What the PI Cognitive Assessment scores and how a target decides

0 The short answer

The PI Cognitive Assessment is The Predictive Index's 12 minute test of verbal, numerical and abstract reasoning, with 50 multiple choice questions, that an employer sends you and that returns one scaled score from 100 to 450, which is then compared with a target the employer set for the job, not with a pass mark. PI's own administrator guide says most people will not answer all 50 questions in the time, so speed is part of the score, and PI's blog puts the average raw score in its reference group at 19.8 of 50, which it ties to the scaled average of 250 (pages opened October 29, 2026). ACIS is a separate way to see a fuller, adult normed profile of your own abilities, with Quick at 15 dollars and about 45 minutes and Full Scale at 50 dollars and about 175 minutes (prices read October 29, 2026), but it gives no score on the PI test and no conversion to it. The decisive limit is that a PI score means something only against the reference group and the job target PI uses, so no outside number can tell you whether yours was good.

1 What Is the PI Cognitive Assessment, According to The Predictive Index?

The Predictive Index describes the PI Cognitive Assessment as a 12 minute, 50 question, multiple choice measure of general cognitive ability that an employer sends to a candidate, and it states plainly that the test is not an IQ test. PI's product page for the assessment, which we opened on October 29, 2026, calls it a 12-minute, 50-question timed test, lists three content areas (verbal reasoning, numerical reasoning and abstract reasoning), says it needs no job specific knowledge, formal education or test preparation, and says its scores reflect speed and accuracy. PI's support page introducing the assessment, opened the same day, adds that the 50 questions are drawn from a larger pool by a linear-on-the-fly engine that keeps content and difficulty equivalent across administrations, so that no one receives exactly the same test twice.

Three further facts come from PI's help article on the assessment, opened on October 29, 2026. It lists versions of 12, 18 and 24 minutes, and up to three attempts. It says the assessment supports 64 languages. And it says the original assessment was developed in 2010, following the Standards for Educational and Psychological Testing and the SIOP (2003) principles for personnel selection. The same article says the test is meant for candidates, or for existing employees applying for a new position with different demands, and that it should not be used as a development tool. The product page adds that it can be given in proctored and unproctored formats, so whether someone watches you depends on the employer's choice and not on the test.

The test can also be taken on an ordinary device. PI's 2018 administrator guide, a PDF that PI still hosts and that we opened on October 29, 2026, says the assessment can be taken on computers, laptops and larger tablets, shows five items on screen at a time, lets the taker move between pages, and displays the time remaining. The guide is eight years old and some details have changed since, which this page flags where we found a difference. It also says the assessment's instructions prohibit calculators while placing no explicit restriction on other aids, and that each company decides whether to allow things like scratch paper. If you have been sent the test, the practical step is to ask the employer, in writing, what aids are allowed.

What PI says it measures deserves a careful reading. The help article says the assessment measures general cognitive ability, often called "g," and then says that it does not measure IQ, previously acquired knowledge, skills, or willingness to learn. It adds that valid IQ tests are longer and measure intelligence, which it calls closely related to "g" but not the same. The product page puts the same idea differently: unlike traditional IQ tests, it says, the assessment measures how quickly someone processes new information. The page on what IQ measures explains why "g" and an IQ score are related but are not the same object, and why a 12 minute test cannot stand in for a battery. That is a limit on what the test can claim, and it applies in both directions: a PI score is not an IQ, and nothing here converts one into the other.

2 How Is the PI Cognitive Assessment Different From the PI Behavioral Assessment?

The two PI assessments answer different questions: the Behavioral Assessment is an untimed, roughly six minute survey of behavioral drives with no right answers, while the Cognitive Assessment is a timed test with right answers that yields a score against a job target. PI's page listing its assessments, opened on October 29, 2026, says the Behavioral Assessment takes six minutes or less, is untimed, is a free-choice, stimulus-response tool, and measures an employee's natural behavioral drives and needs. PI's blog on how its assessments work, also opened that day, names the four drives as Dominance, Extraversion, Patience and Formality, says takers receive two lists of adjectives, and says each person is assigned one of 17 Reference Profiles. The same blog describes the Cognitive Assessment as timed, with a unique set of questions for each candidate across numerical, abstract and verbal categories.

Both results are read against a target and not against a pass mark. PI's blog on scores says that neither assessment is a test you pass or fail and that candidates are compared with a job target and not with each other. PI's help article on setting a cognitive target says an employer can set the cognitive target after a behavioral target exists for the role, which may be why the two often arrive together. The product page recommends pairing them and claims that adding the Behavioral Assessment leads to a 40 percent improvement in job satisfaction; the page gives no study behind that figure, so we report it only as a claim.

For a candidate, the difference changes how to think about each: the survey asks how you tend to act, and the test asks what you can do within a clock. The research on how personality traits and measured ability relate to one another is summarized on the page about IQ and personality, and its main lesson applies here: typical behavior and maximum performance are different things, and a profile that combines them is not a single number.

3 How Is the PI Cognitive Assessment Scored, and What Is a Good Score?

PI reports one scaled score from 100 to 450 based on the number of questions answered correctly, says that the average is 250, and says that no score is good or bad on its own, because a score is judged against the target for the job. The pieces of that answer come from four PI pages opened on October 29, 2026, and the table below keeps each figure next to its source so that you can see which statements are current and which come from the 2018 guide.

Two cautions apply to that table. First, the two reference group sizes differ and PI does not explain the difference on the pages we opened. The blog reports more than 288,000 assessments; the 2018 guide reports over 900,000 working age adults. We do not know whether they describe the same reference set at different dates, so we cite each with its source and do not combine them. Second, a percentile on this test is a percentile among working adults who were assessed, as the guide says, and not among the general adult population; the guide's own example places a 340 at the 86th percentile of the global workforce. Our arithmetic on the blog's figures is simpler: an average of 19.8 correct out of 50 is about 40 percent, which fits a test designed so that most people run out of time.

What does PI say about "a good score"? PI's blog on its scores, opened on October 29, 2026, states that neither assessment is a test you pass or fail, that scores at or above the cognitive target earn a match score of 10, that scores below the target lose points on a sliding scale, and that there is no penalty for scoring above the target. PI's scoring page says a person who scores 270 is always a good match for a job with a target of 270, and that the norm for a particular job may differ from the population norm. In other words, a good score is whatever meets the target that the employer chose for that role, and the same 260 can be a full match for one job and a shortfall for another. The page on how a percentile differs from a score explains why even the percentile context cannot be averaged or compared across tests without knowing the reference group.

The subscores are a source of confusion. The help article says the Verbal, Numeric and Abstract Reasoning subscores appear in the candidate's cognitive report. PI's scoring page and the 2018 guide both say the subscores are not valid or reliable enough to support any hiring decision, and the guide gives the reason: with 12 minutes, differences between content areas can reflect skipping or guessing rather than ability. The guide also lists the maximum items per category as 14 verbal, 18 numerical and 18 abstract, while the current scoring page lists maximums of 16, 18 and 16. Both sum to 50, and the difference suggests the item mix was revised between 2018 and today.

4 Why Does a 12 Minute Clock Make Speed Part of the Score?

PI built the 12 minute limit on purpose so that speed counts: its own administrator guide says most people will not answer all 50 questions, that this is acceptable, and that the limit brings information processing speed into the measurement of general cognitive ability. The guide, opened on October 29, 2026, says the timing is designed to introduce a speed factor and that the limit requires the taker to move through the items as quickly as possible. PI's help article gives a second reason: without the timed element, it says, the questions would be easier to answer and the test would no longer measure how quickly someone learns. It adds that a shorter test gives a better candidate experience.

Our arithmetic shows how tight the clock is. Twelve minutes is 720 seconds, and 720 divided by 50 is 14.4 seconds per question, with nothing left over for rereading or checking. The help article describes the content as simple math, basic logic and language comprehension. A short, simple item set under a short clock suggests that the clock, more than the difficulty of any one question, creates the spread between people. That is our inference from two PI statements, not something PI says in those words.

The scoring rule shapes what a rushed candidate gains and loses. The 2018 guide says incorrect answers do not count against the respondent and that there is no other penalty or weighting, and the support page says the number of questions attempted is irrelevant to interpreting the score. The score is the count of correct answers. Our reading of that rule is that a question left blank and a question answered wrongly cost the same, so time spent stuck on one item is time taken from items you might have answered correctly. PI's pages do not give test-taking advice, so treat that sentence as our reading of the scoring rule and nothing more.

The measurement literature has a precise word for this design. The glossary of the Standards for Educational and Psychological Testing (AERA, APA and NCME, 2014), which we read in its open access edition, defines speededness as the extent to which test takers' scores depend on the rate at which work is performed as well as on the correctness of the responses. The Standards also note, in their chapter on reliability, that internal consistency estimates are likely to be inflated for highly speeded tests. That caution matters for any short, fast test: a reliability figure computed in the usual way would not, by itself, show that the score is precise. Kyllonen and Zu (2016), in the Journal of Intelligence, review response time in cognitive ability measurement, including the speed-level distinction and the speed-accuracy tradeoff that a design like this raises.

In the CHC model, which names the broad abilities that cognitive batteries sample, performing simple tasks quickly is processing speed (Gs), a different broad ability from fluid reasoning (Gf) or comprehension-knowledge (Gc). A timed reasoning test mixes them, so the same score can come from strong reasoning with moderate speed or from moderate reasoning with high speed. The page on why IQ tests are timed explains the difference between speed tests and power tests, and the page on the processing speed test shows how a battery treats speed as its own index. ACIS reports processing speed as the PSI, built from Symbol Search and Coding, separately from the reasoning and knowledge indices, which is a different design from one 12 minute score in which the abilities are blended.

5 What Do the Verbal, Numerical and Abstract Sections Sample?

PI names three content areas, and in the CHC model the most natural reading is that they sample language-based comprehension-knowledge (Gc), numerical ability (Gq and quantitative reasoning within Gf) and abstract pattern reasoning (Gf), although PI describes the test as one measure of a single trait and the pages we opened publish no CHC mapping. The product page, opened on October 29, 2026, defines the areas in everyday terms: verbal reasoning is understanding and analyzing written information, numerical reasoning is working with numbers and interpreting data, and abstract reasoning is identifying patterns and solving novel problems. The help article describes the content as simple math, basic logic and language comprehension. We do not describe or reproduce any PI question.

The CHC mapping is our interpretation, and we label it for a reason. The 2018 guide says the assessment is designed to measure a single trait, cognitive ability, and that the content-area subscores are not valid or reliable enough to support any hiring decision. A candidate should therefore not read a verbal, numerical or abstract subscore as a strength or a weakness, because PI says not to. The three labels are best understood as a way of sampling different content so that the single score does not depend on one skill. The page on the CHC model lists the broad abilities and their narrow abilities, and the page on cognitive domains in IQ testing shows how a battery groups them into indices.

Each content area has a counterpart in an adult normed battery that reports the ability on its own line. For the verbal side, the ACIS Vocabulary subtest measures depth and precision of word knowledge and is read as comprehension-knowledge (Gc); its page notes that vocabulary accumulates through reading and education and peaks later in life than speeded or novel reasoning tasks. For the abstract side, the ACIS Matrix Reasoning subtest asks a person to infer the rule in a pattern and select the option that completes it, and it is read as fluid reasoning (Gf). For the numerical side, the ACIS quantitative subtests, Arithmetic and Mathematical Achievement, are in the Full Scale form only.

A practical point follows for anyone who feels that one section went badly. Because the test returns a single scaled score, you will not know, and the employer is told not to infer, whether you did better on words or on numbers. The page on verbal and quantitative ability explains why the gap between two scores is a less reliable quantity than either score, and why a tilt toward one or the other should be read with caution even in a long battery. A 12 minute test cannot support that comparison, and PI's own guide says so.

6 How Does a Job Target Turn the Score Into a Verdict?

A PI score becomes a hiring signal only when it is compared with a cognitive target that the employer sets for the job, and the match that comes back runs from 0 to 10 and gives full credit to any score at or above the target. PI's help article on setting a cognitive target, opened on October 29, 2026, says the target gauges a person's ability to handle the essential demands of a role. The employer sets it after a behavioral target exists, chooses the minimum level of each of 20 cognitive skills the job needs, selects "none" for a skill the job does not require, and can adjust the result manually. The article says the target cannot be reset, only replaced by creating a new job, and that the cognitive data are "just one data point" that should not be weighted more or less than other hiring criteria.

The 2018 guide explains the logic. It says the assessment is designed to be criterion-referenced: respondents are judged by how their scores compare with a target score for a job, not by rank among applicants. It says a target should come from a methodical, defensible process, names three common routes (a standard-setting study with an assessment professional, a validity study, or the PI Job Assessment), and describes the most common one. The employer picks a job family, which sets a baseline, and then answers questions in the PI Job Assessment that move the target up or down; if no job family matches, the baseline is the average score of 250. When raters disagree widely the software shows a caution, and the guide says targets may need adjusting as job demands and performance data change.

PI's blog on scores states the match rule: at or above the target earns a 10, below it loses points on a sliding scale, and scoring above the target carries no penalty. The blog also says PI recommends against cutoff scores and uses match scores instead. The 2018 guide adds that people whose scores do not match the target should not be disqualified on that basis alone, especially if they fit well in behavior, experience or education, and PI's support page on the assessment says match scores are for ranking and prioritizing and are not designed to be used as a sole decision-making measure.

PI is also explicit about where the test should not be used. The help article on setting a target lists roles with no potential for advancement, positions of minimal complexity, and cases where a candidate's disability would compromise access to or completion of the assessment in the allotted time. The product page frames the use case differently, as finding fast learners and reducing hiring risk in cognitively demanding roles. Both statements come from the publisher, and the second is marketing; the first is the more useful guide to when a target is sensible.

For a candidate, this changes what to do with a result. A raw number has no meaning until you know the target, and you will rarely see the target. The 2018 guide says each company should decide whether respondents see their own scores and should tell them before the assessment, and PI's scoring guidance recommends keeping scores confidential and communicating fit or rank instead. So ask the employer one question before you sit: will you be told your score, or only the outcome? The page on pre-employment cognitive tests covers how employers use these tests in general, and this page stays with how PI does it.

7 How Well Does a Score Like This Predict Job Performance?

Independent research finds a smaller link between general mental ability test scores and overall job performance than the older literature reported, and none of it tests PI's own validity claims, which the publisher states on the pages we opened without showing the studies. PI's product page, opened on October 29, 2026, calls the test the leading scientifically validated cognitive assessment and a top predictor of job performance, and cites "383+ validity studies" and "over 70 years of science." Its FAQ says the test is validated to be fair across age, gender, ethnicity and language. We report those as the publisher's claims. The pages we opened give no coefficients and no study list, and PI's own scoring page says its scaled score is a strong predictor of workplace performance but "not the only predictor."

The independent evidence is about the category of test, not the product. Sackett, Zhang, Berry and Lievens (2022), in the Journal of Applied Psychology, revisited earlier meta-analyses of selection procedures. Their abstract concludes that corrections for range restriction had often overcorrected, so that the validity of many selection procedures for predicting job performance had been substantially overestimated, with mean validity estimates reduced by .10 to .20, and that structured interviews emerged as the top-ranked procedure. A follow-up by the same authors, Sackett, Zhang, Berry and Lievens (2023) in Industrial and Organizational Psychology, reproduces the comparison in a table: for general mental ability tests, the Schmidt and Hunter (1998) estimate was .51 and the 2022 estimate is .31. The 1998 estimate comes from Schmidt and Hunter's review of 85 years of selection research in Psychological Bulletin; we cite its .51 only as the 2023 table reports it.

The same 2023 paper reports a more recent check. It notes that Griebe and colleagues (2022) meta-analyzed 113 validity studies of general cognitive ability and overall job performance from the 21st century and found a mean observed validity of .16 and a mean corrected validity of .23, and that they hypothesize the lower value reflects fewer manufacturing jobs and more team-based work with broader definitions of performance. The 2023 authors also write that their findings should not be read as a broad refutation of the value of cognitive ability measures in all settings, and that a predictor composite has about the same validity whether or not cognitive ability is included. A correlation of .31 squared is about .10, so at that validity a score accounts for roughly a tenth of the variation in rated performance, and .23 squared is about .05; both are our arithmetic, and both use corrected correlations.

Group differences are the other half of this literature, and they deserve a careful sentence. The 2023 paper reports that, among the predictors it compares, cognitive ability tests show the largest mean difference between Black and White applicant groups. That is a statement about group averages in applicant data and says nothing about any individual, and it is a reason employers monitor adverse impact when they use such tests. PI's claim of fairness across groups is a claim we cannot evaluate from the pages we opened. The practical reading for a candidate is modest: a test like this adds some information, an employer who uses it well treats it as one data point, as PI's own help article says, and a low result on one 12 minute sitting is not a verdict on your ability.

The Standards (AERA, APA and NCME, 2014) speak directly to this. Standard 11.1 says that before an employment test is developed and implemented, a clear statement of the intended interpretations of test scores for specified uses should be made, and that validation should be designed to determine how well this has been achieved for all relevant subgroups. Standard 11.4 says that when multiple test scores or test scores and nontest information are integrated for a decision, the role played by each should be clearly explained and the inference from each source supported by validity evidence. The page on reliability and validity explains the difference between a score that is consistent and a score that predicts something, and the page on IQ and job performance gives the wider research on ability and work outcomes.

8 Can You Practice for It, and What Do Retakes Do?

PI offers its own official sample questions, which give no score, and the research on retesting shows that scores on cognitive tests tend to rise when people test again, although PI says it gives each candidate a different set of questions and lets the employer decide whether you may retake. PI's support page on the sample questions, opened on October 29, 2026, says the purpose is to give candidates an idea of what the assessment experience is like and to let them practice under timed conditions. It says candidates do not receive a score at the end, that the sample tests have not been equated, that the questions should not be used to make talent-related decisions, and that candidates can see which questions they answered correctly and incorrectly along with the correct responses. We do not reproduce or describe any of those questions here, and we do not link to sites that sell or post practice material. The 2018 guide tells employers that their policies should prohibit candidates from accessing content that provides answers or guidance specifically related to the assessment, such as cheat sheets or content from an exam preparation company, so material of that kind can work against a candidate who uses it.

The format is the thing worth practicing, and PI's own sample is built for it. Forum threads in which people report their scores or the questions they remember are a different matter. They are anecdotes from self-selected posters, they cannot be verified, and they cannot tell you what a typical result is, which is why this page uses PI's published reference figures and cites no thread.

What happens on a retake? On the employer side, the 2018 guide says it is up to the company whether to allow retakes, that respondents see a different configuration of items each time, that nobody should take the assessment more than three times, and that the highest score should be used if someone does. It also says that giving someone a second attempt only because they did worse than expected is not fair unless every respondent gets the same opportunity, while a retest is reasonable after a technical problem such as an internet disconnection or a serious distraction. The help article says candidates get up to three attempts and that an assessment that expires or is abandoned uses one of them. If something goes wrong during the sitting, tell the employer at once and in writing.

Hausknecht, Halpert, Di Paolo and Moriarty Gerrard (2007), in the Journal of Applied Psychology, meta-analyzed 50 studies of practice effects on cognitive ability tests, covering 107 samples and 134,436 participants, and reported an adjusted overall effect size of .26. Effects were larger when practice came with coaching and when identical forms were used. PI says no participant gets exactly the same test twice, so the part of a retake gain that comes from seeing identical items should be smaller on this test; that is our inference, and format familiarity would remain. The page on retaking an IQ test covers the wider evidence on how much a second sitting changes a score and how long to wait.

9 What If You Need Extra Time, or the Score Comes Back Low?

PI offers 18 and 24 minute versions for candidates who are eligible for accommodations, and its own guidance cautions employers against over-reading a low score, which is a reason to treat one 12 minute result as one data point and not as a ruling on you. The product page, opened on October 29, 2026, lists accommodation variants of 18 and 24 minutes under the ADA. PI's support page on the assessment says extended-time versions are reported on the same score scale as the standard format through equating, that eligibility follows company policy and the law, and that for disability-based requests it is typically a medical professional's decision what accommodation is needed, which neither PI nor the client should suggest. The 2018 guide says the same thing in more detail, and adds that extending the time slightly changes how the instrument measures cognitive ability, because speeded response is one part of the ability, which is why extended time should be given only when it is required. If you think you need an accommodation, raise it with the employer before you start; this page cannot tell you what the law requires where you live.

On low scores, the support page is more careful than most commentary about this test. It says the scaled score is the only figure to consider, describes lower-scoring individuals as people who will likely learn at an even and consistent pace and who suit tactical, short-term decisions, and cautions against over-interpreting low scores. The 2018 guide says that people whose scores do not match a target should not be disqualified on the score alone, especially if they are a strong fit in behavior, experience or education. The product page notes that the test needs no job-specific knowledge, formal education or test preparation, so the score is not a record of what you have studied.

What a low score can reflect is also broader than ability. A clock of 14.4 seconds per question, an interruption, a technical problem, or an unfamiliar timed format can all change the number, and the 2018 guide treats a disconnection or a serious distraction as grounds for a retest and says retesting may help assess people who are less familiar with timed, computerized assessments. A fast, timed format makes the conditions of the sitting part of the measurement. If you want to understand your own strengths on a slower, fuller footing, the next sections explain the options.

10 How Does It Compare With the Other Ways to Measure Cognitive Ability?

The PI Cognitive Assessment is a 12 minute, employer-sent screening score judged against a job target, which is a different instrument from a psychologist's full battery and from a normed online battery you take for your own understanding, and the table below sets the options side by side. Facts about PI come from the pages opened on October 29, 2026, and facts about ACIS from its published site on the same day.

ACIS also has an Optimized form (13 subtests in five domains, about 110 minutes, 30 dollars) and a Custom Test at 5 dollars per subtest, where any domain you complete returns its full index score; prices were read on October 29, 2026 and can change. The Quick form has no processing speed or quantitative subtests, so it covers three domains, and the Full Scale form is the one that covers all six domains and reports the Full Scale IQ from all 20 subtests.

Other employer tests are separate instruments with their own publishers. The pages on the CCAT, the Korn Ferry test, verbal reasoning tests and the IQ test for a job interview cover them on their own terms. The Wonderlic has its own page on converting a Wonderlic score to IQ, which shows where such conversions come from and where they break; PI publishes no conversion for its own scale, and we have not made one.

If you want to see your own profile on a normed, adult basis, the page on IQ tests for adults explains what to look for, the guide to the best online IQ tests compares the market, and the page on where to take an IQ test lists every route from a psychologist to an online battery. A short form is a legitimate choice if you accept its limits, and the page on the quick IQ test works through what a shorter battery costs in precision. The page on cognitive ability tests describes the free, online, adult normed route.

11 How Do You Read a Score From a Test Like This?

A score is a position in a reference group, so the first question about any number is who it was compared with, and the second is how much error surrounds it; a PI score answers the first with working adults assessed by PI and publishes no band for the second on the pages we opened. PI's blog says the 250 average comes from a reference group of more than 288,000 assessments, and the 2018 guide describes a sample of over 900,000 working age adults. PI describes both as working-adult or workforce reference groups and not as samples of all adults. A working population leaves out many people, so a percentile among workers is not a percentile among all adults. The page on how IQ scores are normed explains how reference groups are built and why the same performance gives different standings against different groups.

The scale is the second thing to read. On an IQ-style scale the mean is 100 and the standard deviation is 15, which the page on the standard deviation of 15 explains, and a percentile follows from distance in standard deviations. PI's scale runs from 100 to 450 with a mean of 250, and the pages we opened do not state its standard deviation. The guide gives one example: a 340 at the 86th percentile. If the curve is normal, as the guide's figure shows it, an 86th percentile is about 1.08 standard deviations above the mean, which would make a standard deviation about 83 points; that is our arithmetic from one published example and not a PI statistic. It is also not a conversion to IQ. A working-adult reference group, a 12 minute speeded test and a battery normed on the general adult population are different yardsticks, which is why scales should not be mixed, and why the calculators on how rare a given IQ is apply to IQ-scale scores and not to a PI scaled score.

The third thing to read is precision. Every score carries measurement error, and a test reports it as a standard error or a confidence interval; the PI pages we opened publish neither for the scaled score, and report scores in 10-point steps. A short test, a fast test and a single sitting each make a score less precise than a long, unhurried battery. An adult normed battery handles this by reporting the interval, so a reader can see a range and not a single point. On a battery with six indices, differences between indices are also common in the general population, which the page on index score differences puts in numbers, so a profile that is uneven is not a sign of a defect.

12 Where ACIS Sits, and What a Buyer Should Do

ACIS is an online, unsupervised, adult normed battery for understanding your own abilities, and it is not a stand-in for the PI test, for a clinical evaluation or for any employer decision. It has 20 subtests in six CHC domains and reports a Full Scale IQ with its 95 percent confidence interval and six primary indices (verbal comprehension, fluid reasoning, quantitative reasoning, visual spatial, working memory and processing speed), on the standard scale with a mean of 100 and a standard deviation of 15. Adult norms cover ages 16 to 90. Quick costs 15 dollars and takes about 45 minutes, Optimized costs 30 dollars and takes about 110 minutes, and Full Scale costs 50 dollars and takes about 175 minutes, all one time with no subscription and prices read on October 29, 2026. Breaks are allowed, progress is saved between subtests, and you have 30 days to finish. The free trial needs no card, and the site refuses a second trial from the same email address or browser.

Its limits are plain. It is online and unsupervised, it is not a clinical or diagnostic instrument, and it is English only; the technical manual says verbal and knowledge tasks are affected by English proficiency and educational exposure. Its research page says ACIS "is not intended for clinical diagnosis, hiring decisions, accommodation determinations or admission to high IQ societies." A result is a profile for your own understanding and nothing else. Three of the subtests, Digit Span, Alphanumeric Sequencing and Arithmetic, present items as audio and need working sound. The technical manual documents the battery for readers who want the detail, and this page quotes none of its statistics.

The Standards for Educational and Psychological Testing (AERA, APA and NCME, 2014), published jointly by the three bodies, give the rule that governs how to hold all of this. Standard 1.4 says that if a test score is interpreted for a given use in a way that has not been validated, it is incumbent on the user to justify the new interpretation for that use, providing a rationale and collecting new evidence if necessary. Standard 11.1, described above, says the same for employment tests: state the intended interpretation first. For you, the consequence is simple. A PI score is built to be read against a PI target by the employer who set it, and no number from a different instrument has been validated for that reading, so none should be offered as one.

What should a buyer do? If your question is what the PI test is and how to take it calmly, the answers above, the employer's instructions and PI's own sample are the sources. If your question is what your abilities look like across reasoning, knowledge, memory, spatial skill and speed on an adult normed basis, start with the free trial, open the Full Scale example report to see what a report contains, and then choose a form; the Full Scale form is the one that covers all six domains.

13 Sources Behind This Page

Every fact about the PI Cognitive Assessment on this page comes from The Predictive Index's own pages, which we opened on October 29, 2026 and report as the publisher's statements, and every research figure comes from the papers below, with any figure we calculated labeled in the text as our arithmetic. Where we relied on a table printed in a later paper, or on an abstract, the entry says so. PI's pages change, and its 2018 administrator guide predates them, so check the current pages before you rely on a detail.

1. The Predictive Index (2026). Cognitive Assessment Test For Employment, product page. Questions, time, content areas, accommodation versions, validity and fairness claims. Read on October 29, 2026. https://www.predictiveindex.com/assessments/cognitive-assessment/
2. The Predictive Index (2026). What is the PI Cognitive Assessment?, help article. Versions, attempts, languages, development, what the test does not measure, subscores. Read on October 29, 2026. https://docs.predictiveindex.com/en/articles/10579885-what-is-the-pi-cognitive-assessment
3. The Predictive Index (2026). Introduction to the PI Cognitive Assessment, support page. Question pool, scale, match scores, extended time, interpretation guidance. Read on October 29, 2026. https://www.predictiveindex.com/learn/support/introduction-to-the-pi-cognitive-assessment/
4. The Predictive Index (2026). Cognitive Assessment scoring, support page. Average of 250, scaled score use, subscores. Read on October 29, 2026. https://www.predictiveindex.com/learn/support/cognitive-assessment-scoring/
5. The Predictive Index (2026). Understanding the Predictive Index score: Is yours "good"?, blog. Scale, average raw score, reference group, match score rule. Read on October 29, 2026. https://www.predictiveindex.com/blog/predictive-index-scores/
6. The Predictive Index (2026). Setting a Cognitive Target, help article. How a target is set and where not to use the assessment. Read on October 29, 2026. https://docs.predictiveindex.com/en/articles/10579871-setting-a-cognitive-target
7. The Predictive Index (2026). Our assessments, support page. Length, timing and purpose of the Behavioral and Cognitive Assessments. Read on October 29, 2026. https://www.predictiveindex.com/learn/support/our-assessments/
8. The Predictive Index (2026). How does The Predictive Index test work?, blog. Behavioral drives, adjective lists, 17 Reference Profiles. Read on October 29, 2026. https://www.predictiveindex.com/blog/how-does-the-predictive-index-test-work/
9. The Predictive Index (2026). The PI Cognitive Assessment Sample Questions, support page. Purpose and scoring of the official sample, read for format only; no question was reproduced. Read on October 29, 2026. https://www.predictiveindex.com/learn/support/the-pi-cognitive-assessment-sample-questions/
10. The Predictive Index (2018). Administrator's Guide to the PI Cognitive Assessment, PDF, 29 pages, copyright 2018, hosted by PI in its legacy archive. Speed factor, scoring, reference group, targets, retests, subscores, accommodations. Read on October 29, 2026. https://media.predictiveindex.com/legacy/explore/wp-content/uploads/2019/03/Cognitive-Assessment-Admin-Guide.pdf
11. Sackett, P. R., Zhang, C., Berry, C. M., and Lievens, F. (2022). Revisiting meta-analytic estimates of validity in personnel selection: Addressing systematic overcorrection for restriction of range. Journal of Applied Psychology, 107(11), 2040 to 2068. Abstract read through PubMed; full text not opened. https://doi.org/10.1037/apl0000994
12. Sackett, P. R., Zhang, C., Berry, C. M., and Lievens, F. (2023). Revisiting the design of selection systems in light of new findings regarding the validity of widely used predictors. Industrial and Organizational Psychology, 16(3), 283 to 300. Full text read as a PDF; Table 1 and the discussion of Griebe and colleagues (2022). https://doi.org/10.1017/iop.2023.24
13. Schmidt, F. L., and Hunter, J. E. (1998). The validity and utility of selection methods in personnel psychology: Practical and theoretical implications of 85 years of research findings. Psychological Bulletin, 124(2), 262 to 274. DOI verified through Crossref; the .51 figure is cited only as printed in item 12. https://doi.org/10.1037/0033-2909.124.2.262
14. Hausknecht, J. P., Halpert, J. A., Di Paolo, N. T., and Moriarty Gerrard, M. O. (2007). Retesting in selection: A meta-analysis of coaching and practice effects for tests of cognitive ability. Journal of Applied Psychology, 92(2), 373 to 385. Abstract read through PubMed; full text not opened. https://doi.org/10.1037/0021-9010.92.2.373
15. Kyllonen, P., and Zu, J. (2016). Use of response time for measuring cognitive ability. Journal of Intelligence, 4(4), 14. Abstract read through Crossref; full text not opened. https://doi.org/10.3390/jintelligence4040014
16. American Educational Research Association, American Psychological Association, and National Council on Measurement in Education (2014). Standards for Educational and Psychological Testing. Washington, DC: AERA. Open access edition; Standards 1.0, 1.4, 11.1 and 11.4, the glossary entry for speededness, and the chapter on reliability. https://www.testingstandards.net/open-access-files.html

14 Frequently Asked Questions