WAIS-IV CHC Factor Structure in Older Adults

The published four-factor structure of the Wechsler Adult Intelligence Scale—Fourth Edition (Verbal Comprehension, Perceptual Reasoning, Working Memory, Processing Speed) is the structure most clinicians actually use. The published structure was also developed and validated mostly on samples that under-represented older adults. Whether the same factors operate, and operate the same way, in 80-year-olds as in 30-year-olds is not a foregone conclusion — cognitive abilities change with age, and the structure of those changes is itself an empirical question. Niileksela, Reynolds, and Kaufman (2013) addressed this directly, fitting an alternative five-factor model to the WAIS-IV standardisation sample’s older participants and testing whether the model held invariant from young adulthood into the eighth and ninth decades.

The standardisation problem at age 70+

The WAIS-IV standardisation sample includes participants aged 16–90, which is broader than any prior Wechsler edition. But the test does not give every subtest at every age. Several supplemental subtests — Letter-Number Sequencing, Figure Weights, Cancellation — are not administered to participants aged 70 and above. The motivation is practical (older participants are less likely to complete a long battery, and some subtests have ceilings that compress the older score range), but the methodological consequence is that any factor analysis using the full WAIS-IV battery has to drop participants over 70 from the analysis. Most published factor-structure work on the WAIS-IV has done exactly that, leaving the question of how the test’s structure looks in older adults under-investigated.

The earlier rounds of WAIS-IV factor-structure work also produced a methodological tension that post-2010 papers had to resolve. Benson, Hulac, and Kranzler’s (2010) independent factor analysis of the standardisation sample — published in Psychological Assessment — found that a five-factor Cattell-Horn-Carroll (CHC) model fit the data better than the published four-factor structure, separating visual processing (Gv) from fluid reasoning (Gf) within what the WAIS-IV calls Perceptual Reasoning. Weiss, Keith, Zhu, and Chen (2013) extended that finding to a clinical sample and argued for clinical interpretation of the five-factor approach. The resulting five-factor CHC model became the de facto research-grade alternative to the WAIS-IV’s published structure — and was eventually adopted by the WAIS-V in 2024, as covered in the broader history of the WAIS.

What Niileksela et al. (2013) did

The 2013 paper bridged the two threads. Niileksela, Reynolds, and Kaufman built an alternative five-factor CHC model that uses only the core WAIS-IV subtests — the subtests administered uniformly across the entire 16–90 age range — and tested whether this restricted model held invariant across age groups. The five factors mapped to standard CHC broad abilities: crystallised knowledge (Gc), fluid reasoning (Gf), visual processing (Gv), short-term memory (Gsm), and processing speed (Gs).

The results were broadly reassuring. The alternative five-factor model fit the standardisation data well in the 70–90 age band specifically, indicating that older adults’ WAIS-IV performance can be organised into the same theoretical factors that describe younger adults’ performance. More importantly, the model met criteria for partial strict measurement invariance across the lifespan, with the only significant non-invariance falling on the Similarities subtest. Partial strict invariance means that the underlying constructs measured by the model are essentially the same across ages, the relationships between subtests and factors are the same, and the precision of measurement is comparable — with one specified exception that does not derail the broader interpretation.

Why partial strict invariance matters clinically

The clinical implication is concrete. When a 75-year-old patient takes the WAIS-IV and produces an index profile, the clinician interpreting that profile relies on the assumption that the indices measure the same constructs in older adults that they measure in the standardisation reference. If the WAIS-IV’s factors operated differently in older adults — if Working Memory in 75-year-olds tapped different cognitive resources than Working Memory in 30-year-olds — then comparing an older patient’s profile against the population norms would be psychometrically suspect. Niileksela et al. (2013) showed that this concern is largely unfounded for the five-factor CHC interpretation: the constructs hold up.

The single non-invariance on Similarities is consistent with what cognitive-aging research has independently found about verbal-conceptual abilities: vocabulary and general knowledge are well-preserved into late life, but the abstract-categorisation demand of Similarities (“how are X and Y alike?”) may pull on slightly different resources at different ages. The honest reading is that one item set requires interpretive caution; the rest of the battery can be read against the standard CHC framework regardless of patient age.

The cognitive-aging context

The Niileksela et al. result sits inside a broader literature on cognitive change with age. Salthouse’s (1996) processing-speed theory of adult age differences in cognition, published in Psychological Review, argued that age-related declines on complex cognitive tasks can be largely accounted for by age-related slowing of basic information-processing operations. The factor-invariance finding is compatible with that account: the constructs do not change with age, but mean performance on speed-loaded constructs (Processing Speed especially, and to a lesser extent the speed-mediated portion of Working Memory and Fluid Reasoning) declines steadily across the lifespan. The shift is in the population mean, not in what the factors measure.

This framework also clarifies why interpreting older adults’ WAIS-IV profiles requires age-graded norms rather than an age-blind comparison. The same construct produces lower scores in older adults on average, and the index scores are normed against age-matched peers precisely so that “average for age” reads correctly. The link between age, processing speed, and intelligence is part of a broader pattern that the reaction-time and intelligence literature has documented for decades, and that average IQ scores across the lifespan reflect at the population level.

Cross-version evidence

Niileksela and Reynolds (2019) eventually extended the invariance question well beyond the WAIS-IV. Their multi-group factor analysis across six Wechsler scales (WPPSI-III, WPPSI-IV, WISC-IV, WISC-V, WAIS-III, and WAIS-IV) tested for factor invariance across versions and revisions, not just across ages within a single version. The result was the same broad message: the constructs measured by the Wechsler family are largely invariant, with successive revisions updating items and norms without disrupting what the underlying factors represent.

That finding is what makes longitudinal and cross-cohort research with the Wechsler scales possible. A WAIS-IV Working Memory Index measured in a 30-year-old in 2010 can be meaningfully compared against a WAIS-IV Working Memory Index in an 80-year-old in 2010, and against a WAIS-V Working Memory Index in a comparable participant in 2024 — not because the scoring is identical (it is not, owing to renorming and item updates), but because the latent ability the score indexes is the same.

Limitations and what remains open

Two limitations are worth flagging. First, the Niileksela et al. (2013) analysis is necessarily restricted to the core WAIS-IV subtests administered to participants over 70; the model could not test invariance for the supplemental subtests not given to that group. Whether the supplemental subtests would have shown invariance had they been administered is an open empirical question, though the consistency of the core-subtest result is suggestive. Second, the standardisation sample, while large and demographically representative for the United States, is a single national sample at a single time point; cross-cultural and longitudinal extensions are still relatively rare in the WAIS-IV factor-structure literature, with subsequent work in Greek, Australian, and other cohorts suggesting broadly similar but not identical patterns.

What this means in practice

For clinicians working with older adults, the Niileksela et al. (2013) finding licenses interpreting WAIS-IV index profiles in patients aged 70–90 against the standard five-factor CHC framework, with a small interpretive note around Similarities. For researchers designing studies that include both younger and older participants, the same finding licenses pooling factor-level analyses across ages, again with caution around Similarities. For the broader project of comparing the WAIS-IV against the WAIS-V, the older-adult invariance finding is one of several arguments that motivated the WAIS-V’s eventual move to a five-index structure: if the five-factor CHC model holds across the WAIS-IV’s full age range, formalising it in the next edition’s published structure was a natural step.

Frequently asked questions

What is the WAIS-IV’s published factor structure?

Four indices: Verbal Comprehension, Perceptual Reasoning, Working Memory, and Processing Speed. These index scores feed into a Full Scale IQ. The published four-factor structure is the basis for routine clinical interpretation.

How does the alternative five-factor CHC structure differ?

It splits the published Perceptual Reasoning index into two CHC broad abilities — visual processing (Gv) and fluid reasoning (Gf) — based on independent factor analyses (Benson et al., 2010; Weiss et al., 2013) showing a five-factor model fits better than four. The WAIS-V (2024) formally adopted this split.

Does the same factor structure work for older adults?

Largely yes. Niileksela, Reynolds, and Kaufman (2013) showed that a five-factor CHC model using only the core WAIS-IV subtests met criteria for partial strict measurement invariance across the lifespan into ages 70–90, with the only non-invariant indicator being the Similarities subtest.

Why are some WAIS-IV subtests not given to participants over 70?

Three supplemental subtests (Letter-Number Sequencing, Figure Weights, Cancellation) are not administered to participants aged 70 and older. The decision reflects practical considerations about test length, completion rates in older participants, and ceiling effects on certain subtests. The methodological consequence is that factor analyses incorporating supplemental subtests must restrict to participants under 70, which is why models like Niileksela et al. (2013) deliberately use only core subtests.

Do scores still mean the same thing in an 80-year-old as in a 30-year-old?

The constructs hold across ages (Niileksela et al., 2013), but the mean performance on speed-loaded constructs declines with age. This is why WAIS-IV index scores are normed against age-matched peers — an “average for age” score reads correctly for an 80-year-old precisely because the comparison is to other 80-year-olds, not to the population at large.