The g Factor Explained: The One Thing Every Cognitive Test Has in Common

If you remember a single idea from the science of intelligence, make it this one: the g factor. It rests on one of the most consistently replicated patterns in the study of individual differences, and it is the reason a single number can say anything meaningful about how someone reasons. Most of the concepts people argue about, from IQ to "brain training" to fluid versus crystallized ability, are downstream of g.

The discovery

In 1904 the psychologist Charles Spearman noticed something he didn't expect. Children who did well on one kind of mental test tended to do well on quite different ones. The tasks looked nothing alike, yet performance on them was positively correlated. To explain it, Spearman proposed a shared factor behind all of them: the general intelligence factor, or g (Spearman, 1904).

A century of data has kept the pattern intact. When researchers extracted g from three entirely different test batteries given to the same people, the three g factors correlated almost perfectly (Johnson et al., 2004). Whatever g is, it is not an artifact of one test.

The positive manifold: the evidence in one phrase

The core evidence has a name: the positive manifold. Give a large group of people many different mental tests and essentially all the correlations come out positive. Being capable in one cognitive area makes you statistically more likely to be capable in others. That does not make everyone a polymath; it tilts the odds. The correlations are moderate, which leaves plenty of room for real strengths and weaknesses within one person.

What g might be

So what is it? Here the honest answer is that nobody knows yet. Several accounts compete:

• One underlying capacity. The classic view: a general resource, perhaps related to processing speed, working memory or how efficiently the brain is organized, that every task draws on. Brain-imaging research has found correlates of g, but no single mechanism (Deary et al., 2010).
• Mutualism. Separate abilities that start out independent may boost each other during development, which would produce a positive manifold without any single cause (van der Maas et al., 2006).
• Process overlap. Every test taps several executive processes at once, and the overlap between tests produces the shared factor (Kovacs & Conway, 2016).

The summary most researchers would sign: g is well described statistically but not yet well explained biologically.

How a single score falls out of many tests

If g is invisible, how is it measured? Through factor analysis. By examining what a whole battery of subtests has in common, the variance running through all of them, psychometricians estimate a general factor. On a clinical battery such as the WAIS-5 or WISC-V, the Full Scale IQ is the closest practical estimate of g, because it averages across many varied subtests and cancels out task-specific strengths. An online estimate does not work this way; more on that below.

Fluid and crystallized: two faces of g

The most important refinement of Spearman's idea splits general ability into two broad expressions:

• Fluid intelligence (Gf) — reasoning about new problems; peaks in early adulthood and is the ability most closely tied to g.
• Crystallized intelligence (Gc) — accumulated knowledge and skill; vocabulary keeps growing into the 60s.

Both relate strongly to g. Fluid ability helps build the knowledge that becomes crystallized ability, which is part of why g predicts outcomes across a lifespan rather than just on test day. Modern models (Cattell–Horn–Carroll) place g at the top of a hierarchy, with a set of broad abilities such as fluid reasoning, knowledge, memory and speed beneath it (Carroll, 1993).

Why g matters in the real world

General ability predicts education strongly: in Strenze's (2007) meta-analysis of longitudinal studies, intelligence measured in youth correlated about .56 with years of education, .43 with occupational level and .20 with income. For job performance, the estimates have been revised. Schmidt and Hunter (1998) estimated ~.51; a 2022 reanalysis (Sackett et al.) lowered it to ~.31, still useful, no longer the clear leader, and below structured interviews in their analysis.

None of this makes g the whole story. Personality, motivation, health and opportunity do substantial work, and plenty of high-ability people underachieve. For a fuller treatment of the evidence and its limits, see the g factor, with the evidence and its limits on our sister site, My IQ Path.

See your profile

My Path's IQ test has a 12-item short form and a 60-item full form, both free, covering six areas: fluid, verbal, numerical and spatial reasoning, working memory and processing speed. Each area gets a percentile among people who took the same form. The overall figure weights the areas by published g-loadings rather than loadings factor-analysed from our own data — on the full form it ranks that weighted combination against everyone who took the full form, on the short form it averages the area percentiles — so it is a summary of your profile, not an extracted g. It is shown with a range and re-expressed as an estimated IQ (≈ N). It is an estimate, not a clinical IQ.

Take the free IQ test →

Because general ability sets part of the picture and personality steers what you do with it, pair your result with the Big Five and EQ assessments for the fuller picture.