Q005 — Self-Audit¶


Research	R0029 — Plural Voice Attribution
Run	2026-03-27
Query	Q005

ROBIS 4-Domain Audit¶

Domain 1: Eligibility Criteria¶

Rating: Low risk

Criterion	Assessment
Defined what counts as "documented cases"	Yes — formal cases, survey data, and self-report studies all accepted
"Quantitative data" threshold applied	Yes — opinion pieces and anecdotes excluded

Notes: Clear eligibility maintained throughout.

Domain 2: Search Comprehensiveness¶

Rating: Some concerns

Criterion	Assessment
Multiple search strategies	Yes — academic and workplace searches conducted separately
All three contexts searched	Partial — academic and workplace well-covered; software engineering searches returned adoption data but not misrepresentation data
All results dispositioned	Yes
Source diversity	Partial — workplace relies heavily on one study (KPMG)

Notes: The software engineering gap is a genuine finding, not a search failure. Four targeted queries were attempted. Workplace context relies primarily on one (very large) study.

Domain 3: Evaluation Consistency¶

Rating: Low risk

Criterion	Assessment
All sources scored	Yes — 4 scorecards
ACH matrix applied	Yes
Diagnosticity analysis	Yes

Notes: Consistent application.

Domain 4: Synthesis Fairness¶

Rating: Low risk

Criterion	Assessment
All hypotheses tested	Yes
Counterpoint evidence included	Yes — Stanford stable cheating rates
Software engineering gap honestly reported	Yes — reported as a gap, not glossed over
Framing bias surfaced	Yes — noted that "misrepresentation" framing may not apply to SE

Notes: The Stanford counterpoint was included despite weakening the "epidemic" narrative.

Overall Assessment¶

Overall risk of bias: Low risk (with one concern)

The main concern is workplace evidence concentration — the 57% figure is powerful but comes from a single study. No independent workplace survey was found to corroborate it. However, the study's size (48K+) and methodology partially compensate for this.

Researcher Bias Check¶

Framing bias (acknowledged): The query's "submitting as their own work" phrasing embeds a judgment of dishonesty. In software engineering, AI code assistance may be expected. This was surfaced and addressed.
Anchoring bias (low risk): The 57% headline figure is attention-grabbing and could anchor perceptions. The self-report nature and potential measurement issues were noted.