The impact of coercive, normative, and mimetic Stress on Chinese teachers' continuance intention to use generative AI: An integrated perspective of the Expectation-Confirmation Model and Institutional Theory

Huimin He; Kai Cui; Kunjie Jia; Yiran Du

arxiv: 2605.00522 · v1 · submitted 2026-05-01 · 💻 cs.HC

The impact of coercive, normative, and mimetic Stress on Chinese teachers' continuance intention to use generative AI: An integrated perspective of the Expectation-Confirmation Model and Institutional Theory

Kunjie Jia , Kai Cui , Huimin He , Yiran Du This is my paper

Pith reviewed 2026-05-09 18:54 UTC · model grok-4.3

classification 💻 cs.HC

keywords generative AIcontinuance intentionExpectation-Confirmation ModelInstitutional TheoryChinese teacherscoercive pressurenormative pressuremimetic pressure

0 comments

The pith

An integrated model combining expectation confirmation and institutional pressures explains Chinese teachers' continued use of generative AI.

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

This study combines the Expectation-Confirmation Model, which links met expectations to satisfaction and perceived usefulness, with Institutional Theory's three types of pressures to predict teachers' ongoing adoption of generative AI. Data from 437 Chinese teachers via structural equation modeling shows that confirmation, usefulness, and satisfaction drive continuance intention, and so do coercive mandates, normative expectations from colleagues, and mimetic copying of successful practices. Interviews reveal teachers apply the tool practically for lesson preparation and idea generation yet remain wary of inaccuracies in AI outputs. Readers should care because these findings point to ways both personal experience and organizational context shape technology persistence in education settings.

Core claim

The paper establishes that an integration of the Expectation-Confirmation Model and Institutional Theory accounts for Chinese teachers' continuance intention to use generative AI, with confirmation, perceived usefulness, and satisfaction from the former and coercive, normative, and mimetic pressures from the latter all exerting positive effects, as confirmed by quantitative analysis of survey data and qualitative interpretation from interviews where teachers describe pragmatic yet cautious usage for educational tasks.

What carries the argument

The integrated Expectation-Confirmation Model and Institutional Theory, which combines individual cognitive evaluations of AI performance with external institutional forces to explain sustained technology use.

Load-bearing premise

The assumption that self-reported data from a sample of Chinese teachers accurately represents their actual continuance behavior and that no unaccounted cultural or systemic factors in the education system alter the relationships in the model.

What would settle it

A study that measures teachers' actual logged usage of generative AI over time and finds it uncorrelated with the reported levels of institutional pressures or satisfaction would challenge the model's validity.

Figures

Figures reproduced from arXiv: 2605.00522 by Huimin He, Kai Cui, Kunjie Jia, Yiran Du.

**Figure 1.** Figure 1: The Conceptual Model view at source ↗

**Figure 2.** Figure 2: Structural Model Results Confirmatory factor analysis was conducted to assess the measurement model. As shown in Appendix F, the measurement model demonstrated satisfactory model fit (χ²/df = 2.21, CFI = 0.95, TLI = 0.94, RMSEA = 0.053, SRMR = 0.046), meeting recommended criteria. The results for reliability and view at source ↗

read the original abstract

This study investigates Chinese teachers' continuance intention to use generative artificial intelligence (AI) by integrating the Expectation-Confirmation Model with Institutional Theory. A sequential explanatory mixed-methods design was employed. Questionnaire data from 437 teachers were analysed using structural equation modelling, followed by semi-structured interviews with 15 teachers to further interpret the findings. The results indicate that confirmation, perceived usefulness, and satisfaction play important roles in shaping teachers' continuance intention, while institutional pressures, including coercive, normative, and mimetic influences, also contribute to continued use. Qualitative findings further reveal that teachers often use generative AI pragmatically to support tasks such as lesson preparation and idea generation, while simultaneously exercising caution and critically evaluating the reliability of AI-generated content. These findings highlight the combined influence of individual evaluations and institutional contexts on teachers' sustained engagement with generative AI in education.

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.

Desk Editor's Note private letter to a colleague

read the letter

This paper applies the Expectation-Confirmation Model and Institutional Theory to Chinese teachers using generative AI and finds that both personal satisfaction factors and coercive/normative/mimetic pressures shape reported continuance intention. The mixed-methods setup with 437 survey responses analyzed via SEM plus 15 interviews is the main concrete output. It gives a clear picture of how teachers use the tools for lesson prep and idea generation while still checking outputs for reliability. That qualitative layer helps make the quantitative paths feel less abstract and shows the pragmatic side of adoption in Chinese schools. The integration itself is not new, but the specific population and technology combination adds some fresh data points on how institutional forces interact with individual confirmation and usefulness judgments. The soft spot is the exclusive reliance on self-reported intention. No usage logs, login records, or observed task data back it up, so the link to actual continued use stays untested. In a setting where schools may push AI adoption from above, social-desirability effects could easily inflate the numbers without matching behavior. The abstract also skips the path coefficients, fit indices, and effect sizes, which makes it harder to judge how strongly the institutional variables really add beyond the classic ECM paths. This work is for researchers who study technology continuance in education or who need comparative data from non-Western contexts. A reader running similar surveys elsewhere could use the Chinese results as a benchmark. It deserves a serious referee because the sample size and mixed-methods design are solid enough to warrant external scrutiny, even if the behavioral validation needs tightening. I would send it for review with a note to address the self-report limitation.

Referee Report

2 major / 2 minor

Summary. The paper investigates Chinese teachers' continuance intention to use generative AI by integrating the Expectation-Confirmation Model with Institutional Theory. It uses a sequential explanatory mixed-methods design: structural equation modeling on questionnaire data from 437 teachers, followed by semi-structured interviews with 15 teachers. The results indicate that confirmation, perceived usefulness, and satisfaction shape continuance intention, while coercive, normative, and mimetic institutional pressures also contribute; qualitative data highlight pragmatic use for tasks like lesson preparation alongside caution about AI content reliability.

Significance. If the empirical results hold after addressing reporting and validation gaps, the study meaningfully extends technology adoption research by combining individual cognitive factors from ECM with institutional pressures in the specific context of Chinese education. The mixed-methods design and focus on generative AI provide timely insights for educational technology policy and practice.

major comments (2)

[Methods and Results] Methods and Results sections: the central claim that institutional pressures and ECM constructs contribute to continued use rests on self-reported continuance intention from the 437-teacher survey, with no cross-validation against objective behavioral data such as usage logs, login frequency, or task completion records. This is load-bearing because social-desirability bias and institutional expectations in the Chinese education setting could inflate reported intentions without corresponding actual behavior.
[Results] Results section: the manuscript provides no specific SEM path coefficients, standardized loadings, model fit indices (e.g., CFI, RMSEA, SRMR), or effect sizes for the hypothesized relationships among confirmation, perceived usefulness, satisfaction, institutional pressures, and continuance intention. Without these statistics or the full measurement instrument and validity tests (convergent/discriminant), the statistical support for the claims cannot be fully assessed.

minor comments (2)

[Abstract] Abstract: lacks any quantitative indicators (e.g., key path significance or R² values) that would allow readers to gauge the magnitude of the reported effects.
[Discussion] The integration of quantitative and qualitative findings could be strengthened by more explicit mapping of interview themes back to the specific ECM and institutional constructs tested in the SEM.

Simulated Author's Rebuttal

2 responses · 1 unresolved

We thank the referee for the thoughtful and constructive comments on our manuscript. We have carefully reviewed the feedback on the methods and results and provide point-by-point responses below. Where appropriate, we have made revisions to strengthen the paper while remaining faithful to the original study design and data.

read point-by-point responses

Referee: [Methods and Results] Methods and Results sections: the central claim that institutional pressures and ECM constructs contribute to continued use rests on self-reported continuance intention from the 437-teacher survey, with no cross-validation against objective behavioral data such as usage logs, login frequency, or task completion records. This is load-bearing because social-desirability bias and institutional expectations in the Chinese education setting could inflate reported intentions without corresponding actual behavior.

Authors: We appreciate the referee's concern regarding reliance on self-reported continuance intention. This approach aligns with the predominant methodology in Expectation-Confirmation Model studies and technology adoption research, where objective usage logs are frequently unavailable, especially in K-12 educational contexts. We agree that social-desirability bias and institutional pressures in China represent a valid threat to validity. In the revised manuscript, we have expanded the Limitations section to explicitly discuss this issue and to recommend that future work incorporate behavioral trace data where feasible. The sequential explanatory mixed-methods design offers partial mitigation through the interview data, in which teachers described their actual, pragmatic usage patterns (e.g., lesson preparation) alongside expressed caution. We cannot, however, retroactively obtain usage logs or login records because the study collected only survey and interview data. revision: partial
Referee: [Results] Results section: the manuscript provides no specific SEM path coefficients, standardized loadings, model fit indices (e.g., CFI, RMSEA, SRMR), or effect sizes for the hypothesized relationships among confirmation, perceived usefulness, satisfaction, institutional pressures, and continuance intention. Without these statistics or the full measurement instrument and validity tests (convergent/discriminant), the statistical support for the claims cannot be fully assessed.

Authors: We regret that the detailed statistical reporting was insufficient in the original submission. We have revised the Results section to include all requested information: unstandardized and standardized path coefficients with significance levels and standard errors for every hypothesized relationship, factor loadings, model fit indices (CFI, RMSEA, SRMR, chi-square/df), and effect sizes (R² values). We have also added the complete measurement instrument as an appendix and reported convergent validity (AVE > 0.5, CR > 0.7) and discriminant validity (Fornell-Larcker criterion and HTMT ratios). These additions enable readers to fully evaluate the statistical support for the integrated ECM-institutional model. revision: yes

standing simulated objections not resolved

We cannot supply objective behavioral data (usage logs, login frequency, or task records) for cross-validation because no such data were collected in the original study.

Circularity Check

0 steps flagged

No circularity: standard empirical SEM application of established models

full rationale

The paper conducts a sequential explanatory mixed-methods study: it collects questionnaire data from 437 teachers, applies structural equation modeling to test paths drawn from the pre-existing Expectation-Confirmation Model and Institutional Theory, then follows up with 15 interviews. No equations, derivations, or 'predictions' are presented that reduce by construction to fitted parameters or self-defined constructs. Institutional pressures and continuance intention are measured via validated scales and analyzed as independent statistical relationships; the results are reported as empirical outcomes rather than tautological restatements. Self-citations, if present, serve only as background references to the source models and do not form a load-bearing chain that justifies the central claims. The derivation chain is therefore self-contained hypothesis testing against external data.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 0 invented entities

The study rests on standard assumptions of structural equation modeling and the applicability of the two cited theories to this context; no new entities are introduced. Because only the abstract is available, a full parameter audit is not possible.

free parameters (1)

SEM path coefficients and loadings
The model estimates multiple relationships from the 437-teacher survey data; these are fitted rather than derived from first principles.

axioms (2)

domain assumption The Expectation-Confirmation Model and Institutional Theory constructs are valid and measurable via self-report in the Chinese teacher population.
The integration assumes these frameworks transfer without modification to generative AI use in education.
standard math Structural equation modeling assumptions (linearity, multivariate normality, no omitted variable bias) hold for the collected data.
Standard for any SEM analysis; not tested or discussed in the abstract.

pith-pipeline@v0.9.0 · 5464 in / 1429 out tokens · 38756 ms · 2026-05-09T18:54:12.697206+00:00 · methodology

discussion (0)

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

Examining University Students' Artificial Intelligence-Generated Content (AIGC) Verification Intention from a Protection Motivation Perspective
cs.HC 2026-05 unverdicted novelty 5.0

Survey of 432 students finds protection motivation positively predicts AIGC verification intention, with perceived severity, vulnerability, response efficacy, and self-efficacy as positive influences and maladaptive r...

Reference graph

Works this paper leans on

15 extracted references · 7 canonical work pages · cited by 1 Pith paper

[1]

These tools can support lesson planning, instructional design, explanations, feedback, and other teaching tasks (C

Introduction Generative AI has rapidly become important in education, especially after the release of large language models such as ChatGPT. These tools can support lesson planning, instructional design, explanations, feedback, and other teaching tasks (C. Wang et al., 2026; Lee et al., 2026; Du et al., 2025; Aljohani, 2026). However, concerns remain abou...

2026
[2]

smart education

Literature Review 2.1 Generative AI in Education Generative artificial intelligence has rapidly emerged as a transformative technology in education following the public release of large language models such as ChatGPT (C. Wang et al., 2026). These systems can generate human-like text, explanations, feedback, and instructional materials, enabling new forms...

2026
[3]

Theoretical Framework This study integrates the Expectation-Confirmation Model and Institutional Theory to examine teachers’ continuance intention to use generative AI in the Chinese educational context. The Expectation-Confirmation Model explains users’ post-adoption behaviour based on their experiences with a technology (Bhattacherjee, 2001), whereas In...

2001
[4]

According to this model, continuance intention is influenced by users’ confirmation of expectations, perceived usefulness, and satisfaction

Research Questions and Hypotheses 4.1 The Impact of Confirmation, Perceived Usefulness, and Satisfaction Based on the conceptual model, the first research question (RQ1) asks: How do confirmation, perceived usefulness, and satisfaction influence Chinese teachers’ continuance intention to use generative AI? To address this question, this study draws on the...

2001
[5]

Methods 5.1 Research Design This study used a sequential explanatory mixed-methods design, with quantitative data collection and analysis followed by qualitative inquiry. First, a questionnaire survey examined relationships among constructs in the conceptual model, including Expectation–Confirmation Model factors—confirmation, perceived usefulness, and sa...

2025
[6]

derived from the Expectation-Confirmation Model, and coercive pressure (Sukoco et al., 2022), normative pressure (Hom et al., 2025), and mimetic pressure (Latif et al.,

2022
[7]

Measurement items were adapted from established scales in prior research and slightly modified to fit the context of teachers’ use of generative AI in teaching (Y

derived from Institutional Theory. Measurement items were adapted from established scales in prior research and slightly modified to fit the context of teachers’ use of generative AI in teaching (Y . Du, 2024). The questionnaire was administered in Chinese to ensure participants’ comprehension, and a translation and back-translation procedure was conducte...

2024
[8]

Discussion 7.1 The Impact of Confirmation, Perceived Usefulness, and Satisfaction The findings reinforce the Expectation–Confirmation Model by showing that confirmation, perceived usefulness, and satisfaction are key determinants of teachers’ continuance intention to use generative AI. Consistent with prior continuance research (Bhattacherjee, 2001), conf...

2001
[9]

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

Conclusion This study examined Chinese teachers’ continuance intention to use generative AI by integrating the Expectation–Confirmation Model with Institutional Theory. The findings show that continued use is shaped by both experiential evaluations and institutional influences. Confirmation, perceived usefulness, and satisfaction were important predictors...

work page doi:10.1016/j.ssaho.2025.102357 2026
[10]

https://doi.org/10.2307/3250921 Brislin, R. W. (1970). Back-translation for cross-cultural research. Journal of Cross-Cultural Psychology, 1(3), 185–216. https://doi.org/10.1177/135910457000100301 Chai, C. S., Liang, S., & Wang, X. (2024). A survey study of Chinese teachers’ continuous intentions to teach artificial intelligence. Education and Information...

work page doi:10.2307/3250921 1970
[11]

https://doi.org/10.3390/su17136082 Kim, J. (2024). Types of teacher-AI collaboration in K-12 Classroom instruction: Chinese teachers’ perspective. Education and Information Technologies, 29(13), 17433–17465. https://doi.org/10.1007/s10639-024-12523-3 Kline, R. B. (2023). Principles and practice of structural equation modeling (Fifth edition). The Guilford...

work page doi:10.3390/su17136082 2024
[12]

https://doi.org/10.3390/su12114506 Lee, S., Choe, H., Zou, D., & Jeon, J. (2026). Generative AI (GenAI) in the language classroom: A systematic review. Interactive Learning Environments, 34(1), 335–359. https://doi.org/10.1080/10494820.2025.2498537 Lu, H.-P., & Wang, J.-C. (2023). Exploring the effects of sudden institutional coercive pressure on digital ...

work page doi:10.3390/su12114506 2026
[13]

https://doi.org/10.3390/educsci12120858 Obeid, A., Ibrahim, R., & Fadhil, A. (2024). Extended model of expectation confirmation model to examine users’ continuous intention toward the utilization of E-learning platforms. IEEE Access, 12, 40752–40764. https://doi.org/10.1109/ACCESS.2024.3373190 Ode, E., Nana, R., Boro, I. O., & Ikyanyon, D. N. (2025). A cr...

work page doi:10.3390/educsci12120858 2024
[14]

https://doi.org/10.1016/j.ijer.2026.102963 Tsai, H

International Journal of Educational Research, 137, 102963. https://doi.org/10.1016/j.ijer.2026.102963 Tisdell, E. J. (2025). Qualitative research: A guide to design and implementation (Fifth edition). Jossey-Bass. Wang, C., Du, Y., & Zou, B. (2026). Learners’ acceptance and use of multimodal artificial intelligence (AI)‐generated content in AI‐mediated i...

work page doi:10.1016/j.ijer.2026.102963 2026
[15]

https://doi.org/10.3390/systems13040302 Zheng, W., Ma, Z., Sun, J., Wu, Q., & Hu, Y . (2025). Exploring factors influencing continuance intention of pre-service teachers in using generative artificial intelligence. International Journal of Human–Computer Interaction, 41(16), 10325–10338. https://doi.org/10.1080/10447318.2024.2433300 Zhuang, J., & Sun, H. ...

work page doi:10.3390/systems13040302 2025

[1] [1]

These tools can support lesson planning, instructional design, explanations, feedback, and other teaching tasks (C

Introduction Generative AI has rapidly become important in education, especially after the release of large language models such as ChatGPT. These tools can support lesson planning, instructional design, explanations, feedback, and other teaching tasks (C. Wang et al., 2026; Lee et al., 2026; Du et al., 2025; Aljohani, 2026). However, concerns remain abou...

2026

[2] [2]

smart education

Literature Review 2.1 Generative AI in Education Generative artificial intelligence has rapidly emerged as a transformative technology in education following the public release of large language models such as ChatGPT (C. Wang et al., 2026). These systems can generate human-like text, explanations, feedback, and instructional materials, enabling new forms...

2026

[3] [3]

Theoretical Framework This study integrates the Expectation-Confirmation Model and Institutional Theory to examine teachers’ continuance intention to use generative AI in the Chinese educational context. The Expectation-Confirmation Model explains users’ post-adoption behaviour based on their experiences with a technology (Bhattacherjee, 2001), whereas In...

2001

[4] [4]

According to this model, continuance intention is influenced by users’ confirmation of expectations, perceived usefulness, and satisfaction

Research Questions and Hypotheses 4.1 The Impact of Confirmation, Perceived Usefulness, and Satisfaction Based on the conceptual model, the first research question (RQ1) asks: How do confirmation, perceived usefulness, and satisfaction influence Chinese teachers’ continuance intention to use generative AI? To address this question, this study draws on the...

2001

[5] [5]

Methods 5.1 Research Design This study used a sequential explanatory mixed-methods design, with quantitative data collection and analysis followed by qualitative inquiry. First, a questionnaire survey examined relationships among constructs in the conceptual model, including Expectation–Confirmation Model factors—confirmation, perceived usefulness, and sa...

2025

[6] [6]

derived from the Expectation-Confirmation Model, and coercive pressure (Sukoco et al., 2022), normative pressure (Hom et al., 2025), and mimetic pressure (Latif et al.,

2022

[7] [7]

Measurement items were adapted from established scales in prior research and slightly modified to fit the context of teachers’ use of generative AI in teaching (Y

derived from Institutional Theory. Measurement items were adapted from established scales in prior research and slightly modified to fit the context of teachers’ use of generative AI in teaching (Y . Du, 2024). The questionnaire was administered in Chinese to ensure participants’ comprehension, and a translation and back-translation procedure was conducte...

2024

[8] [8]

Discussion 7.1 The Impact of Confirmation, Perceived Usefulness, and Satisfaction The findings reinforce the Expectation–Confirmation Model by showing that confirmation, perceived usefulness, and satisfaction are key determinants of teachers’ continuance intention to use generative AI. Consistent with prior continuance research (Bhattacherjee, 2001), conf...

2001

[9] [9]

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

Conclusion This study examined Chinese teachers’ continuance intention to use generative AI by integrating the Expectation–Confirmation Model with Institutional Theory. The findings show that continued use is shaped by both experiential evaluations and institutional influences. Confirmation, perceived usefulness, and satisfaction were important predictors...

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

[10] [10]

https://doi.org/10.2307/3250921 Brislin, R. W. (1970). Back-translation for cross-cultural research. Journal of Cross-Cultural Psychology, 1(3), 185–216. https://doi.org/10.1177/135910457000100301 Chai, C. S., Liang, S., & Wang, X. (2024). A survey study of Chinese teachers’ continuous intentions to teach artificial intelligence. Education and Information...

work page doi:10.2307/3250921 1970

[11] [11]

https://doi.org/10.3390/su17136082 Kim, J. (2024). Types of teacher-AI collaboration in K-12 Classroom instruction: Chinese teachers’ perspective. Education and Information Technologies, 29(13), 17433–17465. https://doi.org/10.1007/s10639-024-12523-3 Kline, R. B. (2023). Principles and practice of structural equation modeling (Fifth edition). The Guilford...

work page doi:10.3390/su17136082 2024

[12] [12]

https://doi.org/10.3390/su12114506 Lee, S., Choe, H., Zou, D., & Jeon, J. (2026). Generative AI (GenAI) in the language classroom: A systematic review. Interactive Learning Environments, 34(1), 335–359. https://doi.org/10.1080/10494820.2025.2498537 Lu, H.-P., & Wang, J.-C. (2023). Exploring the effects of sudden institutional coercive pressure on digital ...

work page doi:10.3390/su12114506 2026

[13] [13]

https://doi.org/10.3390/educsci12120858 Obeid, A., Ibrahim, R., & Fadhil, A. (2024). Extended model of expectation confirmation model to examine users’ continuous intention toward the utilization of E-learning platforms. IEEE Access, 12, 40752–40764. https://doi.org/10.1109/ACCESS.2024.3373190 Ode, E., Nana, R., Boro, I. O., & Ikyanyon, D. N. (2025). A cr...

work page doi:10.3390/educsci12120858 2024

[14] [14]

https://doi.org/10.1016/j.ijer.2026.102963 Tsai, H

International Journal of Educational Research, 137, 102963. https://doi.org/10.1016/j.ijer.2026.102963 Tisdell, E. J. (2025). Qualitative research: A guide to design and implementation (Fifth edition). Jossey-Bass. Wang, C., Du, Y., & Zou, B. (2026). Learners’ acceptance and use of multimodal artificial intelligence (AI)‐generated content in AI‐mediated i...

work page doi:10.1016/j.ijer.2026.102963 2026

[15] [15]

https://doi.org/10.3390/systems13040302 Zheng, W., Ma, Z., Sun, J., Wu, Q., & Hu, Y . (2025). Exploring factors influencing continuance intention of pre-service teachers in using generative artificial intelligence. International Journal of Human–Computer Interaction, 41(16), 10325–10338. https://doi.org/10.1080/10447318.2024.2433300 Zhuang, J., & Sun, H. ...

work page doi:10.3390/systems13040302 2025