Prospective job-analysis study — preregistration candidate

Record ID: LL-FDE-JA-PREREG-001
Provisional study ID: FDE-JA-002
Protocol version: 0.1.0-draft
Status: draft registration candidate; not registered; recruitment and data collection are not authorized
Prepared: September 1, 2026
Registration service, URI, and DOI: not selected or assigned

1. Registration boundary

This document is a prospective registration candidate for the first new job-analysis study proposed under the LockedIn FDE Benchmark research program. It records decisions that must be made before recruitment so that later choices can be distinguished from the original plan.

It is not a completed preregistration, ethics approval, study launch, dataset, finding, validation study, or authorization to evaluate people. The provisional ID FDE-JA-002 is used to avoid conflating this planned study with retrospective study record FDE-JA-001, which concerns an unverified prior interview claim. The study register must formally assign the permanent identifier before registration.

2. Organizational responsibility

Role	Current appointee	Required disposition before registration
Organizational sponsor	LockedIn Labs Research	Confirm sponsor authority and role
Principal investigator	Unassigned	Name, qualifications, affiliation, conflicts, and accountability acceptance
Study operations owner	Unassigned	Own recruitment, consent, participant support, and protocol execution
Methods lead	Unassigned	Freeze design, sampling, instruments, and analysis
Data steward	Unassigned	Approve access, retention, deletion, and release controls
Ethics/privacy authority	Unassigned	Record applicable determination before intake
Accessibility reviewer	Unassigned	Review administration and alternate-demonstration paths
Independent methods reviewer	Unassigned	Review the frozen candidate and disclose independence/conflicts
Publication guarantor	Unassigned	Accept final study-publication accountability

An open role remains open; no person, qualification, approval, or independence is implied by repository access or contribution.

3. Intended use

The proposed study is intended to inform only:

the occupational boundary of Forward Deployed Engineering (FDE) work;
a versioned inventory of work activities, decisions, work products, contexts, and consequences;
revisions to the proposed ten-domain, 55-family ontology;
a content blueprint linking job evidence to observable behaviors and future assessment exercises; and
research questions for later content, response-process, reliability, fairness, and criterion studies.

The study is not designed to establish a pass mark, prevalence estimate, population norm, certification, score validity, hiring recommendation, job-performance prediction, or claim that the existing ontology is field validated.

4. Primary research questions

The confirmatory research questions proposed for registration are:

Which work activities and consequential decisions are performed in FDE contexts, and under what operating conditions?
Which activities distinguish FDE work from adjacent technical and consulting roles, and which are shared?
Which capabilities are judged important, consequential, difficult, frequent, or entry-critical within declared role contexts?
Which observable behaviors and work products could provide construct-relevant evidence of those capabilities?
Which proposed ontology families are unsupported, redundant, context-specific, inaccessible, or overdependent on a vendor or delivery model?
Which credible negative cases, refusals, reversals, incidents, or unsuccessful engagements change the proposed construct?

Exploratory questions, themes, and post-hoc analyses must be labeled exploratory in every result package.

5. Target population and sampling frame

The target population is not yet frozen. The registration candidate is expected to include qualified perspectives from:

practitioners doing hands-on forward-deployed work;
engineering, product, delivery, security, and business leaders who directly observe that work;
customer or partner stakeholders who can describe accepted outputs and consequences; and
practitioners in adjacent roles used for boundary comparison.

Planned strata include role context, seniority/responsibility, organization stage, delivery model, technical emphasis, regulated versus non-regulated setting, geography/language where relevant, and successful as well as failed or stopped work.

The sampling frame, recruitment channels, allocation by stratum, outreach volume, incentive policy, nonresponse treatment, replacement rule, and stopping rationale are unresolved. The sample size is also unresolved and must be frozen with a design-based rationale before registration. No target count is asserted by this document.

A large convenience sample must not be described as representative. The publication must disclose frame coverage, selection channels, exclusions, missingness, nonresponse, observed strata, and limits on transfer.

6. Eligibility and exclusion

Eligibility rules must be fixed before recruitment and identity review. At minimum, a participant record must establish:

perspective and relationship to the work;
a plain-language account of relevant responsibilities;
observation window and recency;
organization/context categories needed for analysis;
recruitment channel, incentive, and conflicts;
consent and permitted-use status; and
sufficient verified context to interpret the contribution without publishing private identity.

A job title, employer, public reputation, LockedIn relationship, or training completion is insufficient on its own. Exclusion rules must cover duplicate participation, failed eligibility, unavailable consent, unverifiable context, unusable records, protocol violations, and conflicts that prevent the planned use. Every exclusion class and adjudication rule must be frozen before analysis.

7. Prospective evidence collection

7.1 Critical-incident interviews

Use a versioned structured guide to elicit specific episodes, including initial condition, missing or contested information, decisions, rejected alternatives, hands-on work, dependencies, artifacts, failures or change events, measurable outcomes, attribution limits, and evidence that would distinguish effective from ineffective behavior.

Positive, negative, incomplete, refused, reversed, failed, and ambiguous incidents are eligible when they meet the frozen criteria.

7.2 Work-product inventory

Collect metadata about classes of work products without requiring confidential contents. For each class, record purpose, creator/reviewer, point in the work, quality indicators, common failure modes, confidentiality barriers, accessibility barriers, and construct-preserving alternate demonstrations.

7.3 Structured ratings

After qualitative coding stabilizes, eligible respondents may rate frozen activity or capability statements on predeclared anchored scales such as importance, consequence of error, difficulty, frequency/exposure, entry criticality versus learnability, context applicability, and assessment feasibility/fairness.

Scale wording, anchors, order, missing/not-applicable handling, aggregation, uncertainty, and action rules remain unresolved and must be frozen before collection.

7.4 Adjacent-role comparison

Apply the same work statements and declared context variables to qualified observers of adjacent roles. The analysis will examine shared activities and discriminating conditions; it will not classify people solely from titles.

8. Measures and outcomes

Proposed primary outputs are:

coded job activities, decisions, work products, operating contexts, and consequences;
distributions of predeclared ratings by declared context, with uncertainty and missingness;
traceability from admitted records to retained, revised, context-limited, or rejected ontology families;
agreement and disagreement among qualified coders and reviewers;
negative-case and contradictory-evidence findings;
an inclusion/exclusion flow and observed sampling coverage; and
a proposed content blueprint with evidence and dissent references.

No single rating average automatically creates a universal requirement or weight.

9. Analysis plan to freeze before registration

The methods lead must freeze:

unit-of-analysis and unitization rules;
codebook development, coder training, blinding, and conflict handling;
the independently double-coded sampling rule;
agreement statistic, uncertainty interval, and action threshold appropriate to the data;
adjudication and substantive-dissent retention;
qualitative stopping or information-power rationale;
missing-data, not-applicable, outlier, and sensitivity analyses;
rating summaries, uncertainty, multiplicity treatment, and context comparisons;
ontology and blueprint decision rules;
confirmatory versus exploratory outputs; and
software, environment, seed, and artifact-digest requirements.

The chosen statistics and thresholds are intentionally not invented in this draft. They require a qualified methods decision linked to the final instruments, scale properties, sampling design, and intended interpretation.

10. Privacy, ethics, security, and accessibility

Before recruitment or data intake, the release package must contain:

an ethics or human-participant determination from the accountable authority;
lawful basis, versioned notice/consent, withdrawal and correction routes;
minimum-necessary collection, pseudonymization, access roles, retention, deletion, incident response, and controlled transfer;
treatment of recordings, transcripts, contact data, employer/customer information, confidential artifacts, and cross-border transfer;
participant accessibility review, supported formats, accommodations, and language approach;
protection against employer, sponsor, or platform retaliation; and
a publication/disclosure plan that prevents re-identification and unsupported subgroup inference.

Real participant records, identity evidence, confidential customer artifacts, and contact information must not be committed to the public repository.

11. Amendments and deviations

The registered record must be immutable. Any later change receives an append-only amendment with date, authority, rationale, affected outcomes, knowledge of the data at the time, and impact on interpretation. Deviations must be reported even when they do not change results.

Stopping, analysis, exclusion, or reporting changes made after outcome inspection must never be presented as preregistered. A superseding registration must link back to this record without deleting it.

12. Publication and availability plan

A study publication is expected to include the frozen protocol and amendments, recruitment and sample flow, instruments, codebook, analysis plan, deidentified aggregate outputs, traceability matrix, analysis code, environment lock, source and result manifests, limitations, conflicts, funding, AI-use disclosure, reviewer records, corrections route, and a data/materials availability statement.

Participant-level data will be released only if consent, privacy, security, re-identification risk, contractual rights, and governance allow it. Otherwise the release will describe the controlled-access procedure or explain why access is unavailable. “Available on request” is insufficient without an owner and process.

Registration, archive, canonical URL, identifier, and embargo decisions are pending.

13. Activation checklist

Permanent study ID entered in the study register.
Principal investigator, methods lead, study owner, data steward, reviewers, and guarantor appointed.
Complete protocol, population, frame, sample-size rationale, recruitment, and stopping rules frozen.
Instruments, scale anchors, codebook process, analysis plan, and decision rules frozen.
Funding, incentives, sponsor role, conflicts, recusals, and AI-use plan disclosed.
Ethics/privacy determination, consent, security, retention, incident, and participant-support controls approved.
Accessibility and language review completed.
Independent methods review completed and disagreements retained.
Registration service selected and immutable registration receipt recorded.
Recruitment authorization signed for the exact registered version.

Current disposition: all activation gates remain open; no recruitment or data collection is authorized.