The diagrammatic reasoning test: what it measures and what retests change

0 The short answer

A diagrammatic reasoning test is an employer aptitude test in which a candidate infers rules from shapes, symbols or flowchart style steps, and it mainly samples fluid reasoning (Gf) with a working memory load, so it is not an IQ test. The publisher pages we opened do not agree on the name or the format: Test Partnership describes flowchart style questions in which a shape passes through a series of operators, while SHL's fact sheet describes an inductive test with shapes and alphanumeric sequences, 18 minutes and no more than 15 questions. An adult who wants their own profile on adult norms can take ACIS, which has no flowchart or operator task, at 15, 30 or 50 dollars for about 45, 110 or 175 minutes (prices read October 29, 2026), and the decisive limit is that ACIS is not preparation for, a predictor of or a substitute for any employer test.

1 What Is a Diagrammatic Reasoning Test?

"Diagrammatic reasoning test" is a label that test publishers attach to timed puzzles built from shapes and symbols, and the label is a commercial convention more than a separate psychometric category. On an academy page that shows an update date of July 6, 2026 (read October 29, 2026), Test Partnership, a vendor of employment tests, says that abstract, inductive, diagrammatic and non-verbal reasoning are four labels for essentially the same kind of test. The page states that "The naming difference is a commercial one, not a psychometric one" and that the tests measure fluid intelligence (Gf). The page is promotional, because it ends by recommending the vendor's own inductive reasoning test, so we use it for how a vendor describes its field and not as independent evidence.

A second publisher shows that even one company uses the word for more than one product. Saville Assessment's aptitude tests page (read October 29, 2026) describes its aptitude assessment as "Verbal, numerical and diagrammatic reasoning, matched to the level you are hiring at." Its practice tests page lists a practice test for Diagrammatic Analysis under its Analysis Aptitude Range and a practice test for Diagrammatic Reasoning under its Technical Aptitude Range. Neither page says what the item format is, how many items there are or how long the test lasts, so a candidate who reads only those pages learns the names and not the task.

The word also has an older meaning that is easy to confuse with the test. Larkin and Simon (1987) distinguished diagrammatic from sentential representations of information in Cognitive Science. In their account, sentential representations are sequential, like the propositions in a text, and diagrammatic representations are indexed by location in a plane. They argued that a diagram often shows information that a sentence leaves implicit, and that problem solving with a diagram can proceed through a smooth traversal with very little search. That is a claim about how information is represented, not about a hiring test. Employers use the word more loosely, for reasoning problems whose content happens to be drawn.

Two points follow for a candidate. The first is that the name on the invitation does not tell you the format, and the invitation or the publisher's candidate guidance is the only reliable source of it. The second is that a diagrammatic reasoning test is not an IQ test. It is one timed sample of reasoning on drawn material, scored against whatever comparison group the publisher chose. The page on how abstract reasoning tests relate to IQ sets out the same point for the closest label, and this page does not repeat that argument.

2 What Formats Do Publishers Describe?

Publishers describe two formats under this label, a flowchart in which a figure passes through operators and a sequence in which shapes or symbols follow a hidden rule, and the pages we opened give each only a sentence or two. The flowchart format is described by Test Partnership as "flowchart-style questions where a shape passes through a series of operators", with rotating, flipping and resizing named as examples of what an operator can do, and the page contrasts this with a standard sequence completion question. The vendor attributes the flowchart format particularly to SHL's version and says the candidate has to trace what the shape looks like at the end. The SHL pages we read describe an inductive test and do not use the word diagrammatic, and we found no description of the flowchart format on a Saville or SHL page. What follows about how such items work is our own task analysis from that one vendor sentence, and it is labeled as such.

The sequence format has better documentation. SHL's cognitive assessments page (read October 29, 2026) describes its inductive reasoning test as asking candidates to work out the pattern in a set of images and identify the correct next image. The Verify Interactive Inductive Reasoning fact sheet, copyrighted 2022 and read October 29, 2026, says the test "presents candidates with information presented as shapes and alphanumeric sequences" and measures the ability to infer rules, patterns or sequences and then use them to predict outcomes. SHL does not call this test diagrammatic in either document, which is why this page treats it as the nearest published description and not as the diagrammatic test.

Our reading of what a flowchart item demands, and not a publisher's statement, has three steps. The candidate must work out, from the examples shown, what change each operator applies to a figure. The candidate must then apply those changes in the order the chart specifies. Finally, the candidate must either predict an output or work backward to what an operator must have done. The first step is induction, because the rule is not given. The second is sequential reasoning, because the order of the steps matters. Both steps place information in working memory, which section six takes up. Whether a given product asks for the output, the input or the operator is something only the publisher's own guidance can say.

The sequence format makes the same demands in a different order: induce a rule from the positions shown, then apply it to the next position. That is the inductive reasoning format, and the page on the inductive reasoning test in this series owns it. The page on the deductive reasoning test owns the other half of the pairing, reasoning from stated premises to a conclusion that must follow. Readers whose search was really about shape matrices and odd one out puzzles will find those formats described on the page about the pattern recognition test, and the number version of the sequence format appears on the Visual Number Series subtest page.

3 Which Roles Use It, and Under What Conditions?

The publisher pages we opened place these tests in hiring for professional and technical roles, and they say more about the level of the job than about its field. SHL's fact sheet describes the Verify Interactive Inductive Reasoning test as an assessment for job candidates applying to mid to senior level positions that require inductive reasoning. It lists troubleshooting, creative problem solving and innovative idea generation as sample tasks and names Product Development, IT, Professional Services and Management as potential roles. Saville's aptitude tests page describes its aptitude assessment as "Used across entry roles and early careers through to professionals and managers." Neither page names engineering. A candidate who is an engineer and wants to understand scores in that field will find the evidence on the page about the IQ test for engineers, which separates quantitative, visual spatial and fluid reasoning scores.

The conditions of sitting are more specific. SHL's fact sheet gives an allowed time of 18 minutes, a maximum of 15 questions, one sitting and a design for an unsupervised environment. It describes the question format as interactive and adaptive and says the test can be deployed on any device type, including mobile phones. The listed benefits of adaptive technology are shorter testing times, the ability to offer the test unsupervised and "a more precise score". Saville's practice tests page says most of its tests can be completed on a mobile phone and lists Diagrammatic Analysis among the exceptions. Saville's FAQ on its aptitude page says candidates receive the same number of items at the same difficulty levels, drawn from a randomized bank. These are statements about different products from different publishers, and the table in the next section keeps them apart.

Two practical consequences follow. A short, timed, adaptive test cannot be treated like an untimed puzzle: the clock is part of what is measured, and the page on why IQ tests are timed explains in general terms what time limits add and what they remove. And the result belongs to the hiring process. The fact sheet and the Saville pages are written for the organization that buys the test, and neither tells a candidate what they will see afterwards. The wider category of employer cognitive tests, including the numerical and verbal formats that often arrive in the same invitation, is covered in the guide to the pre-employment cognitive test, and a different publisher's battery is covered on the page about the Korn Ferry test.

4 How Does Taking One Compare With the Other Ways to Be Tested?

A candidate has four real ways to be measured on reasoning with diagrams, and they differ most in who supervises the session, who chose the comparison group and who receives the result. The table holds only what we could read on a publisher or vendor page on October 29, 2026, and where a page is silent the cell says so, because a guessed cost or time would be worse than a blank. The employer row combines what two publishers state for different products and should not be read as one test.

The WAIS-5 row comes from Pearson's product page, which describes an individually administered clinical instrument and gives the times and the qualification level shown. The practice row comes from Saville's practice tests page, which says the practice tests exist to familiarize candidates with the format and with working to strict time limits, that candidates can complete the questions as many times as they like and that the difficulty of the example questions does not necessarily represent the difficulty level of the test itself. Saville's cognitive ability page adds that practice tests and worked examples are free.

Commercial sites that sell diagrammatic practice packs also exist. We did not use them as sources and do not link them, because a seller of practice material has a stake in how much practice seems to matter. The practical order for a candidate is the publisher's free practice first, then the invitation. Readers who want the full map of places that give a real IQ score, including supervised routes, will find it on the page about where to take an IQ test, and a ranking of online batteries is on the page about the best online IQ tests.

5 Which CHC Abilities Does a Diagrammatic Reasoning Test Draw On?

A diagrammatic reasoning test draws mainly on fluid reasoning (Gf), specifically induction and sequential reasoning, with working memory, visual processing and speed contributing in ways that vary by item and by publisher. In the Cattell-Horn-Carroll framework described on the page about the CHC model, Gf is the ability to solve unfamiliar problems, and its narrow abilities include Induction (I), which is discovering underlying rules or patterns from specific instances, and General Sequential Reasoning (RG), which is logical step by step reasoning. A test in which the rule of each operator or each sequence must be found, and then applied in order, maps onto those two. The worked families of Gf tasks are laid out on the page about fluid intelligence tests, and this page does not repeat them.

The table separates what each demand is, which ability it points to, and how firm the link is, because several of the links are our reading of a one sentence format description and not a published finding.

Visual processing deserves a sentence of its own. Operators that rotate, flip or resize a figure are spatial transformations, and a candidate who is strong at them has an advantage that is separate from rule finding. The page on the spatial reasoning test in this series owns that format, and a profile with a high fluid score and a lower visual spatial score would be a reason to read the two separately. Processing speed matters because every product described here is timed. How much of a diagrammatic score is speed varies by test, and we found no primary estimate for the products employers use. The construct itself is explained on the page about the Processing Speed test.

Two abilities are less central than candidates fear. Crystallized knowledge (Gc) is not what the task rewards, because the content is drawn and rules are discovered from the item, although the instructions are written in language and a candidate who reads them slowly pays a price that is not a reasoning price. Quantitative knowledge (Gq) enters only when the sequences contain numbers. The ACIS page for Figure Weights describes its balance scale tasks as sitting between Gf and Gq for that reason, and the same logic applies to any numeric sequence.

6 Why Does Working Memory Load Matter?

Working memory load matters because a diagram problem is solved by holding rules and intermediate results in mind while the next step is worked out, and the research ties working memory to reasoning strongly but not completely. The best account of this for a matrix test is Carpenter, Just and Shell (1990). They analyzed the processes in the Raven Progressive Matrices Test using verbal protocols, eye fixation patterns and errors, and concluded that the processes distinguishing individuals were primarily the ability to induce abstract relations and the ability to dynamically manage a large set of problem solving goals in working memory. That is a finding about a matrix test, not about flowcharts. Our inference is that a chart with several operators asks for the same two things, and that each additional operator or intermediate figure adds to what must be managed.

A second line of work explains why complexity, and not only length, drives difficulty. Halford, Wilson and Phillips (1998) defined working memory limits in terms of the complexity of the relations that can be processed in parallel, which they define as the number of related dimensions or sources of variation. Their review of studies of working memory limits suggests a soft limit corresponding to the parallel processing of one quaternary relation. More complex concepts, they argue, are handled by segmentation, in which a task is broken into components processed serially, or by conceptual chunking, which reduces load at the cost of making some relational information inaccessible. For a candidate the implication, which is ours and not the authors', is that operators that act one after another can be handled step by step, while operators whose effects interact must be held together, and the second kind should be harder for the same number of symbols.

How strong is the link between working memory and reasoning? Two papers from the same 2005 issue of Psychological Bulletin give two answers, and both are best read as measuring different things. Ackerman, Beier and Boyle (2005) meta-analyzed 86 samples relating working memory to intelligence and reported that the average correlation between true score estimates of working memory and g is .479, "substantially less than unity", so the constructs are related but not identical. Kane, Hambrick and Conway (2005) agreed that working memory capacity is not isomorphic with fluid intelligence or reasoning, but reanalyzed 14 data sets from 10 published studies, representing more than 3,100 young adults, and found a median correlation of .72 between working memory and Gf or reasoning factors, which they describe as about 50 percent shared variance. The two figures differ in what was correlated: g in one case, Gf and reasoning factors from latent variable studies in the other. The page on fluid intelligence tests, linked earlier, presents that dispute at more length.

For the person who sat or will sit the test, three conclusions follow, and each stays within what the sources support. A low diagrammatic score does not by itself say whether rule finding or working memory was the limit. A separate working memory result narrows the question, which is why a profile is more informative than one score. And no source we opened measured the working memory share of any employer diagrammatic product, so the honest summary is that the link is real, partial and untested for these specific tests. The ACIS page for Logic Grid makes the same caution about its own task, noting that a weaker result may reflect working memory load and not a general reasoning deficit, and the full explanation of the construct is on the page about the working memory test.

7 What Does Retest Research Show?

Scores on cognitive ability tests rise when people take them again, by about a third of a standard deviation on the first retest, and the rise levels off after about the third attempt, so a higher second score is not proof of higher ability. The two largest syntheses are by Hausknecht, Halpert, Di Paolo and Moriarty Gerrard (2007) and by Scharfen, Peters and Holling (2018). Hausknecht and colleagues note that as many as 25 percent to 50 percent of applicants in organizational and educational settings are retested with measures of cognitive ability. Their meta-analysis of 50 studies covered 107 samples and 134,436 participants and found an adjusted overall effect size of .26, with larger effects when practice was accompanied by test coaching and when identical forms were used. They also wrote that more research was needed on what retesting does to the validity inferences drawn from scores.

Scharfen and colleagues extended the evidence to several attempts. Their literature search found 174 samples from 122 studies, 786 test outcomes and an overall sample of 153,185. A multilevel meta-analysis found a standardized mean change of 0.33 from the first to the second administration, 0.50 from the first to the third and 0.52 from the first to the fourth. The gain from the third to the fourth was 0.02 and was not significant, which is the basis for their conclusion that there are no further score gains after the third administration. The authors illustrate the first figure on an IQ scale: a person scoring 100 on a first test would be predicted to score 105 on a second, 107 on a third and 108 on a fourth.

Four details keep those averages honest. The first is spread: the authors report a standard deviation of the true effect of 0.32 around the 0.33, which gives a prediction interval from minus 0.30 to 0.96, so some people gain a lot and some do not gain. The second is the form. Alternate test forms showed significantly smaller retest effects than identical forms, by 0.15 for the first to second administration and by 0.20 for the first to third. The third is time: the effect fell by 0.0008 per week of interval, which the authors call weak, so waiting a year removes only about 0.04 standard deviations. The fourth is that the study pools many cognitive tests, and we did not find a meta-analysis restricted to diagrammatic products. In their four administration studies, identical forms were used most often for figural tests, which is a reminder that the data on drawn material come from repetition of the same test more than from fresh versions.

Working memory tests show the same pattern. Scharfen, Jansen and Holling (2018) found 234 effect sizes from 95 samples and 68 studies and reported a weighted average of g = 0.28 from the first to the second administration, with effect sizes increasing up to the fourth. That matters here because a diagram test has a working memory component, and the repeat gain is not confined to the reasoning part.

Why do scores rise? Scharfen and colleagues summarize three groups of causes drawn from earlier work: the latent ability itself might be enhanced by retesting, distorting factors such as motivation, test anxiety and unfamiliarity might be reduced, or test specific strategies and skills might develop. They judge the first of these rather improbable. For a candidate this means that a better second attempt is most plausibly a lower anxiety and better strategy effect, which is real for the score but is not a change in the underlying ability. The professional standard on this point is Standard 1.7 of the Standards for Educational and Psychological Testing, which says that if test performance is claimed to be essentially unaffected by practice and coaching, the propensity for performance to change with those forms of instruction should be documented.

8 What About Practice Questions, Answers and PDFs?

Searches for diagrammatic reasoning questions, answers and PDFs ask for items, and a trustworthy page can offer a description of the format, the publisher's own free practice and the research on repeating a test, but not copied items. There are three reasons. A copied item from a live test helps no one choose between tests and can compromise a publisher's bank. Saville says its candidates receive items from a randomized bank, so a fixed list of answers circulating online is unlikely to match what a given candidate sees, which is our inference from that statement. And Saville's own practice page warns that the difficulty of example questions does not necessarily represent the difficulty of the test itself, which means that a practice item cannot tell you how hard the real one will be.

What a candidate can do is use the publisher's own material. Saville's practice tests page says the practice tests are meant to familiarize you with the format of the actual test and with working to strict time limits, that you can complete the questions as many times as you like and that you receive onscreen feedback for each question. Its cognitive ability page says practice tests and worked examples are free. Those are publisher statements, and they are the best answer to the search for a free diagrammatic reasoning test: the free version that matches the real format is the one the publisher provides.

What the research says about practice should set expectations. Hausknecht and colleagues found larger effects when practice was accompanied by coaching, and Scharfen and colleagues found that most of the gain arrives by the first retest, with nothing further after the third. Repeating the same practice items therefore has diminishing returns, and the research gives no reason to expect that reasoning ability itself changes. Saville's cognitive ability page makes a related claim without data, stating that preparation reduces nerves rather than scores, while the meta-analyses measure repeated administration of the same test, which is a different question. The pragmatic reading is familiarity: knowing the screen, the timing and the kinds of change an operator can make reduces the cost of being new to the test. How much to prepare is the candidate's decision, and the wider guide to the category is on the page about pre-employment cognitive tests, which this page does not repeat.

One more rule matters for anyone tempted by "questions and answers" files. An employer asks candidates to complete the test unaided, and the publisher's terms for an invitation govern what a candidate may use. We have not read those terms for any product, and we do not recommend shortcuts. The point of this page is narrower: understanding what the test measures, why scores move on a retake and what a score can and cannot say is legitimate, useful and available without any item.

9 What Does ACIS Offer That Is Close, and What Does It Not Have?

ACIS has no flowchart or operator task, and the closest thing it offers is the Fluid Reasoning Index, five subtests in five formats, reported beside a separate working memory index. We say this plainly because a reader who searched for a diagrammatic reasoning test should not buy anything expecting that format. ACIS is a different instrument for a different purpose, and nothing on this page suggests that it prepares for, predicts or converts to any employer test. What it can show is a profile of the abilities that a diagrammatic test samples, on adult norms and with each domain reported separately.

The Fluid Reasoning Index (FRI) contains Matrix Reasoning, Figure Weights, Visual Number Series, Logic Grid and Complex Relations. Each subtest page, read October 29, 2026, describes the task family and not the items. The page for Matrix Reasoning describes rule discovery in unfamiliar visual patterns, where the person infers how rows or columns change and selects the option that completes the pattern. The page for Figure Weights describes balance scale problems that ask for relational balance reasoning and sits the task between Gf and quantitative reasoning. The page for Logic Grid describes controlled deduction under multiple constraints, in which several rules stay active until one arrangement remains. The Visual Number Series page describes sequential rule discovery, and Complex Relations describes relational integration, the coordination of several relations inside one problem. The page on the Fluid Reasoning Index covers the index itself.

The working memory side is the Working Memory Index, which contains Digit Span, Alphanumeric Sequencing and Visual Sequence. The page for Digit Span describes immediate auditory verbal storage and, in harder versions, active manipulation, and the pages for Alphanumeric Sequencing and Visual Sequence describe working memory under reorganization and temporary storage of ordered visual or spatial information. Digit Span and Alphanumeric Sequencing are in every form, and Visual Sequence is only in the Full Scale form. Digit Span and Alphanumeric Sequencing present their items as audio, so they need working sound.

Three facts from the instructions on those pages matter to a candidate who wants to compare. Several of the FRI subtests are timed per item or per subtest: Matrix Reasoning gives 30 seconds per item plus a shared 120 second extra time bank, Figure Weights, Visual Number Series and Complex Relations give 45 seconds per item and Logic Grid gives 20 minutes in total after two unscored practice puzzles. Matrix Reasoning, Figure Weights, Logic Grid and Visual Number Series stop after three consecutive errors. And all of the pages say that no external aids, such as calculators, notes, scratch paper or drawing tools, are allowed. A diagrammatic test sat for an employer may differ on every one of these points.

The nearest ACIS formats to what a diagrammatic test asks are the sequence like Visual Number Series, the pattern completion of Matrix Reasoning and the constraint tracking of Logic Grid, and none of them is a chart of operators. A high Matrix Reasoning score and a low Logic Grid score, or the reverse, would be informative about the profile and silent about any employer test. That silence is the correct reading, not a gap to be filled by guesswork.

10 How Should You Read a Diagrammatic Reasoning Score?

A diagrammatic score is a rank against a comparison group chosen by the publisher, so the first question to ask is who that group is, and the second is how precise one short test can be. The pages we opened give neither answer. SHL's fact sheet reports the time, the question count and the format and states no norm group and no validity coefficient, and Saville's pages describe the product range without a norm group for the diagrammatic tests. A candidate who is shown a percentile or a band should therefore read it as a position among a particular set of test takers, which may be applicants to one employer or a publisher's general sample, and not as a position among adults in general.

The professional standard is explicit about what a score report should contain. The Standards for Educational and Psychological Testing (American Educational Research Association, American Psychological Association and National Council on Measurement in Education, 2014) say in Standard 6.10 that when scores are released, interpretations appropriate to the audience should describe in simple language what the test covers, what scores represent, the precision or reliability of the scores and how scores are intended to be used. Standard 2.13 says that the standard error of measurement should be provided in units of each reported score. Those are the questions to put to any score you are given, and a candidate is entitled to ask the employer or the publisher for them.

Scale vocabulary helps. A percentile says what share of a comparison group scored at or below you. A standard score such as an ACIS index sits on a scale with a mean of 100 and a standard deviation of 15, and the page on IQ score versus percentile explains why the two cannot be averaged. How a comparison group becomes a norm is covered on the page about how IQ scores are normed, and the meaning of a standard deviation of 15 is on the page about standard deviation 15 explained. Precision belongs to the page on reliability and validity, which explains in general terms why a score from one short test carries more uncertainty than an index built from several subtests.

The retest findings translate into percentiles, and the translation is our arithmetic on published effect sizes, assuming a normal distribution and a starting point at the median. A gain of 0.33 standard deviations moves a person from the 50th percentile to about the 63rd, and a gain of 0.52, the first to fourth administration figure, moves them to about the 70th. Those are averages across many tests and people, and the prediction interval reported by Scharfen and colleagues runs from minus 0.30 to 0.96, so no individual can read their own retest change off the average. The practical rule is that two results from the same test, or from different tests by the same publisher, are not independent measurements of one fixed number.

11 What Should You Take If You Want Your Own Fluid Reasoning Profile?

If the aim is to see your own fluid reasoning and working memory standing on adult norms, take a normed battery that reports those domains separately, because an employer test is built for a hiring decision and the pages we read do not say what the candidate sees afterwards. There are two honest routes. The supervised route is a psychologist giving a full battery such as the WAIS-5, which Pearson describes as individually administered and lists for ages 16 to 90. The second is a normed online battery, and the two comparison pages linked in the table section cover the options, including which ones return a score against adult norms and which are only entertainment.

ACIS is one such online battery, and it has commercial interest in this recommendation, so the facts matter more than the framing. The Quick form is 15 dollars, covers 6 subtests in 3 domains and takes about 45 minutes. It includes Matrix Reasoning and Figure Weights from the FRI and Digit Span and Alphanumeric Sequencing from the WMI. The Optimized form is 30 dollars, covers 13 subtests in 5 domains and takes about 110 minutes, and adds Logic Grid. The Full Scale form is 50 dollars, covers all 20 subtests in six domains and takes about 175 minutes, and is the only form with all five FRI subtests and Visual Sequence. Prices were read on October 29, 2026 and can change. Every form reports a Full Scale IQ with its 95 percent confidence interval, and breaks are allowed, progress is saved between subtests and a buyer has 30 days to complete.

The Custom Test is a different route for a reader who wants only the two domains discussed on this page. It costs 5 dollars per subtest, and any domain you complete returns its full index score. By our arithmetic, the five FRI subtests would cost 25 dollars and the three WMI subtests 15 dollars, and the Full Scale form costs the same as ten subtests. The Full Scale IQ needs the Full Scale battery. Whichever form a reader chooses, the outcome is a profile of abilities, not a prediction about a test given by an employer.

Several entry pages help a reader decide before paying. The page on the IQ test for adults explains what an adult normed test returns, and the page on the cognitive ability test covers the general category. The ACIS free trial needs no card, and one is allowed per person. A reader who is unsure which route suits their purpose will find the comparisons on those pages more useful than any single recommendation.

12 Where ACIS Sits, and What a Buyer Should Do

ACIS is an online, unsupervised, English only cognitive assessment that is not a clinical instrument and not an employer test, so it can show an adult their own fluid reasoning and working memory profile but cannot stand in for, prepare for or predict a diagrammatic reasoning test. We sell ACIS, so this section is a disclosure as much as a recommendation. The limits are as plain as the features. The research page on ACIS states that the instrument "is not intended for clinical diagnosis, hiring decisions, accommodation determinations or admission to high IQ societies" (read October 29, 2026 on the ACIS for research page). It has no flowchart or operator task. The technical manual states that verbal and knowledge tasks are affected by English proficiency and educational exposure.

The Standards for Educational and Psychological Testing (American Educational Research Association, American Psychological Association and National Council on Measurement in Education, 2014) are the standard this page used to read scores. The American Psychological Association is one of the three co-publishers, and the 2014 edition is openly available on the Standards website. They ask that a test's intended interpretations be supported by evidence for each use, that score reports say what a test covers and how precise it is, and that claims about practice and coaching be documented. Applied to a candidate, that means asking the employer what the test is for, treating a single timed result as one sample of reasoning, and treating an online result from any source, including ACIS, as information about performance on those tasks on that day. The page on the technical manual holds the documentation of how the instrument was built and scored, which this page does not summarize.

Guarantee terms are specific, and a buyer should read them exactly: "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." It is not an unconditional guarantee of satisfaction.

What a reader should do next depends on the purpose. If you have been invited to take a diagrammatic reasoning test, read the invitation for device and timing rules, use the publisher's own free practice if one exists, and treat any retest gain as familiarity and not a change in ability. If you want to see your own fluid reasoning and working memory profile, start with the free trial, which needs no card, and look at the example report for the form you are considering before paying for it. If you need a score that an employer, a school or a society will accept, take the route that body specifies. The Korn Ferry page linked earlier covers one such employer battery.

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. Web pages were opened on October 29, 2026 and can change. Where we calculated a figure ourselves, such as a percentile from an effect size, the text says so and the arithmetic is ours, not the authors'.

• Saville Assessment. Aptitude tests. Publisher page, read October 29, 2026.
• Saville Assessment. Practice tests. Publisher page, read October 29, 2026.
• Saville Assessment. Cognitive ability assessment. Publisher page, read October 29, 2026.
• SHL. Verify Interactive, Inductive Reasoning: Assessment Fact Sheet. SHL and its affiliates, 2022, read October 29, 2026.
• SHL. Cognitive assessments. Publisher page, read October 29, 2026.
• Test Partnership. Abstract, inductive, diagrammatic, and non-verbal reasoning: what's the difference? Vendor academy page showing an update date of July 6, 2026, read October 29, 2026.
• Pearson. Wechsler Adult Intelligence Scale, Fifth Edition. Publisher product page, read October 29, 2026.
• Larkin J and Simon H. Why a Diagram is (Sometimes) Worth Ten Thousand Words. Cognitive Science, 1987, volume 11, issue 1, pages 65 to 100.
• Halford G, Wilson W and Phillips S. Processing capacity defined by relational complexity: Implications for comparative, developmental, and cognitive psychology. Behavioral and Brain Sciences, 1998, volume 21, issue 6, pages 803 to 831.
• Carpenter P, Just M and Shell P. What one intelligence test measures: A theoretical account of the processing in the Raven Progressive Matrices Test. Psychological Review, 1990, volume 97, issue 3, pages 404 to 431.
• Ackerman P, Beier M and Boyle M. Working Memory and Intelligence: The Same or Different Constructs? Psychological Bulletin, 2005, volume 131, issue 1, pages 30 to 60.
• Kane M, Hambrick D and Conway A. Working Memory Capacity and Fluid Intelligence Are Strongly Related Constructs: Comment on Ackerman, Beier, and Boyle (2005). Psychological Bulletin, 2005, volume 131, issue 1, pages 66 to 71.
• Hausknecht J, Halpert J, Di Paolo N and Moriarty Gerrard M. Retesting in selection: A meta-analysis of coaching and practice effects for tests of cognitive ability. Journal of Applied Psychology, 2007, volume 92, issue 2, pages 373 to 385.
• Scharfen J, Peters J and Holling H. Retest effects in cognitive ability tests: A meta-analysis. Intelligence, 2018, volume 67, pages 44 to 66.
• Scharfen J, Jansen K and Holling H. Retest effects in working memory capacity tests: A meta-analysis. Psychonomic Bulletin and Review, 2018, volume 25, issue 6, pages 2175 to 2199.
• American Educational Research Association, American Psychological Association and National Council on Measurement in Education. Standards for Educational and Psychological Testing. 2014, Standards 1.7, 2.13 and 6.10.

14 Frequently Asked Questions