CHC Theory: The Model Behind Every Modern IQ Test Score

Ask what an IQ test measures and most people picture a single dial, running from low to high. Ask a psychometrician and they describe a three-level structure that most test-takers never see, because a score report only ever shows the summary at the top of it. That structure has a name — Cattell-Horn-Carroll theory, CHC for short — and it is the model behind the index scores printed on almost every mainstream cognitive ability test given today, including the Wechsler scales this site writes about often.

Three separate ideas that turned out to be one theory

CHC theory is not one person's model. It is a merger of three research programs that developed mostly independently and turned out to describe the same structure. Raymond Cattell proposed, around 1941, that general intelligence splits into two broad components: fluid intelligence, the ability to reason through something unfamiliar in the moment, and crystallized intelligence, the knowledge and vocabulary built up through education and experience. John Horn, working with and after Cattell, expanded the model through the 1960s to the early 1990s into nine or ten broad abilities rather than just two. Separately, John Carroll spent years reanalyzing data from 461 earlier factor-analysis studies and published his own three-stratum theory in 1993 — general ability at the top, broad abilities in the middle, narrow specific skills at the bottom — built from evidence that had nothing to do with Horn's work. After a 1985 meeting connected to the development of the Woodcock-Johnson test, researchers recognized how closely the Horn and Carroll structures matched, and the combined model took the name of all three contributors.

The three levels, in plain terms

• Stratum III — general ability (g), the broad tendency for performance across very different kinds of mental task to correlate with each other
• Stratum II — seven to ten broad abilities, among them fluid reasoning (Gf), comprehension-knowledge (Gc), visual processing (Gv), short-term working memory (Gsm), processing speed (Gs), and long-term storage and retrieval (Glr)
• Stratum I — dozens of narrow abilities, each a specific skill that feeds into exactly one broad ability above it

Put in plain language rather than abbreviations: fluid reasoning is solving a pattern you have never seen before, using logic in the moment rather than facts you already hold — it is what a set of Raven's-style matrix items is built to measure. Crystallized intelligence is vocabulary and general knowledge, accumulated over years of schooling and reading. Working memory is holding and rearranging a small amount of information for a few seconds, the skill behind repeating a string of digits backward. Processing speed is how quickly easy, low-effort decisions get made under mild time pressure — not how hard a problem is, but how fast a trivial one gets cleared.

What sits inside a single broad ability

Stratum I is where the abstractions turn into actual test questions. Comprehension-knowledge (Gc), for instance, is not tested directly — it is built from narrower pieces such as lexical knowledge (the size and precision of someone's vocabulary) and general information (facts picked up through schooling and everyday life), which is why a vocabulary subtest and a general-knowledge subtest can both feed the same verbal index even though they look like different tasks. Fluid reasoning (Gf) breaks down similarly into pieces such as induction, working out the rule that governs a sequence, and quantitative reasoning, applying logical steps to numbers rather than recalling a math fact. A single broad-ability score is really an average across several of these narrow measurements, one more layer beneath the one most readers already know is an average.

Which real tests use which pieces

CHC theory is not confined to one publisher's test. The Woodcock-Johnson battery, now in its fourth edition, was developed in direct collaboration with the researchers building CHC theory, and its structure follows the broad-ability list closely. The Wechsler scales — the WISC-V for children and the WAIS-IV for adults — were substantially reorganized around CHC-style broad abilities in their most recent major revisions, which is why older editions of the same tests reported a simpler split (verbal versus performance) that the field has since moved past. The Stanford-Binet, in its fifth edition, follows a similar five-factor structure. None of these tests measures every broad ability CHC describes — that would make for an extremely long test — but each selects a subset and is explicit, in its own technical manual, about which ones.

Why the index scores on a real report exist at all

This is the direct explanation for something readers of our piece on full-scale IQ versus index scores have already run into: two index scores on the same report can pull apart even when the full-scale total looks unremarkable. CHC theory is the reason those indexes are separate, named quantities in the first place, rather than being folded into one number before anyone sees them. Every major index on a modern Wechsler-family report — verbal comprehension, visual-spatial, fluid reasoning, working memory, processing speed — corresponds to one of CHC's broad abilities. The general ability score at the very top of the model is the same g factor discussed in our explainer on the g factor.

A common misreading of the model

Separating fluid reasoning from working memory from processing speed does not mean the three have nothing to do with each other. They are typically positively correlated — someone who scores well on one broad ability tends to score somewhat above average on the others too, not randomly. That shared tendency across otherwise distinct tasks is exactly what the general-ability factor, g, at the very top of the model is describing. CHC theory does not compete with the idea of general intelligence; the three-stratum structure is what it looks like when general intelligence is broken down into its component parts rather than measured as one thing directly.

What to actually look at on a score report

The practical payoff of knowing this is simple: read past the composite. A profile where the index scores sit close together describes someone with a fairly even ability pattern across fluid reasoning, verbal knowledge, working memory and processing speed. A profile with a wide gap — strong fluid reasoning against comparatively weaker processing speed, for instance — describes something a single full-scale number cannot show at all, and comparing two such profiles across different tests is exactly what our score converter is built to help with, by putting both onto a common scale first.

The same total, two different people

The clearest way to see why the composite hides information is a simple illustration. Picture two test-takers who both land on a full-scale score of 115. The first has strong crystallized knowledge and average processing speed — deep vocabulary and general knowledge, working at an ordinary pace. The second has the reverse profile: strong fluid reasoning and fast processing speed, but a comparatively narrower knowledge base, perhaps because of less formal schooling rather than less reasoning ability. Averaged into one composite, both are "115." Read by index, they look like different people with different practical strengths — one better suited to tasks that reward accumulated knowledge, the other to tasks that reward reasoning through something new quickly. Neither profile is a better or worse kind of intelligence. They are simply not the same thing, and a single number cannot say which one it is describing.

The next time a score report lists several index names above the single composite, that list is not incidental formatting. It is CHC theory, printed in the order the model puts general ability, broad abilities and the tests built to measure them.

Common questions