The study was grounded in Conservation of Resources theory, which posits that individuals strive to obtain, retain, and protect resources, and that resource gains tend to unfold in caravans or spirals. In this framework, perceived AI acceptance is treated as a cognitive appraisal of whether a new technology is likely to add to or drain personal resources. Digital well-being is conceptualized as a personal-condition resource, reflecting a state of digital skills and low digital strain, while innovative work behavior is viewed as discretionary resource investment. Occupational well-being serves as the affective outcome, representing the pattern of pleasurable and activated affective experiences derived from the work role. This theoretical lens allows for the examination of how a favorable technological appraisal might be consistent with a sequence of resource accumulation, although the cross-sectional nature of the data means these are statistical associations rather than proven causal pathways.

Participants were recruited through online convenience and snowball sampling between June and August 2026. The sample included 807 valid responses after excluding questionnaires from non-physical education teachers, those with completion times outside platform limits, and those with internally inconsistent responses. The study utilized self-report scales to measure perceived AI acceptance, digital well-being, innovative work behavior, and occupational well-being. Perceived AI acceptance was adapted from the Technology Acceptance Model, focusing on perceived usefulness and ease of use. Digital well-being was measured using a researcher-developed instrument based on a published theoretical framework, while innovative work behavior was assessed using a general scale that captures idea generation, promotion, and realization in one’s work role. Occupational well-being was measured using the Job-related Affective Well-being Scale, which evaluates job-related affective states along pleasure and arousal axes.

The results indicated that perceived AI acceptance was significantly and positively associated with occupational well-being, with a total effect coefficient of 0.181. This association was not solely direct; a substantial portion of the effect was mediated by digital well-being and innovative work behavior. Specifically, the indirect effect through digital well-being alone accounted for 23.26% of the total effect, while the indirect effect through innovative work behavior alone accounted for 21.93%. A serial indirect effect, where perceived AI acceptance was associated with digital well-being, which in turn was associated with innovative work behavior, and finally with occupational well-being, accounted for an additional 11.66% of the total effect. The total indirect effect was 0.103, representing 56.91% of the total association, while the direct effect accounted for the remaining 43.09%. All bootstrap confidence intervals for the indirect effects excluded zero, indicating statistically reliable associations.

The measurement model was evaluated using confirmatory factor analysis, which demonstrated good fit to the data. The four-factor model, comprising perceived AI acceptance, digital well-being, innovative work behavior, and occupational well-being, showed standardized factor loadings ranging from 0.703 to 0.817. Composite reliability and average variance extracted values met recommended thresholds for convergent validity at the dimension level, although the average variance extracted for innovative work behavior was only marginally above the criterion of 0.50. Discriminant validity was supported, as the square root of the average variance extracted for each construct exceeded its correlations with other constructs. These psychometric properties provide a foundation for interpreting the structural relationships among the variables, although the use of dimension-level parcels in the analysis limits the generalizability of the measurement model to item-level structures.

Robustness analyses were conducted to evaluate the dependence of the findings on the proposed ordering of mediators and the scoring of occupational well-being. When the mediators were reversed in the serial mediation model, the serial indirect effect remained statistically reliable but was smaller than in the proposed order. This suggests that the data cannot definitively discriminate between the two orderings, and the proposed sequence rests on theoretical grounds rather than empirical temporal evidence. Additionally, the model was re-estimated with occupational well-being computed from a 20-item subset of the Job-related Affective Well-being Scale, yielding virtually identical estimates to the primary analysis. This indicates that the findings are not sensitive to the specific scoring method used for the outcome variable, enhancing the stability of the observed associations.

The study highlights the importance of digital well-being as a potential mechanism linking technology acceptance to occupational well-being. Teachers who perceive AI tools as useful and easy to use may experience lower anticipated resource costs in digital engagement, leading to higher digital well-being. This state of digital resource sufficiency may, in turn, be associated with greater capacity for innovative work behavior, as individuals with fewer digital strains may have more cognitive and motivational resources available for discretionary, proactive behaviors. The successful enactment of innovative work behavior may then be linked to further resource gains, such as competence experiences and positive affect, which contribute to occupational well-being. This pattern is consistent with a resource-based interpretation of how favorable technological appraisal may be linked to occupational adaptation, although the cross-sectional design precludes causal inference.

Several limitations of the study should be noted. The cross-sectional design means that the observed associations cannot be interpreted as causal, and reverse or reciprocal relationships cannot be excluded. All data were collected through self-report measures, which may be susceptible to common method bias and social desirability. The sample was obtained through non-probability sampling, which may limit the representativeness of the findings to university physical education teachers in China. Additionally, the study did not control for potential confounders such as institutional support, professional identity, or workload, which could plausibly shape both perceived AI acceptance and its downstream associations. The measurement of innovative work behavior was general and not specific to teaching or pedagogical innovation, so conclusions about instructional innovation should not be drawn. Future research should employ longitudinal designs and objective measures to test the directionality of these associations and to refine the measurement of the constructs involved.

Despite these limitations, the findings offer valuable insights for higher-education institutions seeking to support physical education teachers during AI integration. The results suggest that supporting teachers’ psychological adaptation to AI may involve more than technology provision and may merit attention to teachers’ digital well-being and their opportunities for innovative work behavior. Institutions might consider exploring structured AI-literacy training aimed at accurate, confidence-building appraisals of AI tools, as well as professional development programs that incorporate guidance on regulating digital workload and information demands. Organizational conditions, such as protected time and peer exchange platforms, could also allow teachers to put favorable technology appraisals into innovative work practices. These suggestions are tentative and should be tested in longitudinal and intervention studies to determine their effectiveness in enhancing occupational well-being.

In conclusion, this study provides evidence that perceived AI acceptance is associated with occupational well-being among university physical education teachers, both directly and indirectly through digital well-being and innovative work behavior. The pattern of results is consistent with a Conservation of Resources-based sequence, although the cross-sectional design means that this sequence is theoretically proposed rather than empirically established. The findings underscore the complex interplay between technology appraisal, digital self-regulation, and professional behavior in shaping teacher well-being. As AI continues to diffuse into higher education, understanding these associations will be crucial for developing strategies that support the psychological adaptation of educators and enhance the sustainability of the educational workforce. Further research is needed to clarify the causal mechanisms and to explore the boundary conditions of these associations in different cultural and disciplinary contexts.

Subject of Research: Perceived AI acceptance and occupational well-being among university physical education teachers: a cross-sectional serial mediation model of digital well-being and innovative work behavior

Article Title: Perceived AI acceptance and occupational well-being among university physical education teachers: a cross-sectional serial mediation model of digital well-being and innovative work behavior

Article References: Miao, J. (2026). Perceived AI acceptance and occupational well-being among university physical education teachers: a cross-sectional serial mediation model of digital well-being and innovative work behavior. Frontiers in Psychology, 17, Article 1999813. https://doi.org/10.3389/fpsyg.2026.1999813

Image Credits: AI Generated

DOI: 10.3389/fpsyg.2026.1999813

Keywords: Perceived, acceptance, occupational, well-being, university, physical, education, teachers, cross-sectional, serial, mediation, model