Academic Integrity Report on 'The Effect of Academic Uncertainty on Anxiety: The Mediating Role of Academic Control and Educational Intervention Pathways' (DOI: 10.16298/j.cnki.1004-3667.2026.04.07)

Academic fraud report · Geng Detector

Summary

This report assesses a paper by Liu Jingyuan and Jiao Ran published in Zhongguo Gaojiao Yanjiu (2026, Issue 4), concluding that the paper contains fabricated or improperly assembled statistical data (verdict: confirmed serious concerns). Five key issues are identified. First, the paper reports an impossible paired-samples t-test: with df=20, t=0.037 is reported alongside p=0.007 and Cohen's d=0.806; such a small t corresponds to p≈0.97, making the claimed significance and large effect size statistically impossible. Second, Experiment 2 reports an implausibly small standard deviation (SD=2.57) on a 0–100 visual analogue scale for n=18 participants on a subjective psychological measure, indicating artificially compressed variance. Third, group labels M_high and M_low in the results section are inverted relative to the verbal description. Fourth, terminal-digit analysis of reported means and SDs across Tables 1–3 shows a non-uniform distribution heavily concentrated on .14, .44, .33 and .56, suggestive of fabricated or scripted data. Fifth, the ethics approval code (20160907) appears to be reused from an unrelated, much older project. Confidence is high for findings 1, 3 and 4; moderate for findings 2 and 5 given reliance on contextual inference.

Verdict

🔴 Confirmed serious concerns — multiple independent indicators of fabricated or improperly assembled quantitative results. Confidence: high for the statistical impossibility (Finding 1) and label inversion (Finding 3); moderate-to-high for terminal-digit anomaly (Finding 4) and implausible SD (Finding 2); moderate for the suspect ethics approval (Finding 5).

Key findings

Finding 1 (Severity 🔴): Page 53, Experiment 3. Paired-samples t-test reported as *t(20)=0.037, p=0.007, d=0.806*. With df=20, t=0.037 yields p≈0.97; the reported p-value and large effect size are mathematically incompatible with the reported t.
Finding 2 (Severity 🔴): Pages 52–53, Experiment 2 manipulation check. On a 0–100 subjective rating scale, the experimental group (n=18) has M=80.61, SD=2.57; the control group (n=18) has M=87.39, SD=4.15. SD=2.57 across 18 participants on a subjective emotional measure is implausibly narrow. The same construct in Experiment 1 has SD=21.08.
Finding 3 (Severity 🟠): Page 52, Experiment 2. The text states that the experimental group scored significantly *lower* than the control, but the parenthetical labels the experimental mean as M_high=80.61 and the control as M_low=87.39 — an inverted label–description mismatch.
Finding 4 (Severity 🔴): Across Tables 1, 2 and 3, reported means and SDs to two decimal places show digit clustering inconsistent with a uniform distribution: ".14" appears 4 times; ".44" appears 2 times; ".33", ".56" each appear 2 times. Examples cited: 39.44, 52.33, 44.83, 86.00, 46.56, 50.14, 32.14, 72.14.
Finding 5 (Severity 🟠): Page 50. The ethics approval code is given as 20160907, while the project funding period is stated as 2024–2025 and cited literature extends to 2023–2025, suggesting reuse of an older, possibly unrelated approval.

Evidence highlights

Direct numeric quote (Finding 1): "t(20)=0.037, p=0.007, d=0.806" — under df=20, a two-tailed p≈0.97 is expected for t=0.037.
Direct numeric quote (Finding 2): experimental group M=80.61, SD=2.57, n=18; control group M=87.39, SD=4.15, n=18.
Direct numeric quote (Finding 3): "M高=80.61, SD高=2.57, M低=87.39, SD低=4.15" with the textual claim that the experimental group is significantly *lower*.
Direct digit-frequency examples (Finding 4): 39.44, 52.33, 44.83, 86.00, 46.56, 50.14, 32.14, 72.14.
Direct numeric quote (Finding 5): ethics approval number 20160907.
DOI preserved: 10.16298/j.cnki.1004-3667.2026.04.07.

Notes

All quotations are reproduced from the report; the underlying PDF was not independently verified line-by-line.
Finding 1 alone is sufficient to render the reported Experiment 3 inferential statistics invalid; combined with Findings 2, 3 and 4, the pattern is consistent with data fabrication or template copying rather than genuine measurement error.
Terminal-digit analysis (Finding 4) is suggestive but not conclusive on its own; it should be combined with the other findings.
Final determination of academic misconduct requires investigation by the journal and the authors' institution; raw data, analysis scripts and original ethics approval documentation should be requested.