Validation gap-assessment.

1. Intended-use lock (this is the control)

Use	Sold?	Validation bar
Career exploration / counseling	Permitted	Content-related design + purpose label. Not a UGESP “selection procedure” if no hire/refer/reject flows from the score.
Internal mobility / development (human review)	Permitted	Decision support. Human decides. Do not auto-route on composite.
Workforce planning / occupational-family language	Permitted	Maps are illustrative, not authorized staffing tables.
Screening, ranking, cutoffs, referral, grade, clearance, hire	Prohibited as sold	Would require current job analysis, local criterion or content validation, adverse-impact monitoring, and the agency’s own UGESP file. ALA does not sell that SKU.

Kickoff labels the sitting: exploration / mobility / development — not a qualification determination. Participants see that on the Federal Fit start screen. OPM: job analysis. EEOC: Uniform Guidelines Q&A.

2. What the product actually measures

• 50 timed work-relevant items. Six domains: executive reasoning, quantitative & spatial, verbal & working memory, visual pattern recognition, workplace social judgment, procedural sequencing.
• Scoring lock: +2.0 accuracy if correct; +0.2 speed if correct and RT ≤ 40 s; 60 s window; max composite 110. Career Fit freeze 2026.08-C. Hiring Fit / Federal Fit remain on their published freeze until rebuilt — methodology.
• Not an IQ classification. Not a medical or psychological exam under the ADA. Not USAJOBS. Not OPM-validated. Not a cutoff.
• Output: six-domain profile + occupational-family map + HTML/CSV the agency keeps. Sample: RWC-RPT-086.

3. Uniform Guidelines (29 CFR 1607) — gap table

UGESP topic	Status for exploration SKU	Status if used for selection
§1607.2 Scope — “selection procedure”	Out of scope as sold if the agency does not use scores to hire, promote, refer, or reject	In scope the moment scores rank or cut
§1607.3 Adverse impact / 4/5ths	Not computed — no federal selection use, no race/sex/national-origin fields collected	Missing. Agency would have to collect and analyze its own file
§1607.5 / .14 Technical standards (criterion, content, construct)	Content-related design (work tasks, O*NET-aligned families). Not a completed content-validity study for a named series	Missing for any series-specific hire tool
§1607.6 Alternative procedures	N/A while not a selection procedure	Would be the agency’s analysis
§1607.7 / .8 Documentation	This gap file, methodology, freeze, sample export, SOW, DPA	Agency job-analysis + local validation file still required
§1607.15 Retention of records	Agency owns the download. ALA retains scores 36 months unless deletion is requested (DPA)	Agency records schedule still governs OPF / hiring files

4. SIOP Principles (2018) and 5 CFR §300.103

Requirement	Have	Gap
Define the work / job analysis	Occupational-family mapping; work-relevant item tasks; public “why this item” lines	No series-specific KSA job analysis for GS-XXXX. That is the agency’s if they ever want selection use
Reliability	Fixed freeze, server key, 50 items, documented formula	No published Cronbach’s α / test-retest for Federal Fit as a federal instrument
Criterion-related validity	Literature class: Schmidt & Hunter 1998; Sackett et al. 2022 — cognitive/work-sample methods predict performance. Research evidence	r ≈ .51 is not an RWC coefficient. No Federal Fit vs. federal performance study
Fairness / DIF	No protected-class data stored. ID-only	No DIF tables. Cannot be produced without the agency collecting that data — which this SKU does not
5 CFR §300.103 rational relationship	Purpose label + human decision-maker for permitted uses	Not a professionally developed selection practice for competitive examining

5. Evidence we do have (do not over-read it)

Evidence	What it is	What it is not
58 of 60 still employed at 9 months	One private employer. Actual headcount. Observational. Case write-up	Not federal past performance. Not CPARS. Not a randomized trial. Not proof the next cohort hits that number. Not UGESP validity
Literature (Schmidt & Hunter; Sackett 2022; Kristof-Brown 2005)	Context for the class of methods	Not “our test is 51% accurate.” Validity is not a percentage
Item rationales / freeze / scoring lock	Construction documentation	Not a technical manual sufficient for OPM examining
IBM Quantum construction record	Form-assembly entropy. Self-disclosed. Record	Not IBM endorsement. Not a bias-free certificate

6. What would be required to lift the selection prohibition

1. Agency job analysis for the specific series/grade (OPM method).
2. Local content or criterion study on this instrument, this job, this population — not a literature citation.
3. Adverse-impact analysis if the procedure is used to select, with the agency holding the demographic file (ALA does not collect it).
4. Written change of intended use. ALA would have to agree to sell a different SKU. Federal Fit as currently sold will not be that SKU.

Until those exist, any agency that ranks, cuts, or auto-refers on Federal Fit scores is using the product off-label. That is the agency’s legal risk, not a hidden ALA feature.

7. Recommendation for a $14,500 Scale buy

Accept for exploration / mobility / development if the contracting officer and CHCO lock the use case in the SOW, participants are told it is not a qualification determination, humans remain the decision-makers, and scores are not written into a competitive examining certificate.

Do not accept as a selection instrument. Do not replace USA Hire, OPM examining, or a validated assessment battery with this file.

Companion files: SOW · packet Q4 · research evidence · DPA.

Not legal advice. Not an EEOC or OPM determination. Advanced Learning Academy LLC · 29 August 2026.