PVQ-TM Methodology

• Transparency: All formulas and scoring thresholds are published. No proprietary black-box algorithms.
• Reproducibility: Given the same inputs and data versions, any implementation must produce identical results.
• SSA Compatibility: Aligns with Social Security Administration regulations governing transferability of skills (20 CFR 404.1568, SSR 82-41).
• Multi-Source Data: Integrates DOT, O*NET, BLS ORS, OEWS, and Employment Projections data with explicit source tracking.
• Confidence Grading: Every result carries a data-quality grade (A-D) reflecting the completeness and provenance of underlying data.

• 20 CFR 404.1568(d) — Transferability of Skills: Skills are transferable when they can be used to meet the requirements of other skilled or semi-skilled work. The degree of transferability depends on the similarity of occupationally significant work activities.
• SSR 82-41 — Work Skills and Their Transferability: Defines skills as knowledge that gives a worker the ability to perform functions of a job acquired through performance of past relevant work (PRW). Skills must involve more than raw ability—they require learned judgment, techniques, or methods with a Specific Vocational Preparation (SVP) of 4 or higher.
• SSR 83-10 through SSR 83-14: Physical exertion requirements and the interplay between strength levels and skill transferability at various age categories.
• Advanced Age Rules: For workers of advanced age (55+) or closely approaching advanced age (50-54), SSA regulations require that transferable skills allow “very little, if any, vocational adjustment” in terms of tools, work processes, work settings, and industry.

The DOT contains 12,726 occupation definitions scraped from occupationalinfo.org, each with:

• DOT code (9-digit occupational classification)
• Title and industry designation
• GED levels: Reasoning (R), Math (M), Language (L) on a 1-6 scale
• SVP (Specific Vocational Preparation) level 1-9
• Strength requirement (S/L/M/H/V)
• DPT worker functions (Data/People/Things complexity levels 0-8)
• GOE (Guide for Occupational Exploration) code
• DLU (Date of Last Update)
• Occupational description

Note: The DOT was last updated in 1991. While the occupational definitions remain the legal standard for SSA transferability determinations, PVQ-TM supplements DOT data with current O*NET data where available.

O*NET provides current occupational data via the O*NET Web Services API, including:

• Tasks and Detailed Work Activities (DWAs)
• Tools and technology requirements
• Knowledge, skills, and abilities with importance scores
• Work context and generalized work activities
• Job zones and SVP ranges
• Related occupations and career changers data

ORS provides statistically derived physical demand, environmental condition, and cognitive requirement data with standard errors. When available, ORS takes priority over DOT for trait demand estimation due to its statistical rigor and recency.

OEWS provides employment counts, wage percentiles (10th, 25th, median, 75th, 90th), and mean wages by occupation and geographic area. Used for labor market scoring.

RHAJ reference definitions provide the canonical descriptions for DPT worker functions, GED levels, SVP training times, GATB aptitudes, temperaments, physical demands, and environmental conditions. These definitions anchor the normalization functions.

Each data source uses different native scales. PVQ-TM applies the following normalization functions to map all sources to the common 0-4 scale:

When multiple data sources provide values for the same trait, PVQ-TM uses the following priority order: ORS > DOT > O*NET. ORS takes priority because it provides statistically measured demand data with standard errors. Each trait in every analysis carries a source tag indicating its provenance.

Each case maintains up to four profile rows:

• Work History Profile: Trait levels documented from past employment records
• Evaluative Profile: Trait levels from clinical evaluation (e.g., FCE)
• Pre-Injury Profile: Baseline trait levels before the date of injury
• Post-Injury Profile: Current residual functional capacity — this is the binding constraint used in all computations

• Task/DWA Overlap (35%): Jaccard similarity between the worker's acquired skill statements and the target occupation's O*NET tasks and detailed work activities (DWAs). Also includes DPT (Data-People-Things) worker function descriptors from DOT where available.
• Work Field/MPSMS Overlap (25%): Jaccard similarity between the source and target DOT work field codes and Materials, Products, Subject Matter, and Services (MPSMS) codes.
• Tools/Technology Overlap (20%): Token-level comparison between the worker's documented tools/software and the target occupation's O*NET tools and technology list.
• Materials/Services Overlap (10%): Token-level comparison of materials, products, and services between source and target.
• Credential/Knowledge Overlap (10%): Comparison of knowledge domains between the worker's background and the target occupation's O*NET knowledge requirements.

Before STQ is computed, a hard SVP gate is applied: the target occupation's SVP must be equal to or lower than the highest SVP among the worker's past relevant work entries. If the gate fails, the occupation is excluded with STQ = 0.

When a worker has multiple past relevant work entries, PVQ-TM computes STQ against each PRW entry independently and uses the highest-scoring match. This recognizes that different PRW entries may provide different transferable skills.

For each of the 24 traits, the worker's post-injury capacity is compared against the occupation's demand level. If the demand exceeds capacity on any single trait, the occupation is excluded entirely. There are no partial credits or trade-offs between traits.

Among occupations that survive the hard exclusion gate, TFQ is computed from the reserve margin—the average surplus capacity across all rated traits:

PVQ-TM maps the following DOT fields to the 24-trait demand vector:

The remaining 20 traits are sourced from ORS when available, or marked as “proxy” (null) when no authoritative data exists. Null traits do not contribute to feasibility exclusion—only traits with measured demands can cause exclusion.

When the evaluator has not provided manual ratings, PVQ-TM auto-estimates each dimension from DOT and O*NET data:

• Tools: O*NET tools/technology overlap between source and target. >75% overlap = 100, >50% = 67, >25% = 33, ≤25% = 0.
• Work Processes: GOE code comparison. Same GOE group (first 4 chars) = 100, same division (first 2 chars) = 67, different = 33.
• Work Setting: Industry designation comparison. Exact match = 100, shared significant words = 67, no overlap = 33.
• Industry: Broader sector comparison. Same primary sector = 100, any word overlap = 67, completely different = 33.

Auto-estimated ratings are clearly marked in the output. The evaluator should review and may override any auto-estimated value with a manual assessment based on their professional judgment.

For workers of advanced age (55+) or closely approaching advanced age (50-54), SSA regulations require that transferable skills require “very little, if any, vocational adjustment.” In PVQ-TM, this means all four dimensions must score 100. Any dimension below 100 results in exclusion of the target occupation.

Compares the target occupation's median wage against the worker's prior earnings using a wage ratio:

If no prior earnings are available, the score is based on absolute wage levels: >$60K = 80, >$40K = 60, >$25K = 40, otherwise 20.

• STQ at 45%: Skill overlap is the primary determinant of transferability per SSA policy.
• TFQ at 25%: Physical/cognitive feasibility is the second most important factor—no transfer is possible if the worker cannot perform the job.
• VAQ at 15%: Vocational adjustment reflects the practical difficulty of transitioning.
• LMQ at 15%: Labor market viability ensures the occupation represents a real employment opportunity.

Three sequential exclusion gates are evaluated before computing the PVQ composite. If any gate fails, the occupation is excluded with PVQ = 0:

1. Gate 1 — STQ/SVP: Target SVP must not exceed source SVP.
2. Gate 2 — TFQ: Worker must meet or exceed all trait demands.
3. Gate 3 — VAQ: For advanced-age cases, all adjustment dimensions must score 100.

Each PVQ result carries a confidence grade (A through D) reflecting the completeness of the underlying data:

Every analysis records the versions of data sources used at the time of computation: O*NET version, ORS release, OEWS survey year, and the DOT data extraction date. This ensures that even as data sources are updated, prior analyses can be understood in the context of their original data.

Every trait comparison in TFQ includes a source tag (ORS, DOT, ONET, or proxy) indicating which data source provided the demand value. This allows reviewers to assess the provenance of each data point.

STQ results include the specific matched items for each component: matched tasks, matched DWAs, matched tools, matched materials, and matched knowledge domains. This allows line-by-line verification of the overlap computation.

VAQ results clearly distinguish between evaluator-provided manual ratings and data-driven auto-estimates. Auto-estimated values include the underlying data used for estimation (GOE codes, industry designations, tool overlap percentages).

To replicate a PVQ-TM analysis:

1. Obtain the same data versions recorded in the analysis metadata
2. Enter the identical worker profile (24 traits, post-injury)
3. Enter the identical PRW entries with DOT codes and acquired skills
4. Run candidate generation with the same parameters
5. Apply the formulas documented in Sections 6-10 above
6. Results must match to within rounding tolerance (0.01)

The PVQ-TM computation engine is implemented in TypeScript with all source code available for inspection. The normalization functions, similarity algorithms, scoring thresholds, and composite formulas are fully specified in the codebase and correspond exactly to the documentation in this article.