Strong Interest Inventory

Status: Documented, not shipped · Evidence: M · Family: Self and team awareness · Verdict: reject (2026-06-11)

Strong Interest Inventory (registered trademark of The Myers-Briggs Company, formerly CPP / Consulting Psychologists Press). E. K. Strong Jr., 1927 (Strong Vocational Interest Blank); David Campbell, 1974 (Strong-Campbell Interest Inventory); 2004/2005 revision by Donnay, Morris, Schaubhut and Thompson.

What it is

The Strong Interest Inventory (SII) is a trademarked vocational-interest assessment. A test-taker rates their preference across roughly 290 items (occupations, school subjects, activities, leisure pursuits, types of people), and the publisher’s scoring engine returns a profile on four scale families: the six General Occupational Themes (GOTs, which are Holland’s RIASEC types - Realistic, Investigative, Artistic, Social, Enterprising, Conventional); the 30 Basic Interest Scales (clusters within the themes, such as “Mechanics” or “Writing”); the roughly 260 Occupational Scales (OSs); and five Personal Style Scales. Its historically distinctive move is the Occupational Scale: each OS is criterion-keyed, built by contrasting how people in a specific occupation answered against a general reference sample, so a high OS score means “your interest pattern resembles that of people who are content in this occupation,” not “you are interested in this work.” The deliverable is a multi-page printed profile that ranks the test-taker’s resemblance to occupational reference groups.

Stripped of the proprietary instrument, the norm tables, and the criterion-keyed occupational scales, the only durable cognitive move left is the same one Holland’s RIASEC carries: score a person, role, or option against a fixed set of pre-named interest dimensions and read where the profile matches and where it clashes. Administering the inventory, contrasting a respondent against reference samples, and computing congruence indices are instrument administration and psychometric scoring - tasks a publisher’s scoring service performs, not reasoning moves an agent executes. There is no agent-runnable Strong-specific operation that survives removing the instrument.

When it helps / when it misleads

Where it has real standing, it is as a vocational-counseling instrument administered by a trained professional to a self-reporting individual who wants structured input on career and major exploration. In that narrow setting, the criterion-keyed Occupational Scales add something the bare RIASEC typology does not: a direct resemblance read against people already in named occupations, with genuine long-range predictive signal (below).

It misleads, and these are the walls. It is an instrument, not a thinking move: nothing about an agent reasoning over a person or a problem is what the SII evidence tests, so a Strong “result” produced without administering the validated, normed instrument inherits the brand and none of the measurement validity. Interest is not aptitude: a Strong profile says nothing about whether a person can do the work, only about resemblance to the interests of incumbents, so it must never function as an ability screen or a hiring gate. Its items carry substantial gender bias: about 70% of the interest items show gender-related differential item functioning (Einarsdottir and Rounds 2009), so the profile can quietly funnel men and women toward sex-typed occupations, and using it to steer a real person’s choices imports that bias. Expressed beats measured: a person’s directly stated career intention predicts their eventual occupation at least as well as the inventory does (the expressed-versus-measured tradition, below), so for many practical purposes simply asking “what do you want to do?” is as informative and far cheaper. And forcing a non-vocational decision (which product, which strategy, which vendor) onto the six interest themes imports six irrelevant criteria and omits the ones that matter; that is decision-option-review, not an interest question. Typing a named colleague against occupational reference groups without their participation is a misuse of the instrument, not a thinking move.

What the evidence says

Honest grade: M for the instrument as a measure of vocational interests, with the M concentrated in the criterion-keyed Occupational Scales; the shippable cognitive residue grades only P, transferred. The registry’s preliminary M is the instrument-validity grade and is honest at that level. It is NOT evidence that an agent reasoning about interests decides better; reading the M as a license to ship a “Strong” thinking move would launder instrument-validity evidence onto a move the literature never tests. This is the family’s standing rule, the same one applied to the RIASEC fold.

What the research supports. The Occupational Scales have genuine long-range predictive validity: Hansen and Dik (2005, Journal of Vocational Behavior 67, 365-378) tested whether OS scores predicted the occupation a person actually held 12 years after initial testing, 8 years after a second testing, and concurrently, and found hit rates substantially above chance at all three points, with no significant gender difference and equal prediction for occupations directly and indirectly represented on the profile. This is the strongest independent result in the whole self-and-team-awareness instrument family and is why the SII grades M rather than the X of MBTI or the V of the vendor-only strengths instruments. Reliability is high, though the headline figures come from the publisher’s manual (Donnay, Morris, Schaubhut and Thompson 2005, the revised SII manual, Consulting Psychologists Press): internal-consistency reliabilities reported around .90 to .95 for the GOTs, .80 to .92 for the Basic Interest Scales, and high adult test-retest stability, with lower stability for high-school and college samples (consistent with interests still forming). The GOT layer rests on Holland’s RIASEC structure, which replicates as a circular ordering of the six types (Hansen 1979 on the fit of the Strong-Campbell GOTs to Holland’s hexagon; Rounds and Tracey 1993, Psychological Bulletin, structural meta-analysis), though the strict equilateral-hexagon form is not supported and the structure fits U.S. samples better than samples elsewhere (Rounds and Tracey 1996).

What the research does NOT support, and the transfer. Every supporting study measures a property of the instrument administered to humans: whether a validated, normed profile predicts a person’s later occupation or distinguishes satisfied from dissatisfied incumbents. None tests whether an agent that reasons about interests across six types recommends better, and none tests an instrument-free “Strong” read assigned from a resume or conversation. Two independent lines temper even the instrument grade. First, measured interests do not beat expressed interests: the expressed-versus-measured tradition (Bartling and Hood 1981, “Measured interests versus expressed interests as predictors of long-term occupational membership,” Journal of Vocational Behavior, building on Dolliver’s 1969 review) finds a person’s stated intention predicts eventual occupation at least as well as the inventory, which caps the marginal value of the instrument over simply asking. Second, the items are gender-biased: Einarsdottir and Rounds (2009, Journal of Vocational Behavior 74(3), 295-307) found differential item functioning in about 70% of SII interest items across a sample of 1,860 women and 1,105 men, a sex-type dimension shifting men’s and women’s responses differently. So the honest read is: a moderately-validated vocational instrument whose distinctive predictive signal is real, whose marginal value over an expressed preference is contested, and whose items carry documented gender bias. Every numeric claim above maps to a named author and year; the publisher’s “standard of excellence” marketing language and any “X% of people find their career” style figures trace to no independent primary source and are excluded.

Why it is / is not a skill here

Verdict: Reject (status: flag - documented with the trademark and instrument-versus-move caveat, not shipped). This confirms the registry’s preliminary reject and keeps the preliminary tier M, with the tier explicitly scoped to the instrument and the shippable residue capped at P.

The preliminary entry said the SII “likely folds rather than shipping under the brand” toward Holland’s RIASEC. That read is correct about the mechanism, and the staged riasec-interest-fit dossier completed the chain: RIASEC itself folds into shipped decision-option-review (score an entity against a fixed criterion set and read the fit, which is exactly the option-against-criteria engine, with the six Holland types as a content preset). So the durable move under the Strong brand resolves to: Strong Interest Inventory reduces to RIASEC, and RIASEC folds to decision-option-review. There is no Strong-specific cognitive operation left over once the proprietary instrument and its norm tables are removed - the criterion-keyed Occupational Scale is a measurement procedure (contrast a respondent against employed reference groups), not an agent move.

Why flag rather than a registry fold, and why not ship. A registry foldInto must resolve to a status: shipped entry, and the natural target (riasec-interest-fit) is a candidate, not shipped; so a fold edge cannot be drawn there without misrepresenting the chain. More importantly, this is a famous, branded instrument that the library should document honestly rather than silently subsume: it is the six-thinking-hats and cynefin pattern, a trademarked method with a real but bounded signal, included only with its caveats. It is not excl on the merits the way an under-evidenced or purely redundant entry is, because its Occupational-Scale evidence is genuinely the strongest in this family; flag records that real signal while stating plainly that (a) it is an instrument, not a thinking move, (b) its generic, agent-runnable core is RIASEC, which already folds to decision-option-review, and (c) shipping a “Strong” skill would require controlled evidence that the move - not the instrument - improves an agent’s reasoning, which does not exist.

It is therefore not a Build: no separable artifact is uniquely Strong and not a fixed-criteria fit matrix already covered by decision-option-review. It is not a Recipe: there is one preset-criteria move, not a chain. It leans out-of-scope by domain as well - vocational and career guidance is an applied specialty nearer the sibling pm-skills library than a general thinking move. The learning value of the decision: a famous, decently-validated career instrument can still be the wrong thing to ship here, because instrument validity is not move effectiveness, and the only part an agent can actually run is a content preset of a shipped option-evaluation skill.

Lineage and who to read

Origin: Edward K. Strong Jr. created the Strong Vocational Interest Blank (SVIB) in 1927, pioneering the criterion-keyed Occupational Scale (contrast a target group’s responses against a general sample). David Campbell revised and merged the men’s and women’s forms into the Strong-Campbell Interest Inventory (1974), which incorporated Holland’s RIASEC as the General Occupational Themes. The instrument has been revised repeatedly (1985, 1994, and the 2004/2005 revision by Donnay, Morris, Schaubhut and Thompson). It is a registered trademark; the current publisher is The Myers-Briggs Company (formerly CPP / Consulting Psychologists Press). The generic typology under the GOTs is John L. Holland’s, and is not owned by the Strong publisher.

To read the evidence honestly: Hansen and Dik (2005) for the Occupational Scales’ long-range predictive validity, the instrument’s best independent result; Donnay, Morris, Schaubhut and Thompson (2005), the revised SII manual, for the publisher’s reliability and norming data (read as vendor-sourced); Einarsdottir and Rounds (2009) for the gender differential-item-functioning critique; Bartling and Hood (1981) and Dolliver (1969) for the expressed-versus-measured tradition that caps the inventory’s marginal value over simply asking; and Hansen (1979) and Rounds and Tracey (1993, 1996) for the RIASEC structure under the GOTs and its limits. For the generic, agent-runnable core and why it folds, see the riasec-interest-fit dossier in this library.

Named sources

• Jo-Ida C. Hansen and Bryan J. Dik, “Evidence of 12-year predictive and concurrent validity for SII Occupational Scale scores,” Journal of Vocational Behavior 67 (2005): 365-378. Occupational Scale scores predicted held occupation at 12-year, 8-year, and concurrent points substantially above chance. The instrument’s strongest independent validity result. (M)
• David A. C. Donnay, Michael L. Morris, Nancy A. Schaubhut and Richard C. Thompson, Strong Interest Inventory Manual: Research, Development, and Strategies for Interpretation (Mountain View, CA: Consulting Psychologists Press, 2005). Reliability and norming for the 2004 revision (GOT internal consistency about .90 to .95; high adult test-retest). Publisher manual; read as vendor-sourced. (V for the reliability figures specifically)
• Sif Einarsdottir and James Rounds, “Gender bias and construct validity in vocational interest measurement: Differential item functioning in the Strong Interest Inventory,” Journal of Vocational Behavior 74(3) (2009): 295-307. About 70% of interest items showed gender-related DIF; a sex-type dimension shifted men’s and women’s responses. Independent construct-validity critique. (Independent, S-quality method)
• Carl W. Bartling and Albert B. Hood, “Measured interests versus expressed interests as predictors of long-term occupational membership,” Journal of Vocational Behavior (1982), in the Dolliver (1969, Psychological Bulletin) expressed-versus-measured tradition. Stated career intentions predict eventual occupation at least as well as the inventory. The cap on the instrument’s marginal value. (Independent)
• Jo-Ida C. Hansen, “The fit between Strong-Campbell Interest Inventory General Occupational Themes and Holland’s hexagonal model,” Journal of Vocational Behavior (1979); and James Rounds and Terence J. Tracey, “Prediger’s dimensional representation of Holland’s RIASEC circumplex” / structural meta-analyses (Psychological Bulletin 1993; cross-cultural 1996). The RIASEC structure under the GOTs replicates as a circular order; the strict equilateral hexagon does not, and it fits U.S. samples better. (Structural, S-quality method)