Skip to content
IQCognify

Our Methodology

How the IQCognify test is built, administered, and scored — including the scale we use, how the score is estimated and what it is not, how we keep scoring secure, and an honest account of what an online test can and cannot measure.

Last updated · IQCognify

What we are measuring

IQCognify aims to estimate general cognitive ability — the broad capacity that psychologists often refer to as the g factor. Most of our items are reasoning tasks rather than tests of factual recall: they ask you to identify patterns, hold and manipulate information, reason abstractly, and solve novel problems. General ability is the construct that traditional IQ instruments are designed to capture, and it is the most studied dimension of cognitive measurement. How well our short test captures it has not been independently established.

No online test, including ours, is a clinical instrument. This page describes the methods we use to make our estimate as fair and meaningful as possible — not a claim that the result is a diagnosis.

Item types

The test draws on several reasoning item formats that are widely used in cognitive testing. Some depend more than others on language and schooling — the verbal items require English, and the numerical items require basic arithmetic. We have not run our own factor analysis, so we do not claim that our particular items measure general ability to any established degree:

  • Matrix and pattern-completion items — identify the rule governing a visual sequence and select the missing element.
  • Numerical and logical sequences — infer the underlying relationship in a series and extend it.
  • Spatial and visual reasoning — mentally rotate, fold, or transform figures.
  • Verbal reasoning — analogies and relational logic that test reasoning rather than vocabulary trivia.
  • Working-memory and processing items — short tasks that gauge how much information you can hold and how quickly you can operate on it.

Each item has a difficulty value that we assigned by hand when writing it, chosen so the test spans a wide range — easier items at the lower end, harder items at the top. These values have not yet been calibrated against real response data, so an item we rated hard may in practice be easier or harder than intended.

How the order works: the test uses a fixed, progressive-difficulty form — the same set of items presented from easiest to hardest for every test-taker. It is not a branching computer-adaptive test that changes which question you see next based on your previous answers, and we deliberately avoid calling it 'adaptive' for that reason. The items were written to cover a wide ability range, though how well they actually cover it has not been measured.

The scoring scale

We report results on the familiar IQ scale, which is defined to have a mean of 100 and a standard deviation of 15. On a professionally normed test, that scale describes a real population: 100 is the average of a large, representative sample, roughly two-thirds of that sample falls between 85 and 115, and about 95% falls between 70 and 130.

Our score is placed on that scale differently, and it is important to be clear about how. It is a model-based estimate, not a norm-referenced IQ score. We first estimate your ability from which items you answered correctly, using a standard item-response model (the Rasch model) and the difficulty value of each item. We then convert that ability estimate to the IQ scale with a fixed formula: 100 plus 15 times the estimate. The percentile we show is the one that number would correspond to if abilities were distributed exactly as the model assumes.

What this means in practice: your score has not been compared with the scores of any measured group of people. A result of 100 does not mean you scored at the average of IQCognify test-takers, or of your age group, or of the general population — it means your answers were consistent with an ability level the model places at the centre of its scale. Treat the number, its percentile and its classification band as an estimate from our model, not as a ranking against real people.

Age and norms: current status

Cognitive performance — particularly on speeded reasoning tasks — varies systematically with age. Comparing a teenager's raw score directly against a 60-year-old's would be misleading, which is why professionally administered IQ tests compare each person with others of the same age.

IQCognify does not currently do this. Before the test we ask only whether you are 13 or older, and we do not store that answer. Your age is not used in scoring: a 16-year-old and a 60-year-old who give the same answers receive the same number.

We also do not currently use collected performance data to establish norms. Your score is calculated from your own answers and the model described above; the results of other test-takers play no part in it.

In short, our scale has no empirical norms yet — not by age and not for the population overall. If we introduce norms based on real performance data, we will describe the sample and method on this page before relying on them.

How the eight cognitive abilities are scored

Your result breaks down into eight cognitive abilities (your 'Cognitive DNA'). Every one of them — including Working Memory and Processing Speed — is measured from your performance on dedicated items written for that ability, not inferred indirectly from unrelated questions:

  • Working Memory is measured with short span and n-back-style items that ask you to hold a sequence in mind and operate on it (for example, recalling or reversing a string of digits).
  • Processing Speed is measured with quick symbol-matching and comparison items. Your score on this strand, like every other, is computed from accuracy on those items. We also record how long each answer takes and use it to describe your speed–accuracy style, but response time does not raise or lower any score.
  • The other six abilities — pattern recognition, fluid reasoning, strategic thinking, numerical, spatial, and verbal — are each measured from their own reasoning items.
Because the test is short, each sub-score rests on a small number of items and therefore carries more measurement error than your overall score, which pools every item. Treat the eight-ability breakdown as a profile of relative strengths and weaknesses, and the overall score (with its confidence range) as the more precise figure.

Who the test is for

IQCognify is designed and scored for adults and older teenagers — ages 13 and up. Before the test begins we ask for a self-reported age; anyone who indicates they are under 13 is not shown the assessment and is directed to guidance for parents instead.

Children's cognitive testing is a specialised field that requires individually administered, age-appropriate instruments delivered by a trained professional. An online test is not normed or validated for children and should never drive a decision about a child. See our guide on IQ tests for children for what to do instead.

Server-side scoring and anti-cheat

Scoring integrity matters because a test that can be gamed produces meaningless numbers. We protect it in several ways:

  • Answer keys never reach your browser. Correct answers and the scoring logic live on our servers; the client only ever sends your responses and receives the computed result.
  • Your score is computed on our servers and returned in a signed result token, so a result cannot be altered or forged after it is issued.
  • The scorer only accepts items it recognises. Responses referring to anything outside our item bank are discarded rather than scored.
  • Response times are recorded as reported by your browser. They are used to describe your speed–accuracy style and do not affect any score.
  • The test is a fixed form: everyone sees the same items in the same order. That means it is not protected against someone who has already seen the questions, and a retake will partly measure your memory of the first attempt rather than your reasoning.

These measures protect the integrity of the scoring, not the secrecy of the questions. IQCognify is an unproctored online test: it is built to give an honest reading to someone taking it in good faith, and it cannot prevent someone determined to game it.

Confidence ranges

Every measurement carries error, and a responsible estimate should say so. Alongside your point score we report a range. It is calculated by the same model that produces the score: it reflects how much information your answers carried, given the difficulty values we assigned to the items.

How far to trust the range: it rests on the model's assumptions — that the item difficulties are correct and that the model fits how people actually answer. We have not tested it against real-world data, so we cannot say how often a person's true ability actually falls inside it. Read it as a rough indication of uncertainty, not as a validated 95% interval.

Even so, the range is a better guide than the single number. Two people whose ranges overlap should not be ranked against each other on the basis of their point scores alone.

Accuracy and limitations

We are committed to being honest about what our test is not:

  • It is an estimate. A short online assessment cannot match the precision of a multi-hour, individually administered clinical battery.
  • It is not a norm-referenced IQ score. It is not compared with any measured population, is not adjusted for age, and has not been linked or equated to any clinical IQ test, so it should not be expected to match one.
  • Its reliability and validity have not been established. We have not published evidence of how consistently it measures, or how well it predicts anything else.
  • Conditions are uncontrolled. Distraction, fatigue, screen size, motivation, and test-taking experience all affect a single sitting.
  • Practice effects are real. Taking many similar tests can inflate scores without reflecting a genuine change in ability.
  • It does not diagnose anything. The test cannot identify learning disabilities, giftedness, cognitive decline, or any clinical condition.
If you need an IQ result for a clinical, educational, legal, or employment decision, seek a formal evaluation from a licensed psychologist. IQCognify is intended for educational and entertainment purposes only.