come from the two subtest pages and the technical manual, read October 29, 2026.

The Complex Relations subtest page describes the family as tasks that ask a person to combine several abstract relationships at once rather than solve one obvious rule in isolation, and says the challenge is coordinating multiple transforms, correspondences or analogical steps inside the same item without losing the structure. It places the task most closely with fluid reasoning, Gf. The manual gives the format: the examinee completes an abstract analogy by supplying both missing terms in a frame with two blanks, so the rule linking the given pair must first be induced and then applied twice, and the item cannot be answered from how readily the words associate. It lists 30 analogies at 45 seconds per item with typed answers, and maps the subtest primarily to induction. The page notes that the interactive task stays locked in the public version and that the public page does not disclose items, keys or scoring details, so this page does not either.

The Similarities subtest page describes a task that asks for the shared concept, class or principle behind two things that look different on the surface, and places it most closely with comprehension-knowledge, Gc. The manual's description is more specific: the examinee is shown two words naming objects or concepts and explains how the two are alike, the response is produced rather than selected, and credit is graded, with a superordinate category that captures both members earning full credit and a concrete or functional resemblance earning partial credit. It lists 31 pairs at 1 minute 30 seconds per item, scored 0 to 2 points per item, and maps the subtest to induction and lexical knowledge. The manual says the scoring distinction is what separates induction from lexical knowledge alone.

The two subtests meet in one composite. The manual describes the Verbal Reasoning Index (VRI) as focusing on verbal reasoning through Similarities and Complex Relations, identifying relations among concepts and abstracting shared meaning. The manual says its composites are available when their subtests are available, so a Full Scale report, which contains both subtests, can show it. A buyer who picks only those two subtests in the Custom Test, at 5 dollars each, gets a scaled score for each; the Custom Test description says every subtest returns its scaled score and any domain completed returns its full index, and that the Full Scale IQ needs the Full Scale battery. Whether a given purchase displays a composite is worth checking on the example reports before buying.

Which form contains which subtest decides what a report can show. Similarities is in Quick (15 dollars, six subtests, about 45 minutes), in Optimized (30 dollars, 13 subtests, about 110 minutes) and in Full Scale (50 dollars, 20 subtests, about 175 minutes). Complex Relations is in the Full Scale form and can be added through the Custom Test, and it is one of the seven subtests that exist only in the Full Scale form. The Full Scale form costs the same as ten Custom subtests, so a buyer who wants ten or more should buy the Full Scale form. The free trial needs no card and opens a different set of subtests, so this page does not rely on it for either of these two. The technical manual is the place to read the definitions and the documentation in full.

Complex Relations also belongs to the Fluid Reasoning Index (FRI), alongside Matrix Reasoning, Figure Weights, Visual Number Series and Logic Grid, and Similarities belongs to the Verbal Comprehension Index (VCI) alongside Antonyms, Vocabulary, Information and Synonyms. That placement matters for reading a result. A strong Complex Relations score beside a weak Similarities score is a different profile from the reverse, and the next section explains how the manual reads each.

9 How should an analogy score be read?

An analogy score is an estimate on a scale, it is more precise when it sits inside a battery than alone, and its meaning depends on the other scores beside it. An ACIS subtest returns a scaled score from 1 to 19, with a mean of 10 and a standard deviation of 3. A scaled score of 13 is one standard deviation above the mean, which on a normal curve is about the 84th percentile; that is our arithmetic on the scale, not a figure from any study. An index or composite is reported on the familiar scale with a mean of 100 and a standard deviation of 15, with a percentile and a 95 percent confidence interval. The page on why the standard deviation is 15 explains the scale, and the page on IQ score against percentile explains why the two cannot be averaged. The IQ percentile calculator converts a score to a percentile.

The confidence interval is the honest part of any report. A score is one draw from a distribution of results the same person could have produced on another day, and the interval states a range that likely contains the underlying value. The Standards for Educational and Psychological Testing make the related point about subscores. Standard 1.14 says that when interpretation of subscores, score differences or profiles is suggested, the rationale and relevant evidence in support of the interpretation should be provided, and its comment says the distinctiveness and reliability of the separate scores should be demonstrated. Reading one analogy subtest as a stand alone verdict about reasoning is an interpretation the Standards would ask for evidence about. The page on reliability and validity explains the two terms, and the technical manual documents the evidence for ACIS.

The manual offers rules for reading the profile that involve these two subtests. If Similarities is higher than the other verbal tasks, it says the profile may show strong abstraction despite ordinary lexical breadth. If Similarities is lower than Vocabulary and Information, the user may know words and facts but have more difficulty organizing them into shared conceptual relations. It adds that Logic Grid should be read with Complex Relations, Matrix Reasoning and working memory, so that a low score can be placed in context. These are descriptions of profile shapes, not diagnoses, and each depends on the scores carrying their own uncertainty.

How large a gap has to be before it means something is a separate question. The page on index score differences discusses how often a person's indexes differ, which is the context a single verbal or fluid score needs. Uneven profiles are common, and an uneven profile is not a defect.

The Complex Relations and Similarities scores answer different questions. A Complex Relations scaled score describes performance on verbal analogies inside a time limit, and the manual links it to induction. A Similarities scaled score describes how general the person's explanations of likeness were, judged against a scoring rule, and the manual links it to induction and lexical knowledge. Neither is a measure of intelligence on its own, and neither says why a score is what it is. A reader who wants the most they can get from them should treat the pair as a small, verbal slice of a profile and compare it with the nonverbal fluid subtests in the same report.

10 What conditions matter when you take an analogies test unsupervised?

An unsupervised session changes what a score means unless the conditions match what the norms assumed, and for a typed, timed analogy task the conditions that matter most are a keyboard, a quiet room and uninterrupted attention to each item. The ACIS Complex Relations subtest, as the technical manual describes it, is a typed response task with 45 seconds per item, and the Similarities subtest allows 1 minute 30 seconds per item with a produced, typed explanation. The subtest pages say no external aids are allowed, which means no notes, calculators, scratch paper or drawing tools. The Complex Relations page adds that the test moves on automatically when the countdown runs out.

Standardized conditions are part of what a norm means. The Standards for Educational and Psychological Testing say, in Standard 6.1, that test administrators should follow carefully the standardized procedures for administration and scoring specified by the test developer. Standard 6.3 says changes or disruptions to standardized administration should be documented and reported to the test user. In a supervised session an examiner does both. In an online session nobody does, so the test taker is the only person who knows whether a doorbell rang during item 12. A reader who notices a disruption should treat that subtest as less trustworthy and say so when interpreting the report.

The language of the test matters for the same reason. The ACIS manual says Vocabulary is language and culture sensitive and that its interpretation has to respect the English speaking norm frame, and it says Synonyms can be influenced by reading experience, language proficiency and exposure to the English register sampled. ACIS is English only. A person tested in a second language is measured partly on the language, and the construct-irrelevant variance that Standard 3.2 warns against enters through that door. The page on the cognitive ability test explains how a battery with several domains lets a reader see whether a verbal result is an outlier or the pattern.

Breaks are allowed between ACIS subtests, progress is saved, and a buyer has 30 days to complete the form. That is a practical advantage for a long form, and it does not remove the need to take each subtest in one sitting. For analogies the practical advice is to take Complex Relations and Similarities when alert, in a quiet room, on a device with a real keyboard.

11 Can you practice for an analogies test?

You can get better at the format and at the vocabulary it uses, but the published record on how much practice changes analogy scores is thin and partly contradictory, so any promised gain should be treated with suspicion. Two publisher statements point in opposite directions. The College Board's 2003 research report on the history of the SAT content, by Lawrence, Rigol, Van Essen and Jackson, says that antonyms and analogies were chosen for the early SAT because research of the time, by Coffman in 1962, indicated that they were less responsive to practice than some other question types. The ETS compendium on the revised GRE, decades later, describes analogies and antonyms as easily memorized and says they were removed to reduce the effects of memorization.

Those two statements are not a contradiction in the strict sense. The first reports an older finding that this page could not check at its source, since the report only cites it. The second reports what the GRE program judged to be happening in its own test in the 2000s. Neither is a controlled study of training, and this page did not find one it could verify. So it states no figure for how much practice raises an analogy score.

What can be said is narrower. Familiarity with the format reduces the cost of understanding the task, and the process research above suggests that the encoding step is where better reasoners spend extra time. Wider reading builds the word knowledge that a verbal analogy draws on, and the ACIS manual notes that reading experience influences lexical tasks. Neither of those changes the reasoning ability the format is meant to sample. The broader question of what training moves and what it does not is outside the scope of this page.

A last caution concerns repeating a test. A second attempt at the same format is unlikely to be an independent measurement, because the person remembers the structure of the first. Any gain on a retake mixes practice with real change, and a single analogy score cannot separate them. If the aim is to know where you stand, a first attempt on a normed battery is more informative than several attempts on a free page.

12 Where ACIS sits, and what a buyer should do

ACIS offers one verbal analogy subtest and one verbal concept subtest, normed for adults and scored on a published scale, and a buyer should treat them as a small verbal slice of a profile, taken online without supervision, never as a clinical or selection result. ACIS is a seller in this market, so check this section against the rest of the page.

The strengths are specific. Complex Relations asks for verbal analogies with two missing terms, is mapped by the manual to induction, and sits in the Fluid Reasoning Index. Similarities is graded on how general the shared category is and is in every form. Every form reports a Full Scale IQ with its 95 percent confidence interval, subtests report scaled scores from 1 to 19, and the norms cover ages 16 to 90. The Custom Test lets a buyer take only these two subtests for 10 dollars, and the Full Scale form at 50 dollars contains them with 18 others. Prices were read on the ACIS home page on October 29, 2026 and can change.

The limits are equally specific. The test is online and unsupervised, so no examiner controls the conditions. It is not a clinical or diagnostic instrument, and it is not designed for hiring, school accommodations or admission to high IQ societies. It is in English only, and the manual says verbal and knowledge tasks are affected by English proficiency and educational exposure. Complex Relations is a Full Scale or Custom Test subtest, so the Quick and Optimized forms do not contain it. One subtest score is less precise than an index, and an index is less informative than a profile read as a whole.

The Standards for Educational and Psychological Testing (AERA, APA and NCME, 2014), published jointly by the American Educational Research Association, the American Psychological Association and the National Council on Measurement in Education, are the guidance this page has relied on, and they are available as an open access edition. Four standards bear directly on how an analogy score should be read. Standard 1.12 asks for evidence when a score interpretation rests on claims about the cognitive processes of test takers. Standard 1.14 asks for evidence that subscores and profiles are distinct and reliable before they are interpreted. Standard 3.2 asks developers to minimize construct-irrelevant influences, linguistic and cultural among them. Standards 6.1 and 6.3 ask that administration follow standardized procedures and that disruptions be documented. That is why this page keeps telling readers to compare, not to conclude. The technical manual is where ACIS documents its evidence, and a buyer who wants more than this page can say should read the ACIS technical manual before paying.

On refunds, the ACIS guarantee reads: "Full refund within 5 days if the report does not deliver the features described at checkout or a technical issue prevents you from accessing your results." That is a conditional guarantee, not an unconditional one.

The concrete next step depends on the question. If the question is only how you handle verbal relations, the Custom Test with Similarities and Complex Relations is the most direct route at 10 dollars. If the question is how those scores compare with nonverbal reasoning, memory and speed, the Full Scale form at 50 dollars is the route, and the example reports show what each form returns before any purchase. If the question needs a supervised assessment, for example a school, a clinical or an employment decision, the right step is a qualified professional, and an online score can at most be a conversation starter.

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. Where a figure is our own arithmetic on a published scale, the text says so.

• Sternberg R. Component processes in analogical reasoning. Psychological Review, 1977, volume 84, issue 4, pages 353 to 378.
• Sternberg RJ. The Search for the Elusive Basic Processes Underlying Human Intelligence: Historical and Contemporary Perspectives. Journal of Intelligence, 2022, volume 10, issue 2, article 28.
• Gentner D. Structure-Mapping: A Theoretical Framework for Analogy. Cognitive Science, 1983, volume 7, issue 2, pages 155 to 170.
• Krawczyk DC, McClelland MM, Donovan CM, Tillman GD, Maguire MJ. An fMRI investigation of cognitive stages in reasoning by analogy. Brain Research, 2010, volume 1342, pages 63 to 73.
• Cho S, Holyoak KJ, Cannon TD. Analogical reasoning in working memory: Resources shared among relational integration, interference resolution, and maintenance. Memory & Cognition, 2007, volume 35, issue 6, pages 1445 to 1455.
• Kuncel NR, Hezlett SA, Ones DS. Academic Performance, Career Potential, Creativity, and Job Performance: Can One Construct Predict Them All? Journal of Personality and Social Psychology, 2004, volume 86, issue 1, pages 148 to 161.
• Wendler C, Bridgeman B (editors). The Research Foundation for the GRE revised General Test: A Compendium of Studies. Educational Testing Service, 2014.
• Lawrence IM, Rigol GW, Van Essen T, Jackson CA. A Historical Perspective on the Content of the SAT. College Board Research Report No. 2003-3, College Entrance Examination Board, 2003.
• Cavanagh S. Overhauled SAT Could Shake Up School Curricula. Education Week, July 10, 2002.
• Pearson Clinical Assessments. Important MAT Announcement, Miller Analogies Test page, read October 29, 2026.
• Riverside Insights. CogAT test descriptions, online help, read October 29, 2026.
• Pearson Clinical Assessments. WAIS-5 overview brochure, print code C12396 EL 2/25, and the WAIS-5 product page, read October 29, 2026.
• McGrew KS. CHC Theory Codebook 1: Cognitive Definitions. Institute for Applied Psychometrics, June 2014.
• American Educational Research Association, American Psychological Association and National Council on Measurement in Education. Standards for Educational and Psychological Testing. AERA, 2014.

14 Frequently Asked Questions