Verbal composite

Verbal IQ test that reports a raw score

Twenty-four items in three verbal formats — naming what two things have in common, completing a relation between words, and saying what a sentence actually commits its speaker to — scored as items correct out of items asked, per section and in total. Free, no account, no clock. There is no IQ figure at the end of it, and the panel that would have held one explains instead why a number there would have to be invented.

100% free
No signup
24 items, 3 formats
No IQ number
Every item explained

24 items in three sections of 8, four options each, no clock. What comes out is 24 numbers collapsed into four — one per section and one total. What does not come out is an IQ figure, because the number after the raw score is a conversion table built from a standardization sample, and there is no such sample behind anything on this page.

The three sections

1.: What kind of thing — verbal concept formation — how far up the ladder of abstraction an answer goes
2.: Relations between words — verbal relations — recognizing the relation, then applying it rather than the topic
3.: What a sentence commits you to — sentence comprehension — the scope of only, unless, few, neither and denial

Each section stops and explains itself before it begins. Every item was written for this page: the published verbal subtests are copyrighted, they are administered aloud with the examiner judging an open answer against a rubric, and a multiple-choice version of an open question is a different measurement whatever it is called.

The three worked examples are not in the scored set and tell you the answer afterwards. Every scored item is explained at the end whether you got it or not — that review is the part of this page worth staying for.

How to take the verbal composite

Three formats, eight items each, and a review that is the point of the exercise.

Read the worked example for each format, or skip them
Three examples, one per section, each showing the answer and the reasoning behind it afterwards. They matter more here than on most tests because two of the three formats have a specific convention: the first rewards the most general true class rather than the most obvious shared property, and the third asks what the sentence guarantees rather than what a reasonable person would assume. Neither is obvious from the item alone.
Work through the three sections in order
Each section stops and describes itself before its eight items start, and the instruction stays above every item inside it. Answer with 1 to 4 or by tapping. There is no way back to an earlier item and no clock anywhere, so the pacing is entirely yours — the total time is reported at the end as a fact about the run rather than as part of the score.
Read the profile, then the item-by-item review
The result gives the total, then each section out of eight with a confidence interval that is deliberately wide, then a paragraph on why nothing further is computed from those numbers. Below that, every scored item appears with its answer and the reasoning, whether you got it or not — the reasoning is the part that transfers to the next test you sit, and the score is the part that does not.

Technical specifications

Items	24 scored, three sections of eight, plus 3 optional worked examples outside the scored set. The section order is fixed; the items inside a section and the options inside an item are shuffled from the run seed
The three formats	What kind of thing — name the class two objects share. Relations between words — complete the second pair so the relation matches the first. What a sentence commits you to — pick the statement the sentence guarantees, from four that a reader might assume
Options and chance	Four per item, one correct, so 24 items answered blind average 6 right. The result reports the exact one-sided binomial probability of your own total arising from guessing, which is the only figure on the page that puts your score against anything
What is reported	Items correct out of items asked, per section and overall, with a 95% Wilson interval on each. Nothing is converted, weighted or summed into an index, because a composite score is a rank and a rank needs a sample
Why the intervals are wide	Eight items per section is enough to sort a strong performance from a weak one and nowhere near enough to separate two sections from each other. A gap of one or two items between sections is inside the interval on both, so the profile points at what to run again rather than at a strength or a weakness
The metric this page does not use	A published verbal index converts each subtest's raw count to a scaled score of mean 10 points, SD 3 within the taker's own age band, sums those, and looks the sum up in a second table to produce an index of mean 100 points, SD 15. Both tables come from the standardization sample. This page has neither table and does not approximate one
Item provenance	Every item was written for this page. The published verbal subtests are copyrighted, are administered aloud with an examiner judging an open answer against a rubric, and are controlled precisely so that the items do not circulate — a score here is therefore not comparable to a score from any of them
What leaves the page	Nothing unless you press copy, which puts the section totals and the run seed on your own clipboard. No age is asked for, which is deliberate: the only thing an age could be used for here is a conversion this page will not perform

Frequently asked questions

Why won't this page give me an IQ number?

Because the number would have to be manufactured, and it is the kind of manufactured number people repeat about themselves for years. An IQ figure is a position in a distribution, and the distribution comes from a standardization sample — a few thousand people chosen to match a census, tested one at a time by trained examiners under conditions held constant. Everything after the raw count depends on that sample: which age band you are compared against, how many raw points a scaled point is worth, where the sum lands on the index. No unsupervised page has any of it, and the people who have answered these particular items are whoever happened to find this page, which is not a population. So the run stops at the last defensible number, which is the count.

What exactly is a verbal IQ inside a real battery?

It is an index computed from several subtests, not a test in itself. Each subtest produces a raw count, that count is converted against people of the taker's own age into a scaled score with a mean of 10, the scaled scores of the subtests making up the index are added, and the sum is looked up in a table that turns it into an index score with a mean of 100. Two conversion tables therefore stand between the answers and the familiar number, and both of them are the standardization sample written down. That is also why the same performance can produce different index scores on different batteries: they compose the index from different subtests and normed it on different samples in different decades.

English is not my first language. What does a low score here mean?

Mostly that you have met less English, which is a fact about exposure rather than about reasoning. Verbal measures are the most schooling-dependent and reading-dependent part of any battery — the items are built out of the vocabulary and the sentence conventions of one language, so a person tested outside their strongest language is being measured on their exposure to the test's language first and on anything else second. That is a known limitation of verbal testing rather than a quirk of this page, and it is the reason a level test is the more useful instrument when the language itself is the open question.

The real subtests ask you to explain in your own words. Why is this multiple choice?

Because a browser cannot judge an open answer, and pretending otherwise would be worse than the compromise. In an administered session an examiner scores your explanation against a rubric with partial credit: naming the class two things share earns full marks, naming a shared property earns half, and a wrong or over-specific answer earns nothing. Multiple choice collapses that gradient into right or wrong, and it also hands you the answer set, which makes recognition do work that production used to do. The design response here is to make the best option a genuine class every time and to put the near-miss property in the options where you can see it — so the item still teaches the distinction the rubric was measuring, even though the scoring cannot.

Can I compare this with a score from an assessment I sat at school?

No, in either direction. Your school score came from a normed instrument with an examiner, a fixed administration and a sample behind it; this run came from 24 items written for a web page and answered under whatever conditions you happen to be in. There is no conversion between them and nothing on this page approximates one. What does transfer between the two is format familiarity — knowing that unless does not work backwards, or that a class beats a property, is worth real items on any verbal test, and that is the transferable part of what happens here.

Which of the three sections should I trust least?

Trust the total more than any single section, and treat a gap between two sections as a suggestion rather than a finding. With eight items, a 95% interval on a section score spans a large part of the scale — a 6 and an 8 out of 8 have intervals that overlap heavily, so a two-item difference is well inside the noise. The reason the sections are reported separately anyway is that they point somewhere: if the sentence section is where the items went, that is a specific and fixable thing to work on, and it is worth knowing even when the difference is not statistically anything.

Are these questions taken from a published test?

None of them. The published verbal item banks are copyrighted and several are controlled specifically because a leaked item stops measuring anything — somebody who has seen it is remembering rather than reasoning. The items here were written from the formats rather than from the banks, which is why the page can show you every one of them with its answer at the end: nothing here is worth protecting, because nothing here is the instrument. It also means practicing on this page cannot leak into a supervised administration of a real one.

What the number after the raw score is made of

The interesting fact about an IQ figure is that almost none of it comes from the test questions. Answering items produces a raw count and nothing else; the number people quote is that count run through two conversions, each of which is a table built by testing a standardization sample. The first table converts your count into a subtest scaled score metric — mean 10 points, SD 3 — measured against people in your own age band, because the same raw performance means different things at nineteen and at sixty-four. The second turns the sum of those scaled scores into the index, which is constructed with mean 100 points, SD 15 — The Wechsler scales are constructed with a mean of 100 and a standard deviation of 15. Take the tables away and the questions are just questions. That is not a technicality: the tables are the reason the test is sold under a qualification requirement, and The norm tables are the product. They are copyrighted, sold with the test kit under a qualification requirement, and reproducing them would be both an infringement and an invitation to treat a browser imitation as a Wechsler score.

There is a second reason this page asks nothing about you, and it is the one people find surprising: an age would be the missing input, and having it would make the refusal harder rather than easier. Knowing your age lets a page look confident — it can talk about age bands and adjustment while still having no sample in either — and that is precisely the appearance of rigor that makes an invented figure believable. So the page collects nothing, converts nothing, and reports the count. What it does spend its length on instead is the review: the class-versus-property distinction in the first section, the relation-versus-association trap in the second, and in the third the single move that costs more items than anything else on the page — reading a one-way conditional as though it worked both ways. Somebody who leaves knowing that unless the rain stops, the match is off says nothing whatever about a dry afternoon has got something out of the run that a number could not have given them.

The verbal half of a battery is also the half most sensitive to what you have read and where you were schooled, which is worth holding beside your own result. Vocabulary, verbal relations and comprehension of written English are built out of exposure to written English, so a low section here is at least as likely to be a fact about reading history as about anything more permanent — and if the language itself is the open question, the English level test is the instrument that was built for it. For the pieces separately: word meaning under four options is the synonym test, breadth of vocabulary with a correction for over-claiming is the vocabulary test, the same entailment traps stripped of their verbal dressing are the logical reasoning test, and inference over a passage rather than a sentence is the reading comprehension test.

A reaction time here is the interval between the frame that painted the stimulus and the timestamp the browser attached to your key, both read from the same monotonic clock. What neither can see is the display pipeline behind it, so on a 60 Hz screen roughly 16 ms of every figure below is the machine rather than you. That is the timing floor: two numbers closer together than that are the same number, and this page reports no precision it cannot support.

That floor is why the only figure in milliseconds anywhere on this page is a rule: anything answered within 150 ms of an item being painted is discarded, since four options cannot be read in that time. Everything the run reports about duration is in seconds and minutes, and none of it enters the score — an untimed verbal item answered in ten seconds and the same item answered in ninety are the same item correct.

Not affiliated with, endorsed by, or connected to the owner of Wechsler Adult Intelligence Scale. Wechsler Adult Intelligence Scale is a trademark of its owner and is used here only to name the assessment this page prepares for. The questions on this page are our own: no part of the published test is reproduced, and a score here is not comparable to a score from the real instrument.

This is a measurement exercise, not a clinical assessment. It reports what you did on this page against a stated reference and nothing more — it cannot establish a language disorder, a learning disability or giftedness. Only a qualified professional, working with more than a browser, can make that judgment.

Where the 24 answers are compared

Every number on this page is worked out by JavaScript running in the tab you are reading it in. Your answers, your reaction times and your score are never uploaded, logged or kept — which is also why the test carries on working after you disconnect from the network, and why nothing here can be held back behind an email address.

The items, the answer keys and the explanations all shipped with this page, so every comparison happens in this tab and no answer is sent anywhere. Nothing about you is collected at all — not an age, not an email, and not the section totals, which exist only in this tab until you close it or press the copy button.