Factors influencing Chinese pre-service teachers' behavioral intention and use behavior to adopt VR training system: based on the UTAUT2 model

The study investigates how and why Chinese pre-service teachers intend to adopt and actually use a virtual reality (VR) training system for professional skills development. Against the backdrop of increasing integration of VR in education and China’s emphasis on VR competencies in teacher training, the study adopts and extends UTAUT2 to identify determinants of behavioral intention (BI) and use behavior (USE). Research questions: (1) Which factors influence pre-service teachers’ behavioral intention to adopt a VR training system and what is their relative influence? (2) Which factors influence pre-service teachers’ use behavior of a VR training system and what is their relative influence? The authors propose hypotheses linking performance expectancy, effort expectancy, social influence, facilitating conditions, hedonic motivation, habit, perceived risk, and self-efficacy to BI and/or USE, and test these relationships using PLS-SEM.

The literature review outlines VR as an immersive, multisensory human–computer interface that supports authentic simulations, enhancing learners’ comprehension, motivation, and performance across disciplines. In pre-service teacher education, VR can provide safe, realistic practice for classroom management, communication, reflection, and skills development, mitigating stress and enabling error-tolerant rehearsal. Development efforts include project-based VR curricula, spherical view systems, and broader frameworks for teacher VR training; however, scholars highlight the need for usability and behavior testing to understand user acceptance and guide targeted improvements. UTAUT2 is presented as a comprehensive technology acceptance framework extending UTAUT with hedonic motivation, price value, and habit, plus moderators (age, gender, experience). Prior education studies have used UTAUT2 to examine adoption of MOOC, e-learning, and classroom platforms; in VR contexts, it has revealed both adoption drivers and potential barriers. Building on this, the study tailors UTAUT2 to the VR training context for pre-service teachers by adding perceived risk and self-efficacy, and excluding price value and moderators not pertinent to this sample (age, gender, experience).

Design: Cross-sectional explanatory study using partial least squares structural equation modeling (PLS-SEM). Sampling and participants: Non-probability sampling of undergraduate and postgraduate pre-service teachers at a normal college within a Chinese university. In March 2023 the college introduced a desktop VR training system used in professional development classes. An online questionnaire (Wenjuanxing) was distributed in May 2023; 302 responses were received by end of June; 24 were invalid due to reverse-scoring checks in perceived risk, yielding 278 valid responses (69.42% female; 67.63% undergraduates; majors primarily primary education 53.60% and preschool education 34.17%). VR system: Desktop VR with modules for teacher’s moral practice, teaching practice, comprehensive education ability, self-development ability, and classroom emergency skills; students selected modules for targeted simulation training. Measures: Constructs adapted from UTAUT2 (PE, EE, SI, FC, HM, HA, BI, USE) and added variables perceived risk (PR) and self-efficacy (SE) from established scales. Items measured on seven-point Likert scales (1 strongly disagree to 7 strongly agree). Example items include PE (use improves teaching practice ability), EE (easy to learn/use), SI (important others think I should use it), FC (school provides equipment/training/help), HM (enjoyable/fun), HA (system use has become habitual), PR (effectiveness/health/privacy risks), SE (confidence, determination to accomplish tasks), BI (intend/recommend/plan to use), USE (self-reported capability to use system across modules). Data analysis: PLS-SEM with SmartPLS 4. Measurement model assessed via reliability (Cronbach’s alpha and composite reliability >0.6), convergent validity (loadings and AVE >0.5), and discriminant validity (Fornell–Larcker criterion). Structural model tested via bootstrapping with 5000 resamples; path coefficients, significance (t, p), and R² for BI and USE were reported.

Measurement model showed acceptable reliability and validity (all loadings and AVE ≥ 0.5; Cronbach’s alpha and composite reliability ≥ 0.6; Fornell–Larcker discriminant validity satisfied). Structural model results (n=278): Predictors of Behavioral Intention (BI): SE β=0.281, p<0.001; EE β=0.236, p<0.001; SI β=0.142, p=0.010; PE β=0.139, p=0.020; FC β=0.131, p=0.028; HM β=0.077, p=0.047. Non-significant to BI: HA β=0.026, p=0.653; PR β=-0.015, p=0.646. Predictors of Use Behavior (USE): BI β=0.545, p<0.001; FC β=0.182, p=0.012; HA β=0.171, p=0.005. Explanatory power: R² for BI=0.815; R² for USE=0.712. Relative influence ranking: For BI, SE > EE > SI > PE > FC > HM; for USE, BI > FC > HA.

Findings address the research questions by identifying and ranking determinants of intention and use of a VR training system among pre-service teachers. The strongest driver of intention is self-efficacy, highlighting the importance of confidence, determination, and satisfaction in using VR for accomplishing tasks. Ease of use (effort expectancy) also substantially increases intention, consistent with the need for intuitive interfaces and support. Social influence, performance expectancy, and facilitating conditions reflect utilitarian and social facets: favorable perceptions of benefits, encouragement from important others, and resource/support availability all raise intention. Hedonic motivation has a positive yet relatively small effect, implying that in professional training contexts, learning outcomes and skill acquisition supersede enjoyment. Habit does not raise intention but directly increases actual use, indicating that repeated exposure and routines translate to behavior without necessarily elevating intention. Perceived risk was not a significant inhibitor of intention in this educational setting, possibly due to institutional context and task focus, though effectiveness, health, and privacy still warrant attention. Practical implications include enhancing self-efficacy via guidance and feedback, designing easy-to-use interfaces, ensuring equipment and support, leveraging peer/teacher advocacy, and addressing privacy and health safeguards. The strong BI-to-USE path underscores focusing on intention-building strategies to drive continued use.

The study proposes and validates an extended UTAUT2-based model explaining pre-service teachers’ behavioral intention and use behavior toward a VR training system. The model, which adds self-efficacy and perceived risk and removes price value and moderators not pertinent to the sample, shows high explanatory power (R²=0.815 for BI; 0.712 for USE). Intention is significantly driven (from strongest to weakest) by self-efficacy, effort expectancy, social influence, performance expectancy, facilitating conditions, and hedonic motivation; actual use is significantly driven by behavioral intention, facilitating conditions, and habit. The work extends UTAUT2 application in VR education, offering theoretical insight into intention and behavior and practical guidance for institutions and developers. Future research should broaden samples, consider moderators, and integrate qualitative methods to deepen understanding and generalizability.

The study excludes moderators (age, gender, experience) due to the sample context and removes price value; thus, potential moderating effects remain untested. The sample is limited to pre-service teachers at a single Chinese university, constraining generalizability. The methodology is primarily quantitative and cross-sectional, not capturing rich individual differences or longitudinal effects. Future work should expand sample size across multiple institutions and regions (nationally and internationally), examine additional moderators (gender, experience, culture, interest), and incorporate qualitative approaches to triangulate and deepen findings.