Should I pay for a cognitive test that I can only take on my phone?

Should I pay for a cognitive test that only works on my phone?

Phone-only delivery does not, by itself, make a cognitive test worth buying or worthless. The answer depends on what you expect to receive. If you want a set of reasoning tasks, feedback, or an explanation that helps you practice, judge whether those parts are useful at the offered price. If you want the result to place you against a population, describe your general ability, or guide a consequential decision, ask what evidence supports that score interpretation for people like you and for that use, under the actual delivery and scoring conditions offered. A polished phone experience cannot supply that evidence on its own.

Device research can inform the question, but its findings depend on the tasks, devices, participants, and testing conditions studied. Comparisons involving a particular tablet battery or timed response task do not settle how every phone assessment performs. Treat the activity and the claim made from its score as separate purchase questions: the first may be enjoyable or useful even when the second is not established. That distinction matters even when a single fee covers both parts. Before paying, decide which one matters to you; the sections below examine what each can reasonably offer.

What exactly are you buying when the test is phone-only?

A phone assessment can bundle two different things: an experience and an inference. The experience is what happens while you take it: the questions, the way they appear and respond to touch, the chance to try unfamiliar problems, and any feedback or explanation provided afterward. The inference is what the result is said to mean about you: for example, that your performance is high relative to a defined group, or that it indicates a broader ability. You can value the first without accepting the second. Keeping them separate makes “worth paying for” a more answerable question.

Raw task performance is the record of what happened on the tasks under the conditions of that session: which responses were correct, how quickly someone responded if speed was recorded, or how performance differed across task types. A standard score is a transformed result interpreted against a reference group under a defined scoring approach. The group matters because a comparison has to specify whose performance supplies the reference and whether the test was given in conditions that make the comparison meaningful. Completing a task on a phone does not automatically create such a comparison; the interface can collect answers while the score’s broader meaning remains unsupported or unclear. The joint AERA/APA/NCME *Standards for Educational and Psychological Testing* frame validity around evidence and theory for a proposed interpretation and use. In practical terms, ask what claim the score is meant to support, rather than treating a number as self-explanatory.

Consider two hypothetical buyers. One wants a short set of visual and numerical puzzles to work through during a commute, followed by explanations that show how to approach the problems. That buyer may judge the service by whether the tasks are usable on their phone, the explanations are clear, and the price feels fair; a population ranking may not be needed. A second buyer hopes to use a score to decide who should be selected for a demanding program. For that purpose, engaging questions and a smooth interface are not enough: the buyer needs evidence that the score supports that particular comparison and decision. These are illustrations of different purchase goals, not reports about actual customers or any named test.

The same screen can therefore serve a worthwhile practice activity while providing weak grounds for a broad ability claim. It may present original, challenging problems and useful worked explanations without establishing how a score compares with adults generally. Conversely, a phone is not inherently a barrier to a defensible assessment: if the specific tasks, scoring, administration conditions, comparison group, and intended use have adequate support, mobile delivery can be one practical route to access. The device is part of what should be examined, but it cannot answer the whole purchase question by itself.

Before buying, describe the result you actually need in plain language. “I want to try reasoning problems and see explanations” is an activity goal. “I need a dependable estimate of my standing among adults” is a score-interpretation goal. “I need evidence for a placement decision” adds a consequential use. Those goals call for different grounds for satisfaction: task quality and feedback may be enough for the first, while comparison and decision claims require evidence matched to the intended interpretation. A seller’s confidence, attractive layout, or quantity of questions does not substitute for that match. A careful buyer can ask for sample feedback or a description of the tasks to judge the experience, then seek technical documentation for any population or decision claim. This distinction lets you pay for an experience you genuinely want while withholding trust from a score claim until its basis is clear.

Sources: Standards for Educational and Psychological Testing

What did large online device comparisons actually measure?

The study “Cognitive test scores vary with choice of personal digital device” compared performance among visitors to TestMyBrain.org who chose their own device; it did not randomly assign the same people to phones and computers. Its first sample included 59,587 visitors and its second included 3,818. These unusually large samples make the reported associations worth examining across measures. They do not convert a comparison of device-choosing visitors into a controlled estimate of what changing a device would do to a given person's score.

The relevant contrast was among the devices participants used to complete online cognitive measures, including computers and mobile devices, with smartphone users also distinguished by operating system in the reported analyses. The paper describes measure-specific patterns, including slower reaction-time performance among mobile users and Android smartphone users. The finding is therefore not a single result called “phone performance”: reaction-time measures and other cognitive tasks need to be read separately, and the pattern reported for one outcome should not be extended automatically to another. Nor does a contrast between selected device groups establish that every mobile device, screen, input method, or current test behaves the same way.

Because participants selected how they took part, device choice could travel with other circumstances. Age, access to a computer, prior familiarity with touch or mouse input, the place where someone completes a task, and the decision to visit and finish an online test are plausible sources of selection or context differences. The study's observational setup alone does not show that each factor produced its findings, or quantify how much any one factor contributed. They matter as alternative explanations to consider when interpreting an association, rather than as extra results to attribute to the authors.

The sample size addresses one concern—whether a pattern can be observed across many online visitors—but not the design question of why the groups differ. A very large sample can estimate a device-group association more stably for the people and measures included while leaving device choice entangled with the circumstances that accompany it. Scale also does not establish that the visitors represent all adults who might take a cognitive test, or that a result on the study's tasks transfers to another provider's questions, scoring, and delivery setup. Those are separate steps of inference, not benefits supplied by adding more participants to the same observational comparison.

Device categories also matter to the interpretation of the online comparison. “Mobile” can group devices with different screen sizes, operating systems, and ways of registering a response, while “computer” can include different hardware and input setups. The ledger supports naming the reported mobile and Android smartphone associations, but not treating those categories as if they were a single standardized apparatus. A category-level pattern can motivate a more controlled follow-up that holds the task and participant conditions steady while varying the mode; it cannot supply the outcome of that follow-up in advance. This is why the result has value as a map of where differences appeared, not as a correction factor a provider can simply apply to a score. A conversion would need direct evidence for its own test, devices, participants, and scoring.

The distinction between description and cause is especially important when a test is marketed across settings. A pattern among people who independently arrive at a research website may reflect who selects each device as well as how a device presents the tasks. The study does not let a buyer subtract those influences from an individual result, infer what the same visitor would have scored on another device, or assume that a provider's separate mobile test inherits the comparison. Those questions need evidence designed around the specific comparison being claimed.

This is useful evidence for asking which measures and user groups might warrant closer comparison. For example, a reaction-time association points toward examining the timing-sensitive task and its administration, while a claim about untimed reasoning would need evidence about those tasks. But the visitor comparison cannot by itself isolate a pure device effect, certify a phone version, or show that moving an assessment to a phone necessarily damages it. Its proper contribution is descriptive: among these self-selected visitors, scores on particular measures varied with chosen device in reported ways.

For a buyer, that scope rules out both sweeping sales claims and blanket rejection. Saying simply “mobile works” treats unlike devices, tasks, and participants as one case; saying “mobile fails” makes the same collapse in the opposite direction. The study supports neither universal verdict. It gives a reason to ask whether the specific assessment has been examined on the device and task conditions being offered, while keeping the reported visitor associations in view as a signal for further study rather than a universal adjustment or verdict.

Sources: Cognitive test scores vary with choice of personal digital device

What does a direct PC–tablet comparison add?

A direct comparison can narrow the question because it places two delivery modes beside the same cognitive battery, rather than comparing visitors who chose different devices. “Comparison of PC and iPad administrations of the Cogstate Brief Battery in the Mayo Clinic Study of Aging” examined cognitively unimpaired adults aged 50 and older in the Mayo Clinic Study of Aging setting. It used three designs: a same-day matched PC/iPad retest cohort of 168 people; test-naive baseline cohorts of 1,820 using PC and 605 using iPad; and matched longitudinal cohorts of 63 in each device group. These cohorts answer related but nonidentical questions, so their results should not be pooled into one generic estimate of a device effect.

In the same-day matched comparison, participants completed the battery on both PC and iPad, allowing performance under the two modes to be compared within the same people over a short interval. The baseline comparison instead contrasted separate PC and iPad groups who were new to testing, while the longitudinal comparison followed matched groups over time. The latter design asks whether observed change differed between modes across repeated assessments; it does not erase the fact that PC and iPad are distinct implementations. Together, these designs give more direct mode evidence for this battery than a sample formed solely by participants' device choices, while remaining specific to the clinic population and procedures studied.

The baseline cohorts contribute a different comparison from the matched retest: they show how first-time participants performed in the two administration groups without a same-day second exposure to the battery. The longitudinal groups address change over repeated measurement, and their matched size is much smaller than the baseline samples. Keeping the designs distinct prevents a large baseline count from being mistaken for the sample size supporting every part of the paper, or for proof that the mode comparison captures all conditions of ordinary phone use.

The pattern depended on the outcome. PC performance was faster on several measures, but accuracy usually did not differ or differed only slightly. The Groton Maze Learning Test was a notable exception: the report describes faster performance and more errors on iPad. That combination is informative because speed and accuracy are not interchangeable descriptions of performance. A person may complete a task faster while making more errors, so a timing difference alone does not tell the whole story about how well the task was performed. Nor can one outcome's direction stand in for every other task in the battery.

This divergence changes what a reader should infer from a headline such as “PC was faster.” Faster completion is an observed timing result, not automatically evidence of a broad advantage in cognitive ability or a change in the meaning of a combined score. To understand a battery, one must know how its component outcomes enter the reported result and what interpretation that composite is intended to support. The study reports task-level comparisons; it does not establish that a small mean timing difference necessarily changes a composite or changes any particular participant's classification. That consequence would require evidence tied to the score construction and the decision rule, rather than extrapolation from one component's mean.

The Groton Maze result further cautions against treating speed as a sufficient proxy for accuracy: in that task, iPad responses were faster alongside more errors. A buyer or evaluator who sees only completion time could miss the corresponding error pattern; someone looking only at accuracy could miss a timing difference relevant to a timed task. The study does not give a universal weighting rule for deciding which outcome matters more. That depends on the task's purpose and scoring, and the comparison's value is precisely that it shows why the dimensions should be reported and interpreted separately.

The authors characterized the mode differences as small, which tempers an alarmist reading. Most variables also did not show different rates of longitudinal change between device cohorts. These observations argue against portraying the battery as uniformly transformed by the tablet mode in this sample. At the same time, “small” describes the reported differences; it does not, by itself, decide whether a difference matters for every use, task, or score. The authors noted that consequences for clinical interpretation needed further study, leaving the bridge from measured mode differences to decisions unresolved.

The practical reach is deliberately narrow. This is direct evidence about PC and iPad administrations of the Cogstate Brief Battery among cognitively unimpaired older adults studied in a clinic research setting. It helps distinguish device-group association from a matched mode comparison and demonstrates that timing and accuracy can diverge by task. It is not evidence that a smartphone version is equivalent, because a phone was not the comparison mode; nor does it establish that all computer or tablet assessments share the same pattern. The useful lesson is to preserve the outcome-by-outcome result and the studied setup when judging what this comparison says.

Sources: Comparison of PC and iPad administrations of the Cogstate Brief Battery in the Mayo Clinic Study of Aging

When does response timing make the device part of the task?

A phone's input and display path matters most when the score depends on how quickly a response occurs. In “Characterizing Information Processing With a Mobile Device: Measurement of Simple and Choice Reaction Time,” a 60-person young-adult comparison examined simple and choice reaction time on mobile and traditional platforms. The abstract reports that mobile timing was usable and reliable for the tasks studied, while choice reaction time was faster on the traditional platform. This is a narrow result about timed responses in a small sample; it does not estimate a general phone penalty, show that all mobile tasks are unreliable, or establish how an IQ-style score changes.

The reason timing can be device-sensitive follows from what a latency measure contains. A participant sees a signal, detects it, decides what response is required, makes an input, and the system records that input. In a phone task, the signal must be rendered on the display and the touch must be detected and timestamped. Display refresh, event handling, and response-recording choices are possible parts of the observed interval as well as the participant's perceptual and decision processes. This is measurement reasoning about where elapsed time can enter, not a quantified breakdown reported by the 60-person study. Without implementation-specific evidence, one cannot say how much any one component contributes or subtract a generic device amount from an observed time.

That distinction is easier to see by comparing two task designs. In a simple reaction-time task, the outcome may be the interval between a visual cue and a button or touch response; the delivery and input chain is directly inside the measured quantity. In an untimed matrix choice, the central recorded outcome may instead be which option a person selects after inspecting a pattern. A small difference in when the display refreshes could matter to a latency estimate without changing which pattern option is chosen. But a matrix task can still be affected by screen size, visibility, or interaction design; “untimed” does not mean every aspect of delivery is irrelevant. The point is that relevance depends on what the task records, not on a universal property of phones.

The study's faster traditional-platform choice responses therefore justify a specific question for a timed assessment: has this version's input-to-recording path been evaluated under the conditions in which people will use it? Its report of usable, reliable mobile timing also matters. It indicates that platform involvement does not make the measurement worthless by definition; reliability within a tested setup and a difference between setups can coexist. Neither observation supplies a correction formula for another device, task, or scoring system. A buyer should look for evidence about the actual implementation and the timing outcome it claims to measure, rather than treating either the study's platform difference or its reliability statement as a verdict on every phone test.

This also explains why a timing result should be read with its task definition. A study may record stimulus onset and response timestamps according to a particular software and hardware setup; another assessment may use different screens, touch handling, response windows, or rules for discarding unusually slow responses. Those details determine what the recorded interval represents. The 60-person comparison is not evidence that each component differed, nor that its participants' platform gap was caused by one specific display or input feature. It makes the pathway worth inspecting where speed is the construct. If a provider reports only a total score, a buyer may not be able to tell whether speed contributes to that total; disclosure of task demands and scoring helps distinguish a speed-sensitive result from an untimed accuracy result.

Sources: Characterizing Performance on a Suite of English-Language NeuroUX Mobile Cognitive Tests in a US Adult Sample

What changes when the same mobile battery is repeated?

A repeated phone test is a changing measurement occasion: performance can reflect both a person's standing on the task and what differs between attempts. “Characterizing Performance on a Suite of English-Language NeuroUX Mobile Cognitive Tests in a US Adult Sample” followed 393 English-speaking US adults aged 20–79 using eight smartphone cognitive tests. Participants completed the tests daily for ten days, with five administrations of each test. This schedule makes the study useful for asking how scores behave across repeated use of one mobile battery; it does not turn the battery into a representative reference sample or establish properties of other products.

The authors reported intraclass correlation coefficients (ICCs) above .76 for nine of the twelve scores they evaluated. An ICC is a summary of how consistently people retain their relative score positions across repeated administrations under a specified protocol. It is not a statement that every person's score is identical each time, that the measure has no error, or that a result can be compared with a population norm. Here, the count of nine qualifying scores is evidence of reasonably consistent rank ordering for many scores in this study's repeated setup. It should not be recast as a single reliability value for all eight tests: scores are task-specific, and the remaining scores did not meet that reported threshold.

Consistency also does not mean that repeated exposure leaves performance unchanged. The study found practice gains on four scores. A practice effect is a change in performance associated with prior exposure to the task; it may reflect growing familiarity with instructions, response demands, or task strategies, and by itself does not show an increase in general intelligence. Its presence alongside relatively high ICCs is not contradictory. Participants can preserve much of their ordering compared with one another while scores on particular tasks shift with experience. For a buyer considering repeat use, the relevant question is therefore not merely whether a test is called reliable, but whether its results are expected to change after earlier administrations and how the provider handles that schedule.

The report also describes performance associations with age, phone operating system or device type, testing location, and whether a participant was alone. These associations mean that repeated administration took place in varied personal and environmental circumstances; they do not establish that any one circumstance caused a particular score change. They also underline why “same phone test” does not necessarily mean identical conditions across sessions. A quiet room and an interrupted setting can differ in ways that matter to attention, while a software or device variation may alter the delivery context. The abstract-level findings do not quantify a universal adjustment for these circumstances, so they cannot be used to correct an individual's result after the fact.

This evidence concerns one smartphone battery, its eight tests, and its ten-day remote protocol. It cannot supply norms for adults, transferable reliability for a different phone test, or proof that another provider's retest schedule produces the same practice pattern. The reasonable conclusion is narrower: mobile scores can show meaningful consistency while also varying by task, repeat exposure, and associated context. If a test is intended for one-time practice, those repeated-use questions may have little bearing on the purchase. If a person plans to track change by taking the same test repeatedly, the exact product's retest interval, alternate forms, practice findings, and device conditions become part of what its score can mean.

An ICC is also conditional on the people and range of performance observed. A repeated test can rank a group consistently in one sample and yield a different coefficient in a group with a narrower or broader spread of scores, even if the underlying task is unchanged. That statistical dependence is one reason not to lift the study's count or threshold and apply it to a product with different tasks, participants, or administration conditions. The reported ICCs answer a within-study repeatability question; they do not tell a reader how precise a single score is for a particular person, or whether a change from one sitting to the next is meaningful. Those are separate questions that require the relevant product's measurement evidence and retest design.

Sources: Characterizing Information Processing With a Mobile Device: Measurement of Simple and Choice Reaction Time

Why can matching average scores still hide uncertainty for one person?

A group average and an individual agreement question describe different properties. A group mean is the arithmetic center of scores for a specified set of people under a specified condition. Individual agreement asks how closely the paired scores for the same person correspond across conditions. Similar means can coexist with wide person-to-person differences; a shifted mean can coexist with close correspondence for most people if the shift is consistent. Neither pattern can be read from the mean alone. That is why a buyer should not treat “the groups scored similarly” as a complete answer to whether a particular person's phone score would match their score on another device.

Consider a hypothetical illustration, with no data implied: imagine two versions of a task administered to two groups. Their average results might be alike because some participants do better on the first version and others do better on the second, with those differences balancing when combined. The averages would conceal the direction and size of each person's paired difference. In another hypothetical, the second version might produce a small consistent shift for nearly everyone. The two group averages would differ, but each person's relative standing could remain close. These possibilities are not claims about a study; they show why a summary of group centers cannot answer a question about paired person-level agreement.

A further question is classification: would the score put a person on the same side of a consequential threshold under each condition? Even a small difference may matter for someone whose score lies close to a cutoff, while a larger difference may leave another person's category unchanged if their score is far from it. This is a logical implication of threshold use, not a finding about any specific test. To assess classification, a buyer would need evidence about the relevant score distribution, uncertainty around individual scores, and the decision rule. An average comparison supplies none of those by itself. A report that a mean difference is small therefore cannot establish that no individual could cross a threshold.

The distinction also clarifies what precision evidence should answer. For one person's result, the relevant question is how much observed score variation could reflect measurement error under the intended conditions and for the intended population. A useful account would identify the score being interpreted, the conditions under which its precision was estimated, and how uncertainty affects the use the buyer has in mind. If a decision relies on a boundary, evidence should show how score uncertainty relates to that boundary. This is different from asking whether two broad cohorts have similar averages. It is also different from asking whether a test can rank people consistently over repeated occasions; a ranking statistic can summarize a group pattern without giving a person's likely score range or threshold status.

“Comparison of PC and iPad administrations of the Cogstate Brief Battery in the Mayo Clinic Study of Aging” is useful here only as a reminder to keep the target of an empirical result clear. Its reported comparisons concern outcomes and cohorts in a particular PC/iPad battery study. Cohort-level results can inform whether large systematic differences appeared in those studied outcomes, but they do not by themselves show that the two administrations yield interchangeable values for every person. The study did not test smartphones, and the study title or its mean-level comparisons cannot be converted into a personal precision estimate for a different test. The inferential boundary follows from the kinds of quantities being compared, not from an extra result attributed to the authors.

For a purchase decision, ask what claim the provider makes about a single user's result. If the claim is only that the task offers practice or feedback, a group mean comparison may not be central. If the claim is that the score locates an individual relative to others or supports a cutoff decision, the buyer needs evidence that reaches the person-level interpretation: appropriate precision information, a relevant comparison basis, and a clear account of uncertainty at the score or threshold. Group similarity remains useful evidence against a large average shift in the studied outcomes, but it cannot carry those additional claims alone. A responsible conclusion is therefore bounded: average comparisons describe groups; confidence about one person's result requires evidence designed to support that individual use.

Sources: Comparison of PC and iPad administrations of the Cogstate Brief Battery in the Mayo Clinic Study of Aging

What does a phone result need to support a score claim?

A phone result supports a score claim only when the evidence connects the task, the recorded response, the scoring process, the people being compared, and the proposed use. The “Standards for Educational and Psychological Testing” frames validity as evidence and theory supporting a proposed interpretation and use of scores. That definition shifts the buyer's question from “Is this test valid?” to “What interpretation of this particular result is supported, for whom, under what conditions, and for what decision?” A narrow interpretation may be supportable even when a broader claim is not. A task can be useful for self-guided practice without supporting population ranking, selection, or diagnosis.

The first link is task coverage. A provider should make clear what abilities or operations its questions sample and how those tasks relate to the construct named in the score claim. If the assessment includes a few kinds of pattern or number problems, that description alone does not establish comprehensive coverage of general intelligence. The inference needs a rationale and evidence connecting the chosen tasks to the construct, including what is left out. The narrower the task sample, the more carefully a broad label needs justification. This is not a demand that every short activity measure everything; it is a requirement that the conclusion match what the activity actually samples.

The next link is administration. The evidence should describe the conditions that are part of the tested procedure: device and software versions, screen and input demands where relevant, timing rules, instructions, and any access arrangements. Those details matter because an observed response is produced within a particular setup. If the evidence was gathered on a different interface or under supervised conditions, a provider needs a reason to extend its interpretation to phone-only use. A difference in delivery does not automatically invalidate a result, but unexplained differences leave a gap between the studied task and the task the buyer receives. A claim about phone delivery should therefore be traceable to the actual version and conditions or to a defensible equivalence argument.

Scoring and comparison form another linked step. A raw count describes responses under a scoring rule; it does not become a standard score merely because a site converts it into an IQ-like number. A percentile describes standing against a defined reference group, so the group, sample, and conversion method matter. A validated adult comparison requires relevant evidence for adults and the particular interpretation being offered. These outputs are not interchangeable labels for one quantity. The conversion should be documented well enough to tell what the displayed number means and what population, if any, anchors it. If a conversion uses placeholder parameters or lacks an appropriate reference population, the resulting estimate cannot support claims of established adult standing, regardless of how polished its display appears.

The final link is the consequence attached to the result. Evidence adequate for reflection or choosing another practice exercise may be inadequate for a school placement, employment decision, diagnosis, or eligibility determination. The more consequential the proposed use, the more the supporting evidence must address the relevant population, administration, precision, and effects of error. This follows the Standards' interpretation-and-use framework; the Standards do not certify a particular phone product or supply its missing evidence. A provider should state the intended use and limits rather than letting a score label silently invite broader conclusions.

This chain helps locate exactly where a claim outruns its support. A well-designed set of tasks does not automatically validate a conversion; an appropriate conversion does not automatically establish phone equivalence; and evidence for one population or low-stakes use does not automatically transfer to another population or high-stakes decision. The individual-agreement question in the preceding section is one specific evidence need within this larger chain, not a substitute for it. A buyer can reasonably pay for an engaging activity while withholding belief from a score interpretation that lacks a documented bridge from task to use. The defensible claim is the strongest one each link supports—and no broader.

Sources: Standards for Educational and Psychological Testing

What should a provider disclose about a technology-based test?

Before paying, a buyer should be able to inspect enough information to connect the test being offered with the claims made for it. The “ITC/ATP Guidelines for Technology-Based Assessment” treats technology-based assessment as involving design, delivery, scoring and use, alongside concerns such as accessibility, fairness, privacy and security. That framework suggests distinct questions for a buyer: which version was examined, what evidence exists and for whom, how a displayed result is produced, whether the interface can be used accessibly, and what happens to submitted information. These are useful categories for locating material information. The guidance is professional advice, not a product rating, certification or finding that a provider satisfies each category.

Start with version and technical conditions. A short description should let a buyer identify what is actually being sold: the test or form, its software or delivery version, supported device and operating-system conditions, and any requirements that could affect responding, such as screen or input demands. This does not require a sales page to reproduce a technical manual. It lets the buyer tell whether evidence refers to the current phone experience or to a materially different setup. If technical details are available in a separate document or on request, the provider can say where to find them and identify the version they cover. Without that link, a general claim about a ‘mobile test’ is difficult to compare with the particular product on the screen.

Next, look for a plain account of the evidence and the people it concerns. A useful disclosure identifies what was evaluated, the relevant participant group, the conditions of administration, and the kinds of conclusions the evidence is intended to support. A result from one age group, language group, task set or supervised setting may not answer the same question for a different population using a phone at home. The point is not to demand that every provider publish a full research archive on its checkout page. The buyer needs a route to the underlying technical information, enough description to understand its scope, and a clear boundary around claims that reach beyond it. When no documentation is public, that is a limit on what a buyer can verify; it is not proof that no evidence exists privately or elsewhere.

Scoring notes answer a different question: what do the numbers and labels mean? A provider should explain whether it displays a count of correct responses, domain feedback, a comparison score, a percentile, or another output, and describe the basis for any conversion or comparison. Terms that look familiar can conceal different quantities. A buyer should be able to learn what reference group, if any, anchors a population comparison, and whether an estimate carries a stated uncertainty. Where a provider cannot substantiate a comparison, it should describe the result more narrowly as task feedback rather than inviting the reader to infer a population rank. This disclosure concerns interpretation of the output; it does not require a buyer to resolve every technical question before deciding whether the activity itself is enjoyable or useful.

Access statements should make practical conditions visible: whether the test depends on visual detail, sound, touch gestures, timed responses or other interface features, and what navigation or accommodation options are available. This helps a person judge whether the task can be completed as intended and whether the provider has considered different access needs. A general assurance that a site is ‘accessible’ is less informative than a description of the supported methods and any known limits. Privacy and security notices serve another purpose. They should explain what information is collected, why it is used, how long it is retained, whether it is shared, and what safeguards or user controls apply. These facts matter to the decision to disclose information, but a privacy notice does not establish anything about score quality.

A provider may reasonably keep detailed manuals or study reports outside the sales page, or supply them when asked. The buyer can still look for a clear route to that material and a direct answer to questions that matter for the intended purchase. The “ITC/ATP Guidelines for Technology-Based Assessment” helps organize those questions across the technology and assessment process; it does not certify a test, prescribe a single disclosure format, or establish that a specific provider has implemented every safeguard. A careful review therefore records what can be inspected, what remains unclear, and which uncertainty matters to the buyer's planned use. Missing public detail narrows the claims a buyer can personally verify, while leaving open the possibility that additional documentation is available from the provider.

Sources: ITC/ATP Guidelines for Technology-Based Assessment

What does the Test IQ Free fee buy?

Test IQ Free's offer separates a free practice experience from an optional paid explanation. The free test contains 50 original adaptive questions across pattern reasoning, number patterns and spatial reasoning. It provides worked answers, feedback by reasoning domain, and a provisional estimated IQ range. A person who mainly wants to attempt varied reasoning questions and inspect explanations can therefore try the activity without paying for the estimate or a report. The questions are original and the test adjusts difficulty during the form, but those design features do not by themselves establish that a resulting number supports a population comparison.

The estimate has a specific limitation that should shape how it is read: it comes from an unvalidated conversion with no validated adult reference population. Its range is not a confidence interval and is not an established comparison with the adult population. The display may look like a familiar IQ result, but the label and presentation do not supply validation, norms or evidence for a clinical interpretation. A reader can use the domain feedback and worked answers to see which kinds of items they found difficult and to choose what to practice next. They should not treat the range as a verified measure of general intelligence, use it to identify giftedness or a disability, or rely on it for a consequential decision.

The optional report costs $9. It adds a personalized explanation with the provisional point estimate and percentile produced by the same conversion, a three-domain interpretation, worked examples, everyday and work applications, and a 30-day practice plan. Paying therefore buys organization and interpretation of the exercise, along with practice structure. It does not buy a separate assessment, a more carefully normed score, or independent evidence about accuracy. The point estimate and percentile inherit the same conversion limitation as the free range; adding detail to a result cannot repair the evidence behind its measurement. The plan can structure practice with relevant reasoning operations, but it should not be understood as a promise to raise general intelligence.

The report's practical value is therefore in reducing the work of organizing feedback. A reader can review answers and domain feedback in the free experience, then decide whether a more connected interpretation and scheduled set of exercises would help them use that information. The report's examples and applications may make the material easier to revisit, while its plan can turn a general intention to practice into a sequence of sessions. Those are usability benefits, not measured improvements in ability. A buyer who does not expect to use the explanations or schedule may get little additional value from paying, even if they enjoyed the test itself.

Whether that report is worth $9 depends on what a reader wants from it. Someone who values a personalized account, worked illustrations and a ready-made schedule may reasonably prefer the report, while someone content with the free questions and answers has no need to pay for those materials. Either choice can be sensible when the purchase is judged as an experience or an aid to practice. The important distinction is between paying for explanation and paying for stronger measurement evidence: the report adds the former, not the latter. The product page, “Free IQ Assessment: 50 Adaptive Questions, Estimated IQ Range | Test IQ Free,” describes the offer; it is not independent validation of the conversion.

Sources: Free IQ Assessment: 50 Adaptive Questions, Estimated IQ Range | Test IQ Free

When should a phone-only score change your next step?

Let the consequence of the decision set how much weight you give the score. If the next step is simply to try another puzzle, choose a task that seems well made and use its explanations to decide what kind of problem to attempt next. The result can guide that small choice without establishing a broad claim about your ability. If the result could affect placement, diagnosis, employment, eligibility, or access to a service, do not let this browser score carry the decision. Seek a qualified assessment and consider other relevant evidence about the question at hand. The cost of an error is higher, so the evidence must support that particular use and the person being assessed.

This is a proportionate rule, not a claim that every phone assessment is unsuitable. A phone test could contribute to a consequential process if evidence for the exact version supports the intended interpretation, the administration conditions fit the examinee and task, and a qualified person interprets the result in context. Phone delivery alone neither disqualifies a result nor makes it dependable. The question is whether the full evidence chain described above reaches the decision being made. A device can be convenient while a particular score claim remains unsupported; conversely, careful validation and suitable use may make technology part of an appropriate assessment.

Professional administration by itself is no guarantee either. The assessment still needs to address the right question with an appropriate instrument, suitable conditions, and an interpretation that fits the available evidence. For a consequential decision, ask what other information should be considered and who is qualified to interpret it; do not substitute a polished score label for those answers. For a low-stakes practice choice, that level of scrutiny is unnecessary: the quality of the task and the usefulness of its explanations may be enough. If the result would not change a decision you care about, relying on it more heavily adds little. A score can prompt a useful question for an assessor, but it cannot answer that question on its own. Match reliance to the decision's stakes and the evidence available for that use.

What should you do next?

If you want a low-stakes reasoning activity, try the free test and review its worked answers; use them to notice which operations you want to practice, not to settle what you can achieve. Choose an unfamiliar pattern or number problem and work through why each answer option does or does not fit. That keeps the exercise useful even when you set the estimate aside. Its estimated IQ range comes from an unvalidated conversion with no validated adult reference population. The optional $9 report adds explanation and a practice plan, not validity, so pay only if that structure would help you use the feedback. You can also skip testing if neither practice nor a consequential assessment would be useful to you. If a decision about placement, diagnosis, employment, or eligibility is at stake, speak with a qualified professional about an assessment suited to that question and consider other relevant evidence. If you are unsure whether the decision is consequential, ask the person making it what evidence they need before treating any online result as relevant. Keep the test separate from decisions about your worth, potential, or identity; one task session cannot settle those questions. Take the free 50-question test at /test.

Sources: Free IQ Assessment: 50 Adaptive Questions, Estimated IQ Range | Test IQ Free

Questions readers ask