What are frequently asked questions?

It is the formula derived by Tellegen and Briggs (1967) for computing the mean, standard deviation, and reliability of an unweighted sum of k subtests, given the constituent subtests' means, SDs, reliabilities, and inter-correlations. The result lets an analyst convert subtest scores into a composite scale score with defensible psychometric parameters. When the composite you want to use is not in the published manual. Wechsler short forms, custom research composites, special-population adaptations, and forensic-specific composite indices are the standard cases. If the manual already supplies the composite parameters you need, use those directly; Formula 4 is for the cases the manual doesn't cover.

Mosier, C. I. (1943). On the reliability of a weighted composite. Psychometrika, 8(3), 161–168. https://doi.org/10.1007/BF02288700 Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric theory (3rd ed.). McGraw-Hill. Tellegen, A., & Briggs, P. F. (1967). Old wine in new skins: Grouping Wechsler subtests into new scales. Journal of Consulting Psychology, 31(5), 499–506. https://doi.org/10.1037/h0024963

Tellegen-Briggs Formula 4 for Composite Scores

By Xavier Jouve, Ph.D., Psychometrician

Published December 19, 2023 · Reviewed May 7, 2026

1,719 words · 7 min read · 3 references cited

A psychologist administering a partial Wechsler battery — say, four core subtests rather than the full ten — needs a way to convert the resulting subtest scores into a composite IQ-style scale score. The full battery has published norms with means, standard deviations, and reliability coefficients tabulated by age band; the four-subtest abbreviation does not. Tellegen and Briggs (1967), in Old Wine in New Skins, derived the formula now widely known as Tellegen-Briggs Formula 4 to handle exactly this case: given the means, standard deviations, intercorrelations, and reliabilities of the constituent subtests, compute the mean, standard deviation, and reliability of the unweighted sum and the standard score it implies.

The formula has remained in continuous use for sixty years, with implementations spanning Wechsler short-form research, special-population adaptations of standard batteries, and any context where standardization data for a particular composite were never collected. The Cogn-IQ statistical-tools suite implements Formula 4 as an open web calculator, taking the constituent subtest parameters as input and returning the composite scale-score statistics directly.

What Formula 4 computes

The setup: an analyst has k subtests that they want to combine into a single composite by simple summing. Each subtest has a known mean, standard deviation, and reliability from the full standardization sample, plus the inter-subtest correlations among the k subtests. Formula 4 derives three quantities:

Figure 1. Formula 4 turns constituent subtest parameters into composite mean, SD, and reliability, then onto any standard-score scale (Tellegen & Briggs, 1967).

The mean of the unweighted sum (trivially, the sum of subtest means);
The standard deviation of the unweighted sum (a function of subtest SDs and inter-correlations);
The reliability of the unweighted sum (a function of subtest reliabilities, SDs, and inter-correlations).

From these, the composite can be transformed to whatever standard-score scale the receiving framework uses — IQ-points (mean 100, SD 15), T-scores (mean 50, SD 10), z-scores, or any other linear transformation. The composite reliability, in particular, lets the analyst report a defensible standard error of measurement for the composite score, which is the operationally important quantity for clinical interpretation.

The formula’s structural insight is that composite reliability depends on the pattern of inter-correlations, not only on subtest reliabilities individually. A composite of two perfectly reliable but perfectly independent subtests will have lower reliability than a composite of two moderately reliable but correlated subtests, because the variance attributable to true score scales differently with the correlation structure. Mosier (1943), in On the Reliability of a Weighted Composite, derived the general weighted-composite reliability result decades before Tellegen and Briggs; Formula 4 is the unweighted-summing special case that practitioners typically need.

The 1967 historical context

Tellegen and Briggs’s paper was a methodological response to a practical problem of the period. The Wechsler-Bellevue and the early Wechsler Adult Intelligence Scale (WAIS) had standardization tables for the full battery and for the standard Verbal/Performance subdivision, but applied researchers were increasingly grouping the published subtests into novel composites — sometimes for forensic neuropsychology, sometimes for short-form clinical screening, sometimes for research-specific factor structures. None of these novel composites had standardization tables, and most researchers were either inventing ad hoc weights or reporting raw sums without a defensible scale-score interpretation.

The Old Wine in New Skins title captures the methodological move. The Wechsler subtests were the wine; the new groupings were the skins. The Tellegen-Briggs formulas — there are several, addressing related composite-score problems — let researchers package the old wine in new skins without losing the standardization information that made the original subtests useful in the first place. Formula 4 specifically addresses the common case: an unweighted sum of k subtests that share known parameters.

Sixty years later, the formula is taught in Wechsler training programs, embedded in proprietary scoring software for short-form research, and implemented in open calculators like the Cogn-IQ tool. Its longevity reflects the fact that the underlying problem — building defensible composites from existing standardized components — has not gone away, and the formula’s mathematical economy makes it hard to improve on for routine practice.

Practical use in modern psychometric work

The most common contemporary applications of Formula 4 fall into four buckets:

Wechsler short-form research. Researchers evaluating two-, three-, or four-subtest Wechsler abbreviations (Lace 2022 on WISC-V short forms is the recent canonical example in the WISC literature) routinely need composite reliability and SEM estimates that are not supplied in the full-battery manual. Formula 4 supplies them directly from published full-battery parameters.

Special-population adaptations. When a standard battery is used in a non-standardized population — a non-Western language adaptation, an older age band beyond the manual’s published range, a clinical sample with restricted ability range — the composite parameters need to be recomputed from the population’s empirical correlation structure. Formula 4 is the recomputation tool.

Niche or research-specific composites. Investigators building composites of cognitive measures across instruments — a working-memory composite that draws from WAIS Digit Span, a custom n-back, and a corsi-block task, for example — need a way to combine the components into a single reliable composite score with reportable parameters. Formula 4 handles the heterogeneous-instrument case as long as the constituent measures’ parameters are known.

Forensic and clinical assessment. Forensic neuropsychologists who construct composite indices for specific legal questions — a memory composite for capital-case Atkins evaluations, an attention composite for civil disability determinations — need reliability and SEM estimates that survive expert challenge. Formula 4 produces them with derivation transparency that is difficult to dispute on methodological grounds.

Known limitations and modern alternatives

Formula 4 is not without caveats. The formula assumes that the constituent subtests’ parameters (means, SDs, reliabilities, inter-correlations) are known without error from the standardization sample. In practice, those parameters are estimates with their own sampling variability, and a Formula 4 result computed from estimated rather than known parameters inherits that variability. For most clinical applications the inherited error is small; for research applications with unusually small standardization samples or noisy reliability estimates, the inheritance can be material.

Figure 2. Formula 4 remains the cost-effective routine choice; modern alternatives handle its limitations more gracefully but demand more analytic infrastructure.

The formula also assumes linear scaling and approximately normal subtest distributions. For subtests with strong floor or ceiling effects — the high and low ends of the distribution, where many cognitive tests have sparse standardization data — Formula 4 produces composite estimates that can underestimate scores at the high end and overestimate at the low end by a typical 2-6 points. This is a known property of the formula and is documented in the calculator’s user notes; for routine clinical use it is manageable, for extreme-score cases (gifted assessment above 145, severe-impairment assessment below 55) the analyst should treat the Formula 4 estimate as a starting point rather than a final answer.

Modern alternatives — Bayesian composite estimation with informative priors, full IRT-based composite scoring with item-level calibration, structural equation modeling with latent composite factors — handle the Formula 4 limitations more gracefully but require correspondingly more analytic infrastructure. For the routine case where the analyst has full-battery parameters and wants a reportable composite, Formula 4 remains the most cost-effective approach. For the unusual case where the limitations bite, the modern alternatives are worth the investment.

Where this fits in the broader psychometric tooling landscape

The Cogn-IQ statistical-tools suite implements several composite and reliability formulas that share Formula 4’s general lineage. Cronbach’s alpha is the most widely-cited reliability statistic but applies to single scales; Formula 4 is its composite-of-scales analogue. Equating methods handle the related problem of putting different forms on a common scale; Formula 4 handles the related problem of putting a sum of subtests on a meaningful composite scale.

The unifying lesson across these tools is that psychometric reporting is more than a single statistic. A complete composite report includes the composite mean, standard deviation, reliability, standard error of measurement, and ideally a confidence interval around the obtained score; Formula 4 produces the first four directly and supports the fifth with its computed SEM. The Cogn-IQ calculator automates the computation and supplies a citable output suitable for inclusion in research papers and clinical reports.

Frequently Asked Questions

What is Tellegen-Briggs Formula 4?

It is the formula derived by Tellegen and Briggs (1967) for computing the mean, standard deviation, and reliability of an unweighted sum of k subtests, given the constituent subtests’ means, SDs, reliabilities, and inter-correlations. The result lets an analyst convert subtest scores into a composite scale score with defensible psychometric parameters.

When do I need Formula 4 instead of a published composite score?

When the composite of interest is not in the published manual. Wechsler short forms, custom research composites, special-population adaptations, and forensic-specific composite indices are the standard cases. If the manual already supplies the required composite parameters, those can be used directly; Formula 4 is for the cases the manual does not cover.

How accurate is Formula 4 at extreme score ranges?

It can underestimate scores at the high end of the distribution and overestimate at the low end by a typical 2-6 points. This is a known limitation of linear scaling assumptions in the underlying derivation. For routine clinical use the inaccuracy is manageable; for gifted assessments above 145 or severe-impairment assessments below 55, treat the Formula 4 estimate as a starting point requiring qualitative interpretation.

What inputs does Formula 4 require?

The means, standard deviations, reliabilities, and inter-correlations of the constituent subtests, all from the relevant standardization sample. For Wechsler short-form research, these come from the full-battery manual. For custom composites or research-specific applications, the analyst computes them from the empirical sample.

Is the Cogn-IQ Tellegen-Briggs Formula 4 calculator free?

Yes. The calculator at cogn-iq.org/statistical-tools/tellegen-briggs-formula-4-calculator is web-based, requires no signup, and is intended for routine research and clinical use. The page also documents the formula and its limitations alongside the calculator interface.

References

Mosier, C. I. (1943). On the reliability of a weighted composite. Psychometrika, 8(3), 161–168. https://doi.org/10.1007/BF02288700
Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric theory (3rd ed.). McGraw-Hill.
Tellegen, A., & Briggs, P. F. (1967). Old wine in new skins: Grouping Wechsler subtests into new scales. Journal of Consulting Psychology, 31(5), 499–506. https://doi.org/10.1037/h0024963

Related questions

What is psychometrics?

The discipline of psychometrics emerged from two distinct yet complementary intellectual traditions. The first, championed by figures such as Charles Darwin, Francis Galton, and James McKeen Cattell, emphasized the study of individual differences and sought to develop systematic methods for their quantification. The second, rooted in the psychophysical research of Johann Friedrich Herbart, Ernst Heinrich Weber, Gustav Fechner, and Wilhelm Wundt, laid the foundation for the empirical investigation of human perception, cognition, and consciousness. Together, these two traditions converged to form the scientific underpinnings of modern psychological measurement. Read more →

How do online IQ tests compare to professional assessments?

A quick search for "IQ test" returns dozens of websites promising to measure your intelligence in 10 minutes. Meanwhile, a professional cognitive assessment takes 2–3 hours, costs hundreds of dollars, and requires a trained psychologist. Are the free online versions worth anything, or are they little more than entertainment? The honest answer is more nuanced than either pole of the debate suggests: the gap is real, but it is not principally about online versus in-person. It is about psychometric standards versus their absence, and the modal free quiz on the internet has no psychometric standards at all. Read more →

What's the difference between the WAIS-IV and WAIS-V?

Pearson released the Wechsler Adult Intelligence Scale, Fifth Edition (WAIS-5) in late 2024 — the first major revision since the WAIS-IV appeared in 2008. For the world's most widely administered adult IQ test, sixteen years between editions is a long time, and the new version reflects what cognitive science learned in those years. The headline change is structural: the Perceptual Reasoning Index has been split into two indices, separating fluid reasoning from visual-spatial ability. Five new subtests have been added, the Full-Scale IQ administration has been streamlined to about 45 minutes, and the norms have been refreshed to a 2023-24 sample. For clinicians the question is when to switch. For test-takers and families, the more practical question is what a WAIS-5 score means when the prior reference point was a WAIS-IV — and how the two compare when nobody has yet published a peer-reviewed cross-edition equivalence study. Read more →

How do you interpret IQ test results?

You've received an IQ test report — for yourself, your child, or a client — and what should be a clean answer is a thicket of numbers, percentiles, confidence intervals, index scores, scaled scores, and qualitative descriptors. This guide walks through what each piece actually means and how a psychometrician reads them. The short version: a single Full-Scale IQ number is rarely the most useful piece of information in the report, score discrepancies need both statistical and base-rate scrutiny before they mean anything clinically, and almost every "IQ point" carries a margin of error larger than most readers assume. Read more →

Why does the paper's 1967 historical context matter?

Tellegen and Briggs's paper was a methodological response to a practical problem of the period. The Wechsler-Bellevue and the early Wechsler Adult Intelligence Scale (WAIS) had standardization tables for the full battery and for the standard Verbal/Performance subdivision, but applied researchers were increasingly grouping the published subtests into novel composites — sometimes for forensic neuropsychology, sometimes for short-form clinical screening, sometimes for research-specific factor structures. None of these novel composites had standardization tables, and most researchers were either inventing ad hoc weights or reporting raw sums without a defensible scale-score interpretation.

How is Formula 4 used in modern psychometric work?

The most common contemporary applications of Formula 4 fall into four buckets: Wechsler short-form research. Researchers evaluating two-, three-, or four-subtest Wechsler abbreviations (Lace 2022 on WISC-V short forms is the recent canonical example in the WISC literature) routinely need composite reliability and SEM estimates that are not supplied in the full-battery manual. Formula 4 supplies them directly from published full-battery parameters.