Academic fraud detection report — Misapplication of statistics in scorpiology: a case study (Victoria Tang, 2026)

Academic fraud report · Geng Detector

Summary

This report evaluates the paper titled 'Misapplication of statistics in scorpiology: a case study' by Victoria Tang, published in ELYSIUM — Journal of Informal Scorpiology (No. 3, 2026), DOI 10.5281/zenodo.20371057. Tang's paper is itself a forensic re-examination of an earlier work by Ait Hammou et al. (2026). Our review is therefore directed at the fidelity of Tang's statistical reproduction and the credibility of her accusations against the target paper. Verdict: Tang's paper is judged clean of any detected misconduct and represents a high-quality, mathematically rigorous critique. Key issues identified in the target work include severe morphological data-entry errors (e.g., a recorded carapace length of 6.97 mm against congeneric values of 26.92–40.44 mm, and a telson width of 9.58 mm vs. 2.57–4.94 mm), 23 unadjusted multiple-comparison tests producing a family-wise error rate of approximately 69% with corrected BH q-values rising to ~0.257, a rank-deficient LDA with 23 variables against N=12 yielding a spurious 100% training accuracy, internal legend/figure contradictions, and probable AI-generated prose with formula inconsistencies. Confidence is high due to the explicit numerical evidence and reproducible methodology; limits relate to the absence of the authors' raw measurement files.

Verdict

The reviewed paper (Victoria Tang, 2026) is assessed as clear of detected fraud. Tang's manuscript is a detailed, mathematics-driven reinvestigation of Ait Hammou et al. (2026), and its conclusions against the target paper are well supported by reproducible calculations.

Key findings

Severe morphological data anomalies in the target paper: a recorded carapace length of 6.97 mm for an *A. aeneas* specimen against congeneric ranges of 26.92–40.44 mm, and a telson width of 9.58 mm for *A. liouvillei* against 2.57–4.94 mm. The reconstructed value of 26.97 mm (likely missing leading digit "2") is consistent with biological scaling.
Lack of multiple-testing correction across 23 Kruskal-Wallis tests and 23 ANOVA tests; the family-wise error rate at α=0.05 approaches 69%, rendering the three reported p-values (0.027, 0.030, 0.040) non-significant after Benjamini–Hochberg correction (q≈0.257).
Rank-deficient LDA conducted with p=23 variables against N=12 samples (within-group degrees of freedom N−g=9); the within-group scatter matrix is mathematically singular, yet the authors report a 100% training classification rate.
Internal legend/figure contradiction: text states *A. aeneas* corresponds to red squares and hybrids to green triangles, but Fig. 12 shows 3 red points and 2 green points.
Indicators of AI-assisted writing: contradictory statements between Results and Discussion (ANOVA presence vs. absence), mixing of sample and population SD formulas, and systematic misspelling of *A. liouvillei* as *A. liouvellei*.
Timeline anomaly: the original paper was published in 2026 and the present critique written 2026-05-25, indicating that the detected errors were overt.

Evidence highlights

DOI of the reviewed paper: 10.5281/zenodo.20371057
FWER estimate: 1 − (1 − 0.05)^23 ≈ 0.69
BH-corrected q-values after reanalysis: ≈ 0.257
Dimensional inconsistency for the anomalous specimen: Cara_L = 6.97 mm vs. Cara_PW = 7.39 mm (width exceeds length)
Linear-algebra impossibility: p (=23) > N−g (=9) makes the within-group scatter matrix singular; MASS::lda() would error with "variables are collinear"
Reported telson length-to-width ratio of approximately 1.08 vs. congeneric range 1.68–2.79
Source PDF: Elysium 2026 No 3_260530_055254.pdf

Notes

This report concentrates on evaluating Tang's critique; the ultimate factual adjudication of Ait Hammou et al. (2026) requires an institutional investigation by *Serket* and access to the authors' raw measurements and original R/Python scripts.
All conclusions about the target paper rest on reported numerical values, geometry, and basic statistical theory; any reframing by the target authors would require additional verification.
The reviewed paper itself does not display the typical signatures of fabricated figures or data in Tang's own contribution.