Broad Score or Subscale? How to Choose a Follow-Up Question

What does each score level help you see?

A broad score gives you the report’s wider view; a narrower subscale gives you a more particular angle. Neither is the single correct score for every question. Read them as different levels of description, and check how the assessment itself defines each level before drawing meaning from the labels.

In personality measurement, a domain or trait is a broad pattern summarized across related behaviors or responses. A facet or subscale is a narrower component within that wider pattern. Names vary: one instrument’s “subscale” may not match another instrument’s facet, and a report may use its own groupings. The technical guide, rather than the familiar sound of a label, is the authority on what was measured.

The NEO Personality Inventory–Revised (NEO-PI-R) is a clear example of a hierarchy: it reports five broad domains, each with six specific facet scales. In their 1995 article, Costa and McCrae describe domain interpretation as a rapid overview and facet interpretation as a more detailed assessment. That supports a useful sequence for reading that instrument: see the broad pattern first, then inspect relevant details. It does not establish that every test uses the same structure or that its facets have equal quality.

A broad result is useful when the question is still open. If a report gives a higher score on a domain such as Extraversion, that alone does not tell you which parts of the domain produced the result or how you acted in a particular meeting. A narrower scale may point toward a more focused issue, but only if its description actually covers the behavior you are wondering about.

Think of the levels as map scale. A wide-area map helps you orient; a close view may show the turn you need to inspect. The close view is not automatically more accurate: it may contain more detail while being less stable, or it may capture a feature irrelevant to your question. The first decision is therefore not “Which score is highest?” but “What do I need to understand?”

Sources: Domains and facets: hierarchical personality assessment using the revised NEO personality inventory

When is a narrower score worth following?

Follow a subscale when its written definition closely matches the uncertainty you want to resolve and there is evidence that its score carries interpretable information. Use it as a targeted question, keeping the broad result as context. The fact that a score is narrower does not by itself make it a better explanation.

One reason not to dismiss facets comes from a specific study, not a universal rule. Danner and colleagues analyzed the 60-item Big Five Inventory–2 (BFI-2) with a statistical approach called a bifactor model. In plain terms, it estimates how questionnaire answers relate both to a broad trait and to narrower facets, while separating the overlap between them. Their paper describes a heterogeneous sample of 1,193 U.S. adults; the outcome analyses report N=1,116. Outcomes included education, income, self-rated health, and life satisfaction. For self-rated health, the directly observed correlations between ordinary questionnaire scale scores and health were positive for Energy Level (.40), Sociability (.14), and Assertiveness (.14). These are called manifest correlations: they come from the measured scale scores as printed, which can reflect both shared Extraversion and narrower differences. The model-based facet estimates were .20, −.18, and −.27, respectively. These are called latent estimates because the model treats broad and facet traits as underlying patterns and estimates the facet’s relationship after accounting for broad Extraversion. The two sets answer different questions and should not be read as competing versions of the same score. The study shows that conclusions can change when researchers separate overlapping broad and narrow information; it does not show that a printed Energy Level score is more accurate or predicts an individual reader’s health. Its other selected facet findings are also specific to this measure, sample, model, and cross-sectional self-reported outcomes.

The study also shows why the conclusion must stay conditional. The measured domain and facet scores combine broad-trait overlap, narrower information, and differences in score reliability. In the model, Compassion and Productiveness had near-zero, nonsignificant facet-specific estimates and were left out of later analyses. A subscale label alone cannot show how much distinct, interpretable information it carries; and results from the BFI-2 do not establish the quality of another instrument’s subscales.

For a practical illustration, suppose a broad score raises a question about how someone approaches collaboration. A report’s subscale definitions might distinguish speaking up from seeking frequent social contact. If the real uncertainty concerns whether ideas are voiced during planning, the former definition may give a more relevant prompt than the latter. This is an illustration of how to match a question to a definition, not a claim about any person’s score or behavior.

The next step is to compare the selected subscale with actual occasions. Ask what happened, what the setting required, and whether the same pattern appeared more than once. If the report’s definition does not fit the question, return to the broad result or set the report aside. A focused prompt should make observation easier, not force experience to fit a score.

Sources: Modelling the incremental value of personality facets: the domains-incremental facets-acquiescence bifactor showmodel

What evidence should you check before relying on a subscale?

Check three things: whether the subscale is sufficiently precise for the interpretation, whether the evidence applies to the people and use in question, and whether the report supports comparisons between its subscales. If those details are unavailable, treat a narrow result as a tentative prompt rather than a firm distinction.

Reliability is the consistency of scores under specified conditions; it is not the same as accuracy, validity, or proof that a particular interpretation is true. ETS’s 2014 Standards for Quality and Fairness state that reported scores, including subscores, should be reliable enough to support their intended interpretations. The document also says that the level required depends on the intended use and consequences of a wrong decision. These are general testing principles from ETS, not an evaluation of a personality report or a substitute for that instrument’s own evidence.

Precision evidence should match the score level. A report might give reliability information for its overall score without explaining how consistently each subscale is measured. That is not enough to assume every narrow result is equally steady. ETS guidance calls for reliability estimates appropriate to the aggregation level being reported and notes that reliability depends on the sources of variation considered and the group whose scores are being interpreted. Ask whether the technical documentation reports evidence for the particular subscale, not just the test as a whole.

Be especially cautious when comparing two subscales. A visible gap between two numbers may look like a meaningful personal contrast, but that reading needs evidence about how consistently such differences are measured. ETS explicitly calls for information on the consistency of score differences when users are expected to interpret them. Without it, avoid turning small numerical gaps into a story about competing strengths or weaknesses.

The BFI-2 study illustrates why instrument-specific documentation matters. Its facets had four items each, and its authors analyzed how those items reflected domain and facet variance within a particular model and sample. Those results cannot certify a different questionnaire’s subscales. A four-item facet in that study does not establish the quality of a four-item scale elsewhere; item wording, construction, population, and intended interpretation all matter.

If a report does not supply this information, look for a technical manual or ask the publisher or assessor what the subscale is intended to support, which population its evidence covers, and whether score differences can be interpreted. The absence of accessible evidence does not prove the score is useless. It does mean you should lower the confidence and keep the question modest.

Sources: Modelling the incremental value of personality facets: the domains-incremental facets-acquiescence bifactor showmodel; ETS Standards for Quality and Fairness (2014); Test Reliability—Basic Concepts

What can a report answer about one real situation?

A report can help you frame a question about a recurring tendency; it cannot, on its own, establish why one event happened or what you will do next. Use the broad score for orientation, a sufficiently supported subscale for a focused prompt, and observable examples to check the interpretation against context.

This conclusion gives both score levels a job. A broad result summarizes shared pattern and helps prevent over-reading one narrow result in isolation. A subscale can sharpen attention when its definition maps to the behavior at issue. Neither score substitutes for the situation itself: a missed deadline, a quiet meeting, or a tense exchange may reflect workload, role expectations, preparation, incentives, or other factors as well as a person’s usual tendencies.

A broad summary can hide distinctions that matter to a specific question. The NEO-PI-R account supports detailed facet interpretation, while Danner and colleagues’ BFI-2 analysis found selected associations beyond broad domains in a particular sample and model. That supports considering a well-documented facet as a prompt, not treating it as a cause or individual prediction.

For a decision about work or collaboration, write the question in observable terms: “In which meetings do I hold back a concern?” is more testable than “Am I a poor collaborator?” Then identify which report definition, if any, fits that question. Record an example that supports the interpretation and one that complicates it. A report’s tendency statement should remain open to revision when the setting or evidence points elsewhere.

Verdict: begin broad, move narrow when definition and score-quality evidence justify it, and finish with observation rather than a verdict about the person. This is a reflection method, not a hiring score, diagnosis, or job recommendation. If the remaining question is how several tendencies combine in your work decisions or collaboration, the live Work Pattern Report at /assessment offers a low-stakes self-report structure for reflection; it has no norms and is not validated for employment selection or career matching.

Use this short sequence: name the uncertainty; read the broad result; select a subscale only if its definition fits; check precision and intended use; then compare the prompt with a concrete event and its context. If you cannot find evidence for the subscale or a clear behavioral match, stay with the broad result as background and let direct observation guide the next question.

Sources: Domains and facets: hierarchical personality assessment using the revised NEO personality inventory; Modelling the incremental value of personality facets: the domains-incremental facets-acquiescence bifactor showmodel; ETS Standards for Quality and Fairness (2014)

Questions readers ask