The research team, based at Chengdu University of Technology and Sichuan University's West China Hospital, used a quantitative quasi-experimental design with a pre-test and three parallel instructional conditions. Students were enrolled in two university-wide general-education courses, Outline of Chinese Modern History and History of the People's Republic of China, both delivered as smart courses through the Xueyin Online platform. Nine intact classes were selected — three for each instructional modality — and, crucially, the instructors rather than the students were randomly assigned to the conditions.

A one-way ANOVA of pre-test engagement confirmed that the three groups were statistically equivalent at baseline across behavioural, affective, cognitive and agentic engagement, with all probability values well above conventional significance thresholds. After two-stage screening that removed missing, invalid, duplicate and extreme-value responses, the final sample comprised 394 students in face-to-face classes, 256 in online classes and 293 in blended classes, with demographics spanning urban and rural household registration, Han and ethnic minority students, and science, engineering, humanities and social science majors. Self-determination theory provides the study's conceptual backbone, and its technical vocabulary matters for understanding the results.

The theory treats extrinsic motivation not as a single undifferentiated force but as a continuum of four regulations that vary in their degree of internalisation. External regulation, the least autonomous form, describes behaviour driven by external contingencies such as rewards or punishments; introjected regulation is internal but still controlling, because students act to avoid guilt or protect contingent self-esteem; identified regulation reflects personally endorsed goals; and integrated regulation, the most autonomous form of extrinsic motivation, is congruent with the self. Sustained engagement, in this framework, depends on the satisfaction of three innate psychological needs — autonomy, competence and relatedness.

The theory's history is contested terrain: meta-analytic evidence shows that expected, tangible rewards tend to undermine intrinsic motivation, whereas recognition and non-controlling feedback tend to sustain it, indicating that rewards either frustrate or satisfy psychological needs depending on how they are delivered. Financial rewards relate negatively to intrinsic task motivation while recognition and social rewards relate positively, precisely because the latter convey competence and acknowledge personal value. Whether this mechanism operates similarly across instructional modalities is not self-evident, and the study's design reflects that question directly.

In face-to-face settings, dense social presence readily supports relatedness and competence through immediate teacher feedback and peer interaction. In fully online settings, the same feedback must be deliberately engineered through digital tools, and relatedness is more easily thwarted. In blended settings, students move between the two contexts, so need satisfaction depends on how seamlessly teacher and digital support are integrated across modes.

The three conditions in this study differed markedly in pedagogy: the face-to-face group implemented a teacher-centred, lecture-dominant model with printed textbooks and whiteboards; the online group used the Xueyin Online MOOC platform with AI-enhanced pedagogy, adaptive personalisation and cloud-based multimedia; and the blended group combined in-person sessions with online activities through Small Private Online Courses on the Chaoxing Learning Passport mobile platform, emphasising proactive preparation, self-regulation and instructor facilitation. The questionnaire combined the Self-Regulation Questionnaire–Academic, the Intrinsic Motivation Inventory and the Basic Psychological Need Satisfaction Scale, operationalising 11 latent indicators across three constructs: extrinsic motivation with four regulatory dimensions, basic psychological need satisfaction with autonomy, competence and relatedness, and student engagement with behavioural, affective, cognitive and agentic dimensions. Each of the 11 subscales contained five Likert-type items rated on a five-point scale from strongly disagree to strongly agree.

Illustrative items ranged from the controlling end — "I participate in discussion boards to avoid losing points or falling behind" — to the fully internalised end — "I apply myself to this course because what I learn has become part of who I am." Need satisfaction items asked whether students had meaningful choices in completing assignments, felt competent in self-directed learning, and found their ideas valued during discussions. Content validity was established through expert appraisal by nine practising educators, and cognitive interviews verified item clarity. The psychometric results were strong across the board.

Standardised factor loadings ranged from 0.834 to 0.895 for extrinsic motivation, from 0.715 to 0.754 for need satisfaction, and from 0.831 to 0.865 for student engagement. Average variance extracted exceeded 0.7 for every latent variable, composite reliability ranged from 0.914 to 0.925, and Cronbach's alpha coefficients ranged from 0.869 to 0.891, all exceeding the 0.80 threshold for good reliability. Absolute skewness and kurtosis values of all observed variables fell within accepted thresholds for normality, with a maximum absolute skewness of 0.936 and a maximum absolute kurtosis of 1.994.

Age, household registration, gender, ethnicity, only-child status, academic major, annual family income and region were included as covariates, and all three hypotheses in the structural equation model were supported regardless of whether these covariates were included. Because the distribution of indirect effects is typically non-normal, mediation was tested with 5,000 bootstrap resamples, reporting bias-corrected 95 percent confidence intervals for each group. Measurement invariance was tested in three stages — configural, metric and scalar — and all three were established, with the metric model producing the lowest Akaike information criterion and therefore preferred for the cross-group comparisons.

Nested-model comparisons then formally tested whether structural paths differed across groups: constraining the path from extrinsic motivation to need satisfaction and the path from need satisfaction to engagement to equality did not significantly worsen fit, indicating invariance across modalities, whereas constraining the direct path did significantly worsen fit, confirming that this path differed across groups. The headline results are technically precise and theoretically consequential. Extrinsic motivation positively and significantly predicted need satisfaction in all three groups, with standardised coefficients of 0.569 in the face-to-face group, 0.577 in the blended group and 0.561 in the online group, explaining 31.5 to 33.3 percent of the variance in need satisfaction.

Need satisfaction in turn positively and significantly predicted student engagement, with coefficients of 0.549, 0.545 and 0.535 respectively, and the two predictors jointly explained 36.2 to 40.1 percent of the variance in engagement. The bootstrap mediation results revealed the study's central pattern: the indirect effect accounted for 65.41 percent of the total effect in the face-to-face group, 71.09 percent in the online group and 54.51 percent in the blended group, with all confidence intervals excluding zero. The direct effect was significant in the blended group at 0.262 and in the face-to-face group at 0.165, but non-significant in the online group at 0.122 — meaning that in fully online instruction, extrinsic motivation was associated with engagement only through the satisfaction of autonomy, competence and relatedness.

The authors interpret this pattern as a qualification of the apparent universality of the self-determination sequence: the mediating mechanism holds across modalities, but its magnitude and the availability of a direct motivational route depend on the social presence the modality affords. Online learning structurally weakens social presence and relatedness, so needs must be deliberately engineered through digital design. The blended group produced the largest indirect effect at 0.314, supporting the argument that integrating face-to-face relatedness with flexible, self-regulated online learning can strengthen the internalisation mechanism — flexibility benefits students only when accompanied by coherent pedagogy rather than mere technological access.

The positive association between extrinsic motivation and need satisfaction also challenges the assumption that extrinsic motivation is uniformly detrimental: in these general-education courses, the extrinsic regulations measured were largely internalised rather than experienced as purely coercive, and therefore functioned as a resource for autonomy, competence and relatedness. Practically, the findings offer concrete guidance for higher-education teaching in an era of AI-enhanced courses. Instructors and course designers should use rewards, grades and feedback informationally rather than controllingly, pairing them with meaningful choice, appropriately challenging tasks and opportunities for relatedness.

Online and blended courses merit particular attention: because extrinsic motivation was not directly associated with engagement in fully online settings, digital platforms should deliberately engineer interaction, timely feedback and social presence, using AI tools to scaffold competence and autonomy while preserving the human support that satisfies relatedness. The authors acknowledge limitations: the data came from undergraduates at one university using convenience and quota sampling, the structural relationships rest on self-reported single-source survey data vulnerable to common-method bias, and the study did not examine specific AI features or teacher behaviours in detail. Longitudinal or fully randomised designs, objective trace data and cross-institutional samples would better establish temporal and causal order.

Still, the core message is clear: engagement is not secured by technology or incentives alone, but by the degree to which any learning environment — classroom, MOOC or SPOC — makes students feel volitional, effective and connected.

Subject of Research: Extrinsic motivation and student engagement in higher education

Article Title: Extrinsic motivation and student engagement in higher education: the mediating role of basic psychological need satisfaction across instructional modalities

Article References: Xie, X., Hu, Y., & Wang, S. (2026). Extrinsic motivation and student engagement in higher education: the mediating role of basic psychological need satisfaction across instructional modalities. Frontiers in Psychology, 17, Article 1962667. https://doi.org/10.3389/fpsyg.2026.1962667

Image Credits: AI Generated

DOI: 10.3389/fpsyg.2026.1962667

Keywords: Extrinsic, motivation, student, engagement, higher, education, mediating, role, basic, psychological, need, satisfaction