pith. sign in

arxiv: 2604.10991 · v1 · submitted 2026-04-13 · 💻 cs.HC · cs.AI

Examining EAP Students' AI Disclosure Intention: A Cognition-Affect-Conation Perspective

Yiran Du , Huimin He This is my paper

Pith reviewed 2026-05-10 16:16 UTC · model grok-4.3

classification 💻 cs.HC cs.AI
keywords AI disclosure intentionEAP studentspsychological safetyfear of negative evaluationgenerative AIacademic integritymixed-methods studyhigher education
0
0 comments X

The pith

Psychological safety increases EAP students' intention to disclose AI tool use in writing, while fear of negative evaluation decreases it.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper examines the psychological factors that shape whether English for Academic Purposes students plan to reveal their use of generative AI tools when completing academic writing tasks. It applies the cognition-affect-conation framework to separate enabling influences from inhibiting ones and tests the resulting model with survey data from 324 students at a Chinese university. Structural equation modeling shows that feeling psychologically safe raises disclosure intention while fear of being judged lowers it. Follow-up interviews indicate that clear teacher guidance builds safety and that vague policies or worries about reputation strengthen fear. The results point to practical steps institutions can take to encourage honest reporting of AI assistance.

Core claim

Drawing on the cognition-affect-conation framework, the study finds that psychological safety positively predicts EAP students' intention to disclose generative AI use in academic writing, whereas fear of negative evaluation negatively predicts the same intention. Quantitative analysis of 324 student surveys supports these directional effects. Qualitative interviews with 15 students add that supportive teacher practices and explicit guidance strengthen safety, while policy ambiguity and concerns over reputation intensify fear and reduce willingness to disclose.

What carries the argument

The cognition-affect-conation framework applied to AI disclosure, in which cognitive appraisals of safety and evaluation risk shape affective responses that in turn drive the conative intention to report tool use.

If this is right

  • Clear institutional policies on AI use reduce fear of negative evaluation and thereby support higher disclosure rates.
  • Supportive teacher practices raise psychological safety and increase students' willingness to report AI assistance.
  • Ambiguous or absent guidance on AI policies heightens fear and discourages transparent reporting.
  • Transparent AI use in academic work becomes more likely when both safety is fostered and evaluation fears are lowered.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same safety-versus-fear pattern may operate in non-EAP courses or in disciplines outside English-medium instruction.
  • Targeted workshops for instructors on creating psychologically safe classrooms could be tested as an intervention.
  • Future studies could measure actual disclosure behavior rather than intention to check whether the reported intentions translate into practice.

Load-bearing premise

That the framework and the specific survey items chosen accurately capture the main psychological drivers of disclosure decisions for these students.

What would settle it

A new survey of comparable EAP students that finds no reliable statistical link between measured psychological safety scores and reported disclosure intention would undermine the central quantitative claim.

Figures

Figures reproduced from arXiv: 2604.10991 by Huimin He, Yiran Du.

Figure 1
Figure 1. Figure 1: The Conceptual Model [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Psychological safety had a significant positive effect on AI disclosure intention (β [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
read the original abstract

The growing use of generative artificial intelligence (AI) in academic writing has raised increasing concerns regarding transparency and academic integrity in higher education. This study examines the psychological factors influencing English for Academic Purposes (EAP) students' intention to disclose their use of AI tools. Drawing on the cognition-affect-conation framework, the study proposes a model integrating both enabling and inhibiting factors shaping disclosure intention. A sequential explanatory mixed-methods design was employed. Quantitative data from 324 EAP students at an English-medium instruction university in China were analysed using structural equation modelling, followed by semi-structured interviews with 15 students to further interpret the findings. The quantitative results indicate that psychological safety positively predicts AI disclosure intention, whereas fear of negative evaluation negatively predicts it. The qualitative findings further reveal that supportive teacher practices and clear guidance foster psychological safety, while policy ambiguity and reputational concerns intensify fear of negative evaluation and discourage disclosure. These findings highlight the importance of clear institutional policies and supportive pedagogical environments in promoting transparent AI use.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. This paper examines psychological factors shaping English for Academic Purposes (EAP) students' intention to disclose generative AI use in academic writing. It applies the cognition-affect-conation framework to model psychological safety as an enabling factor and fear of negative evaluation as an inhibiting factor. A sequential explanatory mixed-methods design is used: structural equation modeling (SEM) on survey data from 324 students at a Chinese English-medium university, followed by semi-structured interviews with 15 students. The quantitative results report that psychological safety positively predicts disclosure intention while fear of negative evaluation negatively predicts it; the qualitative data attribute safety to supportive teacher practices and clear guidance, and fear to policy ambiguity and reputational concerns. The study concludes that clear institutional policies and supportive environments promote transparent AI use.

Significance. If the SEM paths and qualitative interpretations hold after proper validation, the work contributes empirical evidence on enabling and inhibiting factors for AI disclosure in higher education, particularly for EAP contexts. The mixed-methods design allows both directional quantitative findings and explanatory qualitative insights, which could inform policy and pedagogy on academic integrity and AI transparency. The focus on a non-Western EMI setting adds contextual value to the literature on human-AI interaction in writing.

major comments (2)
  1. [Methods] Methods section: The manuscript reports SEM results on paths from psychological safety and fear of negative evaluation to AI disclosure intention but provides no model fit statistics (CFI, RMSEA, SRMR, chi-square), no details on the measurement model (factor loadings, AVE, CR, discriminant validity via Fornell-Larcker or HTMT), and no information on how the survey scales were adapted or validated for this population. These omissions are load-bearing because the central directional claims rest on the fitted model.
  2. [Data Collection and Analysis] Data analysis and results: Exclusion criteria for the 324 responses (e.g., incomplete surveys, attention checks, outliers) are not described, nor is the interview coding scheme, thematic analysis procedure, or any inter-rater reliability metric for the 15 transcripts. Without these, the robustness of both the quantitative path coefficients and the qualitative support for the enabling/inhibiting factors cannot be assessed.
minor comments (2)
  1. [Abstract] The abstract states the sample size and directional findings but could briefly note the key fit or validity metrics once added, to allow readers to gauge the quantitative claims at a glance.
  2. [Results] Figure or table presenting the SEM path diagram and standardized coefficients would improve clarity; currently the results are described only in text.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed comments, which help strengthen the transparency and rigor of our work. We address each major point below and will incorporate the necessary additions in the revised manuscript.

read point-by-point responses

  1. Referee: [Methods] Methods section: The manuscript reports SEM results on paths from psychological safety and fear of negative evaluation to AI disclosure intention but provides no model fit statistics (CFI, RMSEA, SRMR, chi-square), no details on the measurement model (factor loadings, AVE, CR, discriminant validity via Fornell-Larcker or HTMT), and no information on how the survey scales were adapted or validated for this population. These omissions are load-bearing because the central directional claims rest on the fitted model.

    Authors: We agree that these reporting details are essential and were omitted from the original submission. In the revised manuscript, we will expand the Methods section to include all model fit indices (CFI, RMSEA, SRMR, and chi-square), a full description of the measurement model with factor loadings, AVE, CR, and discriminant validity assessed via both Fornell-Larcker and HTMT criteria. We will also detail the adaptation process for the survey scales, including their origins in established literature and any validation steps undertaken for the EAP student population. revision: yes

  2. Referee: [Data Collection and Analysis] Data analysis and results: Exclusion criteria for the 324 responses (e.g., incomplete surveys, attention checks, outliers) are not described, nor is the interview coding scheme, thematic analysis procedure, or any inter-rater reliability metric for the 15 transcripts. Without these, the robustness of both the quantitative path coefficients and the qualitative support for the enabling/inhibiting factors cannot be assessed.

    Authors: We acknowledge these procedural details are missing and limit assessment of robustness. In the revision, we will add a clear account of quantitative data screening, including criteria for excluding incomplete responses, any attention checks applied, and outlier handling. For the qualitative data, we will describe the coding scheme, the thematic analysis procedure used, and any inter-rater reliability metrics calculated for the interview transcripts. revision: yes

Circularity Check

0 steps flagged

No significant circularity; empirical model tested on new data

full rationale

The paper draws on the established cognition-affect-conation framework to hypothesize enabling (psychological safety) and inhibiting (fear of negative evaluation) factors, then tests directional predictions via SEM on fresh survey data from n=324 students plus follow-up interviews. Results are direct outputs of the fitted model on independent observations rather than reductions to inputs by construction. No self-definitional steps, no fitted parameters renamed as predictions, no load-bearing self-citations, and no ansatz or uniqueness claims imported from prior author work. The derivation chain is data-driven and externally falsifiable.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Only abstract available; no explicit free parameters, axioms, or invented entities can be identified. The study relies on standard psychological constructs and statistical modeling whose details are not provided.

pith-pipeline@v0.9.0 · 5467 in / 1082 out tokens · 43063 ms · 2026-05-10T16:16:34.249868+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

10 extracted references · 10 canonical work pages

  1. [1]

    Introduction The rapid development of artificial intelligence (AI), particularly generative AI technologies, has significantly transformed language learning and academic writing practices in higher education. In English for Academic Purposes (EAP) contexts, AI-powered tools such as natural language processing systems and generative writing assistants can ...

    work page 2025

  2. [2]

    Literature Review 2.1 AI in EAP Education Artificial intelligence (AI) technologies have become increasingly visible in English for Academic Purposes (EAP) education, particularly with the development of natural language processing systems and generative AI tools capable of producing extended academic discourse (Wang et al., 2026). These technologies can ...

    work page 2026

  3. [3]

    Theoretical Framework This study adopts a cognition–affect–conation framework to examine EAP students’ AI disclosure intention. The framework conceptualises behaviour as a sequential process in which cognitive evaluations influence affective responses, which subsequently shape behavioural intentions (Zhou & Zhang, 2024). It has been widely applied in beha...

    work page 2024

  4. [4]

    Research Questions and Hypotheses 4.1 Enablers of AI Disclosure Intention Based on the conceptual model, the first research question (RQ1) asks: How do psychological safety, perceived fairness, perceived teacher support, and self-efficacy affect EAP students’ AI disclosure intention? Psychological safety refers to the perception that individuals can ackno...

    work page 2023

  5. [5]

    In the first phase, a questionnaire survey was administered to examine the relationships among the constructs in the proposed conceptual model of AI disclosure intention

    Methods 5.1 Research Design This study adopted a sequential explanatory mixed-methods design, in which quantitative data collection and analysis were followed by qualitative inquiry to further interpret the quantitative findings (Cohen et al., 2018). In the first phase, a questionnaire survey was administered to examine the relationships among the constru...

    work page 2018

  6. [6]

    The quantitative results show that when students perceive the learning environment as psychologically safe, they are more willing to report their AI use

    Discussion 7.1 Enablers of AI Disclosure Intention The findings indicate that psychological safety plays a central role in enabling students’ intention to disclose AI use in academic writing. The quantitative results show that when students perceive the learning environment as psychologically safe, they are more willing to report their AI use. The intervi...

    work page 2023

  7. [7]

    The findings indicate that disclosure intention is shaped by the interaction between enabling and inhibiting mechanisms

    Conclusion This study examined the psychological factors influencing EAP students’ AI disclosure intention through a cognition–affect–conation framework using a sequential explanatory mixed-methods design. The findings indicate that disclosure intention is shaped by the interaction between enabling and inhibiting mechanisms. Psychological safety emerged a...

    work page doi:10.1016/j.ssaho.2025.102357 2026

  8. [8]

    G., Harrison, N

    https://doi.org/10.1007/s44217-024-00194-8 Somerville, S. G., Harrison, N. M., & Lewis, S. A. (2023). Twelve tips for the pre-brief to promote psychological safety in simulation-based education. Medical Teacher, 45(12), 1349–1356. https://doi.org/10.1080/0142159X.2023.2214305 Stone, B. W. (2025). Generative AI in higher education: Uncertain students, ambi...

    work page doi:10.1007/s44217-024-00194-8 2023

  9. [9]

    https://doi.org/10.1186/s40594-024-00493-4 Usman, M., Cheng, J., Ghani, U., Gul, H., & Shah, W. U. (2021). Social support and perceived uncertainties during COVID-19: Consequences for employees’ wellbeing. Current Psychology. https://doi.org/10.1007/s12144-021-02293-3 Vella, S. A., Mayland, E., Schweickle, M. J., Sutcliffe, J. T., McEwan, D., & Swann, C. ...

    work page doi:10.1186/s40594-024-00493-4 2021

  10. [10]

    https://doi.org/10.3390/jtaer20020099 Zhou, T., & Wu, X. (2025b). Examining generative AI user disclosure intention: An ELM perspective. Universal Access in the Information Society, 24(2), 1209–1220. https://doi.org/10.1007/s10209-024-01130-1 Zhou, T., & Zhang, C. (2024). Examining generative AI user addiction from a C-A-C perspective. Technology in Socie...

    work page doi:10.3390/jtaer20020099 2024