Integrity review report: 'Class peer effects on adolescents' occupational expectations in county-level areas: The role of class structure and digital media' (DOI: 10.27017/j.cnki.qnyj.2026.02.009)

Academic fraud report · Geng Detector

Summary

Verdict: No evidence of academic misconduct; the paper is assessed as substantively clean. This quantitative social-science article uses a nested logit model with three occupational categories, which expands each respondent to three observations. Apparent inconsistencies between N (e.g., 2123, 2088) and reported observation counts (e.g., 6369, 6264) are fully explained by this expansion (2123×3=6369; 2088×3=6264), and subgroup N totals in Tables 4–5 reconcile cleanly to the full sample. Missing-data handling is conservative: 35 cases dropped when internet-use covariates are added, consistent with survey non-response rather than selective trimming. One unresolved issue concerns timing: the paper lists CASS '青启计划' grant 2026QQJH05, whose identifier contains '2026', yet the article appears in 2026 Issue 2 of 《青年研究》; this likely reflects retroactive attribution of a recently awarded grant to work completed earlier rather than data fabrication. No image-based checks were possible because the source PDF contained no extractable figures. Overall confidence is moderate-high, limited by the absence of raw data and original visual artifacts.

Verdict

No evidence of academic misconduct. The paper is data-coherent and methodologically careful. The only flagged item is a minor timing oddity regarding grant attribution, which is common practice and does not imply fraud.

Key findings

The N vs. observation-count discrepancy flagged by lay readers is the mathematical consequence of a 3-category nested logit model: 2123 × 3 = 6369 and 2088 × 3 = 6264 match exactly.
Subgroup sample sizes and observation totals in Tables 4 and 5 sum precisely to the full-sample totals, indicating no duplication or padded data.
The 35-case drop from Model 1.2 (N=2123) to Models 1.3/1.4 (N=2088) corresponds to the addition of individual- and peer-level internet-use variables, consistent with plausible item non-response.
The grant number 2026QQJH05 carries a '2026' label while the paper appears in 《青年研究》2026 Issue 2, suggesting rapid grant-to-publication turnaround; this is unusual but plausibly explained by pre-existing completed work being attributed to a newly awarded project.
No raw images (Western blots, microscopy, etc.) were available for visual duplication or splicing checks.

Evidence highlights

DOI: 10.27017/j.cnki.qnyj.2026.02.009
Table 3, Model 1.1: N=2123, observations=6369 (2123 × 3 = 6369 ✓)
Table 3, Models 1.3/1.4: N=2088, observations=6264 (2088 × 3 = 6264 ✓)
Tables 4–5: subgroup Ns and observation counts reconcile to full-sample totals.
Missing-value reduction: 2123 → 2088 (Δ = 35) upon adding internet-use covariates.
Funding: 中国社会科学院"青启计划" grant 2026QQJH05; publication venue 《青年研究》 2026年第2期.

Notes

Limitations: analysis is based on text-extracted PDF; raw microdata, code, and figure files were not provided, so independent replication of regression coefficients is not possible.
The 'limit-operation' timing of grant-to-publication cannot be definitively adjudicated without the grant's official award date; treat as unresolved but low-risk.
Confidence: moderate-high for data integrity; low for image-related checks (not performed).
Recommended actions: none required for fraud remediation; readers seeking to verify model output may request the authors' Stata do-files and dataset.