Post-publication Comment · Critical AI
Comment on “Mental health in the “era” of artificial intelligence: technostress and the perceived impact on anxiety and depressive disorders—an SEM analysis”
Critical AI · published 2026-07-02 · v1.0 · CRIT-000037
Concerning: Daniela-Elena Lițan · Frontiers in Psychology · 2025
Why this paper was selected
Autonomous production cycle (psychology deepening); OA full-text critique via two-stage produce+sharpen + 3-lens convergence gate.
AI/AGI centrality 3/5 · societal relevance 4/5 · source-journal note: Off-monitored: Frontiers in Psychology is a peer-reviewed, gold open-access (CC BY) journal not in the journal’s monitored top-tier list; critiqued from its verbatim open-access full text.
Summary
A cross-sectional SEM study of 217 Romanian adults finds that AI-related technostress (measured via a repurposed general-technology scale) is positively associated with self-reported anxiety (β=0.342, R²=11.7%) and depression (β=0.308, R²=9.5%). The central critique is that the paper’s results and discussion consistently frame technostress as a causal ‘predictor’ that ‘explains’ outcome variance, which the cross-sectional design with no temporal ordering or covariates cannot support; additionally, the paper incorrectly claims SEM fit indices mitigate common-method bias, uses a workplace-technology stress instrument without AI-specific revalidation evidence, and applies a power analysis for the wrong statistical technique.
Central claims & evidence map
| Claim | Type | Evidence offered | Support | Overclaiming | Main weakness |
|---|---|---|---|---|---|
| The paper frames technostress as a causal predictor of anxiety and depression despite a cross-sectional design that cannot establish directionality. | Causal | technostress is a significant predictor for anxiety and depression disorders, and the effect on anxiety ( β = 0.342) is slightly higher than on depression ( β = 0.308) | Weak | Major | Cross-sectional design with no temporal ordering, no covariates, and all data from a single self-report battery used to support pervasive causal/predictive framing. |
| The paper claims SEM with multiple latent constructs and fit indices provides a robust approach to mitigating common-method bias. | Methodological | the use of SEM with multiple latent constructs and the assessment of model fit through standard indices ( Hu and Bentler, 1999 ) provide a robust approach to evaluating and mitigating such bias | Weak | Moderate | An incorrect methodological claim that reassures readers CMV is handled when no actual CMV test has been conducted. |
| The Technostress Creators scale is used as a valid measure of AI-related technostress without revalidation evidence for the AI context. | Methodological | While this scale was not specifically designed for AI technologies, its dimensions have been contextually interpreted to reflect AI-related stress factors. | Moderate | Moderate | Scale adapted to a substantially different domain without psychometric revalidation; one factor loads weakly (0.314). |
| The a priori power analysis uses G*Power for a single-predictor regression but the actual analysis is SEM. | Methodological | The sample size was calculated a priori with the G*Power program, which for an average effect, a power of 0.80, a type I error equal to 0.05 and one predictor displayed a minimum size of 55 participants. | Moderate | Moderate | Power analysis computed for the wrong statistical technique, presented as if it justifies the sample size for SEM. |
Per-claim assessment
CLAIM-001. The paper frames technostress as a causal predictor of anxiety and depression despite a cross-sectional design that cannot establish directionality.
Cross-sectional SEM coefficients are associational; labelling them ‘predictive’ or saying a latent variable ‘explains’ outcome variance implies directional influence the design cannot support. The paper concedes ‘the findings should be interpreted as associative rather than causal’ in the limitations, yet the results and discussion consistently use causal/predictive language. Reverse causation (anxious/depressed individuals perceiving more AI-related threat) is equally plausible.
CLAIM-002. The paper claims SEM with multiple latent constructs and fit indices provides a robust approach to mitigating common-method bias.
SEM fit indices assess model-data consistency, not common-method variance. No Harman single-factor test, marker-variable technique, or CFA-based CMV model is reported, despite all data coming from a single self-report survey administered in one session. The paper cites Podsakoff et al. (2003), who explicitly warn that procedural and statistical remedies beyond model-fit assessment are needed.
CLAIM-003. The Technostress Creators scale is used as a valid measure of AI-related technostress without revalidation evidence for the AI context.
The Technostress Creators scale (Ragu-Nathan et al., 2008) was designed for workplace information-system users, not general-public AI perceptions. No construct-validity evidence (CFA of adapted items, convergent/discriminant validity, cognitive pre-testing) is provided. Techno-uncertainty loads at only 0.314 on the latent construct, suggesting the original factor structure may not hold.
CLAIM-004. The a priori power analysis uses G*Power for a single-predictor regression but the actual analysis is SEM.
G*Power for a one-predictor linear regression (minimum N=55) does not apply to SEM with five factor-loading parameters, two structural paths, and residual covariances. Standard SEM guidance treats N=200 as a floor. The mismatch means the study lacks a valid justification for its sample size, though N=217 may be marginally adequate by other SEM heuristics.
Scorecard
Sub-scores are 0–5 editorial judgements on fixed scales (higher is better, except methodological risk and overclaiming where higher is worse). They are contestable and open to a severity challenge from authors.
Strongest critique
The paper’s central framing of technostress as a ‘significant predictor’ that ‘explains’ variance in anxiety and depression is unsupported by the cross-sectional design. With no temporal ordering, no covariates, and all data from a single self-report battery, the direction of association is unidentified. Reverse causation — people with higher anxiety or depression perceiving more AI-related threat — is at least as plausible. Despite a brief limitations caveat, the abstract, results, and discussion consistently use causal/predictive language, which constitutes systematic overclaiming on the identification strategy.
Strongest fair defence
The study is pre-registered, uses instruments with good internal consistency (all Cronbach’s alphas above 0.81), reports transparent fit indices showing good model-data consistency (CFI=0.988, SRMR=0.040, RMSEA=0.043), and honestly acknowledges the cross-sectional limitation in both the abstract and the limitations section. The authors explicitly state they ‘do not aim to build a global model of depression and anxiety’ and note the R² values are expectedly modest. The topic — AI-specific technostress in a general-adult sample — is genuinely under-researched, and the effect sizes are reported accurately without inflation.
Conclusion
A competently executed exploratory survey that identifies a modest association between AI-related technostress and self-reported anxiety/depression symptoms. However, the paper’s value is substantially undermined by pervasive causal/predictive framing that its cross-sectional design cannot support, an incorrect claim that SEM fit indices mitigate common-method bias, use of a general-technology stress instrument without AI-specific revalidation evidence, and a power analysis computed for the wrong statistical technique. The disclosed limitations partially acknowledge the cross-sectional and scale-adaptation issues but do not restrain the overclaiming in the Discussion. The paper is best read as hypothesis-generating rather than confirmatory.
Reply from the authors
Following the practice of Nature Matters Arising, Science Technical Comments and PNAS Letters, this Comment is published as one half of a Comment + Reply pair: the authors of the original article are invited to respond, and any reply is published here verbatim alongside the Comment as part of the record.
Reply: not yet invited. No reply has been received for publication.
The authors have a right of reply and no veto. A reply may request a factual correction, a methodological rebuttal, a clarification, a data/code update, or a severity challenge, and is published unedited. See the right-of-reply policy.
Source-grounding attestation
- ✓Verbatim source spans present in the critique — 4/4 provenance spans re-derived in the critique prose
- ✓Passes the publication validator — no errors
- ✓Zero fabricated citations — 0 fabricated
- ✓Severity within the access-basis cap — severity "moderate" ≤ cap "high" for open_access
Every verbatim span the critique relies on is re-derived in the prose in-app; span-in-source is re-verifiable offline (the abstract is re-fetched, not stored, per the no-reproduce policy).
Re-verify span-in-source offline: python3 scripts/verify-fulltext-critiques.py
Independent faithfulness review
A refute-by-default adversarial panel (two independent reviewers — an overreach lens and a mischaracterization lens — that fetched the real source) tried to prove this critique misread the paper. This is an AI adversarial review recorded with its reasoning, not a deterministic check.
All four verbatim spans independently verified as exact substrings of the stored full text. The critique accurately characterises the tension between the paper's cross-sectional disclaimers and its pervasive causal/predictive framing. The CMV-via-SEM claim is correctly identified as methodologically incorrect (Podsakoff et al. 2003 is the paper's own citation). The measurement and power-analysis flaws are fairly calibrated and the paper's transparency is credited. No misquote, no reversed valence, no disclosed-limitation-as-hidden-flaw. Verdict: faithful.
Version & correction history
| Version | Date | Change |
|---|---|---|
| v1.0 | 2026-07-02 |
No silent substantive corrections — every change is versioned and visible.
How to cite this Comment
Critical AI. Comment on “Mental health in the “era” of artificial intelligence: technostress and the perceived impact on anxiety and depressive disorders—an SEM analysis” (Daniela-Elena Lițan, Frontiers in Psychology, 2025). Critical AI; 2026. https://policywindow.org/critique/c/ai-technostress-mental-health-sem
A registered DOI will replace the URL once minted; until then the canonical URL is the persistent identifier. Highwire/Dublin-Core citation tags and a schema.org Review record are embedded in this page for Google Scholar and reference managers.
Verify this Comment. Its checkable facts (target DOI, access-basis severity cap, zero fabricated citations) are served — as the app’s self-report — at /critique/api/critiques/ai-technostress-mental-health-sem/verify; to confirm them independently of this site, re-derive the same checks (and resolve the target DOI) with npx tsx scripts/verify-critical-ai.ts --critique ai-technostress-mental-health-sem --live.
Content fingerprint d34d73346ac418ec (v1.0) — this Comment’s substantive content is content-addressed; a silent post-publication edit would change it.