Key takeaways
- Check the form version and the remote process before rating completeness.
- Treat an employer review without an authorized reviewer as unsupported.
- Report missing evidence separately from a substantive document mismatch.
Research question and decision use
The review asks whether remote I-9 files contain a complete employee section, an employer review performed under an allowed process, and a retained record that matches the payroll identity. The unit is one new hire file. The result supports training and file repair; it does not decide whether a particular document was genuine.
| Stage | Evidence | Failure mode | Use |
|---|---|---|---|
| Intake | Dated source and owner | Missing record | Route for review |
| Handoff | Named recipient and time | Unconfirmed receipt | Recheck |
| Closure | Completion evidence | Status only | Keep open |
Population and eligibility
Eligible files belong to employees who started during the window and whose work location required a remote process. Record the location, start date, form version, and review method. Exclude files handled entirely in person, but report the excluded count so the remote population is clear.
Sampling protocol
Select all files for a small cohort, or stratify by location, reviewer, and start month for a larger one. Include files with a known correction and files with no exception. Record the seed and any replacement, and do not drop a file because the review is known to be late.
Evidence dictionary
Fields include form version, employee section date, reviewer identity, review date, remote method, document type, document expiration, payroll legal name, and retention location. Missing means no record was found. Unsupported means a review is marked complete but the reviewer or method cannot be verified.
Classification rubric
Rate each file complete, partial, unsupported, or missing. Complete requires both sections, an authorized reviewer, an allowed method, and a retained record. Partial means one required element is absent. A name mismatch is a separate finding from a missing signature.
Reviewer calibration
Two reviewers rate a calibration set that includes a corrected name, a late review, and a file reviewed by an unauthorized coordinator. They compare the failed element, not only the overall rating. Initial ratings remain in the workbook after adjudication.
Analysis plan
Report counts and shares by status, reviewer group, location, and form version. Show the number of files with more than one defect. Do not publish a single pass rate without the defect mix, because one serious defect can matter more than several minor ones.
Sensitivity and exception analysis
Recalculate after excluding files with a documented correction completed before the review date. Compare results when an unsupported review is treated as missing rather than partial. If the ranking of reviewers changes, describe the process difference before drawing a conclusion.
Privacy and role boundaries
I-9 files contain identity and work authorization information. Review in a restricted environment, do not copy document images into the analysis file, and publish only aggregate findings. Apply the retention schedule to the working extract.
Limitations
The review checks whether the record shows a compliant process, not whether the employee was actually authorized to work. A missing document image may reflect a retention choice, and a reviewer note may omit a step that happened elsewhere. The method cannot establish legal sufficiency.
Replication package
Retain the sampled file list, form version key, reviewer roster, rating sheet, adjudications, and aggregate table. A later reviewer should rebuild the results without access to unnecessary personal data.
Conclusion
The practical output is a short repair list: missing employee sections, reviews by unauthorized staff, unsupported remote methods, and name mismatches. Each item gets an owner and a correction date, and the next cohort is checked against the same rules.
Sources and methodology
A bounded review of onboarding records with a frozen population, evidence rules, reviewer calibration, and explicit limitations.
- USCIS Form I-9 current.
- USCIS remote examination alternative procedure current.
- eCFR employment eligibility verification rules current.
- E-Verify current.
- DOL new hire reporting current.
- NIST SP 800-63B digital identity guidance current.
- NARA records management current.
- Plain Language Guidelines current.
- W3C WCAG 2.2 current.
- FTC data security guidance current.
Source count: 10. Last verification date: October 27, 2026.
Questions about the findings
What is the unit of analysis?
One onboarding record followed from its triggering event to a verified disposition.
Does this establish causation?
No. The design is descriptive and limited to the retained administrative record.
How is missing evidence handled?
It remains a visible finding rather than being inferred or excluded.
Put the finding to work
Review the full research library, compare cluster coverage inside Onboarding And Retention Research, and pair these findings with our VA candidate screening support.