Review Reports - Academic Self-Efficacy and Academic Burnout Among College Students: The Chain-Mediating Roles of Self-Control and Learning Engagement

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Dear Authors,

Thank you for the opportunity to review your manuscript.

Academic burnout remains a significant concern for universities worldwide, and understanding the psychological mechanisms that may protect students from burnout is an important research endeavor. The study benefits from a large sample size, a clearly articulated conceptual model, and a coherent analytical strategy. The manuscript is generally well organized, the hypotheses are logically derived from the literature, and the reported findings are consistent with existing theoretical expectations.

At the same time, I believe that several aspects of the manuscript require further development before publication.

1.-Although the proposed serial mediation model is coherent and empirically supported, the manuscript does not sufficiently clarify how it advances current knowledge beyond previous studies, for example, relationships among academic self-efficacy, self-control, learning engagement, and academic burnout have already been extensively documented, and several recent studies have examined comparable mediation frameworks. That´s way I think that you should more explicitly on the identification the specific theoretical gap addressed by the present study and explain how the proposed model contributes new insights rather than confirming already established relationships. may be, -considering the variables included in sample-, a zoom in with new analysis on the classification sample could give more density to results.

2.- consisdering the above mentioned, the use of composite scores allows examination of average relationships among constructs but may conceal important heterogeneity within the sample. Given both the sample size and the multidimensional nature of the instruments employed, the manuscript could have explored more differentiated approaches, such as subgroup analyses, moderation effects, latent profiles, or dimension-level relationships among the constructs. Although such analyses may be beyond the immediate scope of the study, acknowledging these possibilities would help position the manuscript within current efforts to develop more nuanced models of student functioning and academic burnout.

3.- While the proposed model is theoretically plausible, the data support associations rather than causal inferences. I encourage the authors to moderate causal claims throughout the manuscript and more carefully distinguish between theoretical mechanisms and empirical evidence.

4.- I would also encourage the authors to strengthen the abstract, it currently provides very limited quantitative information regarding the empirical findings. Including key statistical results, such as major path coefficients, indirect effects, or proportions of explained effects, would allow readers to better assess the magnitude and practical relevance of the reported relationships.

5.- I think that some methodological issues also deserve attention, for example: The manuscript reports the use of Likert-type scales for all principal variables, yet reliability is assessed exclusively through Cronbach’s alpha and the confirmatory factor analysis procedures are not described in sufficient detail. Because Likert responses are ordinal in nature, the authors should clarify whether ordinal reliability estimates (e.g., ordinal alpha or McDonald’s omega) were considered and whether the CFA was estimated using procedures appropriate for ordinal indicators, such as polychoric correlation matrices and robust estimators. A justification of the psychometric decisions adopted would strengthen the methodological rigor of the study.

Similarly, additional information should be provided regarding the adaptation and refinement of the measurement instruments. The manuscript indicates that several items from the learning engagement and academic burnout scales were removed during the purification process, yet the specific items removed, the criteria for deletion, and the psychometric consequences of these decisions are not sufficiently documented. Greater transparency would enhance confidence in the measurement model.

Although the authors report evidence of discriminant validity using the Fornell-Larcker criterion, I remain concerned about the very strong association observed between learning engagement and academic burnout (r = -.76). Such a correlation raises questions regarding the empirical distinctiveness of these constructs. Additional evidence, such as HTMT ratios, comparisons with alternative measurement models, or a more detailed discussion of potential conceptual overlap, would strengthen the validity argument.

6.-While the findings are clearly summarized, the discussion would benefit from deeper engagement with previous literature, a more detailed interpretation of effect sizes, and greater consideration of the broader implications of the results. In particular, the strong relationship between engagement and burnout deserves further theoretical reflection.

7.-The assessment of common method bias relies exclusively on Harman’s single-factor test, a procedure that has well-known limitations in contemporary methodological literature. The authors may wish to justify this choice or consider complementary approaches.

Finally, the manuscript addresses a relevant topic and presents findings that are potentially valuable for understanding academic burnout among university students, Addressing these concerns would considerably improve the scientific contribution and impact of the study.

Best Wishes

Author Response

Comment 1:

Although the proposed serial mediation model is coherent and empirically supported, the manuscript does not sufficiently clarify how it advances current knowledge beyond previous studies, for example, relationships among academic self-efficacy, self-control, learning engagement, and academic burnout have already been extensively documented, and several recent studies have examined comparable mediation frameworks. That´s way I think that you should more explicitly on the identification the specific theoretical gap addressed by the present study and explain how the proposed model contributes new insights rather than confirming already established relationships. may be, -considering the variables included in sample-, a zoom in with new analysis on the classification sample could give more density to results.

Reply:

We appreciate this constructive suggestion. We agree that the original version did not sufficiently distinguish the contribution of the present model from prior studies that had already examined academic self-efficacy, self-regulation, learning engagement, and burnout. In response, we substantially revised the Introduction to identify three more specific gaps: the predominance of pairwise or parallel-mediation approaches, the insufficient theoretical attention to the sequence from efficacy belief to self-control and then to learning engagement, and the limited consideration of sample heterogeneity.

We also clarified that the novelty of the study does not lie in treating the four constructs as entirely new variables, but in specifying a theoretically ordered process and comparing the relative magnitude of the direct, single-mediator, and serial-mediator pathways. In addition, to give more density to the results based on sample classification variables, we added supplementary sample classification information and a robustness analysis controlling for demographic characteristics.

Action:

Location in the revised manuscript: Introduction; Section 3.1; Appendix A, Tables S1 and S5

“This study offers three contributions. First, it clarifies that the novelty of the model does not lie in treating academic self-efficacy, self-control, learning engagement, and academic burnout as entirely new variables. Rather, the contribution lies in specifying a theoretically ordered process in which academic self-efficacy is linked to burnout-related risk through a staged pathway from cognitive belief to regulatory capacity and then to learning participation.”

“Respondents were also classified by university type, grade level, disciplinary background, place of origin, and student leadership experience. The classification results are summarized in Table S1.”

Comment 2:

consisdering the above mentioned, the use of composite scores allows examination of average relationships among constructs but may conceal important heterogeneity within the sample. Given both the sample size and the multidimensional nature of the instruments employed, the manuscript could have explored more differentiated approaches, such as subgroup analyses, moderation effects, latent profiles, or dimension-level relationships among the constructs. Although such analyses may be beyond the immediate scope of the study, acknowledging these possibilities would help position the manuscript within current efforts to develop more nuanced models of student functioning and academic burnout.

Reply:

Thank you for pointing out the potential limitation of composite-score analysis. We have clarified that our primary model was designed to test the theoretically ordered average process, but that such an approach may obscure heterogeneity across student groups and across dimensions of the constructs.

To respond to this concern, we added detailed sample classification information and a supplementary robustness analysis with demographic controls. We also expanded the limitations and future research section to explicitly recommend subgroup analysis, moderation models, latent profile analysis, and dimension-level pathways as important future extensions of the present chain mediation model.

Action:

Location in the revised manuscript: Introduction; Section 3.3; Section 5; Appendix A, Tables S1 and S5

“The present study therefore positions its model as an average-process explanation and explicitly acknowledges the need for more differentiated subgroup, moderation, latent-profile, and dimension-level analyses in future work.”

“Demographic variables were handled as supplementary controls rather than as core theoretical variables. The primary model was estimated without controls in order to test the proposed Social Cognitive Theory pathway as parsimoniously as possible.”

“Future studies could address these limitations by using longitudinal, cross-lagged, or experimental designs to evaluate temporal ordering more rigorously. Researchers should also examine subgroup, moderation, latent-profile, and dimension-level analyses to clarify whether the proposed pathway operates differently across student populations.”

Comment 3:

While the proposed model is theoretically plausible, the data support associations rather than causal inferences. I encourage the authors to moderate causal claims throughout the manuscript and more carefully distinguish between theoretical mechanisms and empirical evidence.

Reply:

We fully agree with this important methodological concern. Because the study used cross-sectional self-report data, the revised manuscript avoids causal wording wherever possible. We replaced expressions such as “effect,” “influence,” “reduce,” and “predicts” with more cautious terms such as “association,” “is associated with,” “is linked to,” “statistically mediates,” and “theoretically informed pathway.”

We also added explicit statements in the Abstract, Methods, and Limitations sections to distinguish the theoretical logic of the proposed mechanism from the empirical evidence provided by the present data.

Action:

Location in the revised manuscript: Abstract; Section 3.3; Section 5

“Given the cross-sectional and self-report design, the results should be interpreted as evidence of theoretically informed associations rather than causal effects.”

“Because the data were cross-sectional, the mediation model is interpreted as a theoretically informed statistical mediation model rather than as evidence of temporal or causal mediation.”

“Although the model is theoretically grounded in Social Cognitive Theory, the empirical evidence supports associations among variables rather than confirmed causal mechanisms.”

Comment 4:

I would also encourage the authors to strengthen the abstract, it currently provides very limited quantitative information regarding the empirical findings. Including key statistical results, such as major path coefficients, indirect effects, or proportions of explained effects, would allow readers to better assess the magnitude and practical relevance of the reported relationships.

Reply:

Thank you for this useful suggestion. We revised the Abstract to provide a more informative summary of the empirical findings. The revised abstract now reports the sample size, measurement-model fit indices, major standardized path coefficients, bootstrap indirect associations, confidence intervals, and the proportion of the total association accounted for by indirect pathways.

These additions allow readers to better evaluate both the statistical magnitude and practical relevance of the reported relationships before reading the full Results section.

Action:

Location in the revised manuscript: Abstract

“The measurement model showed acceptable fit (χ²/df = 2.84, CFI = 0.958, TLI = 0.947, RMSEA = 0.036, SRMR = 0.031).”

“Regression-based path analysis indicated that academic self-efficacy was positively associated with self-control (β = 0.48, p < .001) and learning engagement (β = 0.33, p < .001), and negatively associated with academic burnout (β = -0.24, p < .001).”

“Bootstrap analyses showed significant indirect associations through self-control (Estimate = -0.096, 95% CI [-0.135, -0.062]), learning engagement (Estimate = -0.165, 95% CI [-0.218, -0.117]), and the serial pathway through self-control and learning engagement (Estimate = -0.074, 95% CI [-0.104, -0.049]). The total indirect association accounted for 58.26% of the total association.”

Comment 5:

I think that some methodological issues also deserve attention. The manuscript reports the use of Likert-type scales for all principal variables, yet reliability is assessed exclusively through Cronbach’s alpha and the confirmatory factor analysis procedures are not described in sufficient detail. Because Likert responses are ordinal in nature, the authors should clarify whether ordinal reliability estimates (e.g., ordinal alpha or McDonald’s omega) were considered and whether the CFA was estimated using procedures appropriate for ordinal indicators, such as polychoric correlation matrices and robust estimators. A justification of the psychometric decisions adopted would strengthen the methodological rigor of the study. Similarly, additional information should be provided regarding the adaptation and refinement of the measurement instruments. The manuscript indicates that several items from the learning engagement and academic burnout scales were removed during the purification process, yet the specific items removed, the criteria for deletion, and the psychometric consequences of these decisions are not sufficiently documented. Greater transparency would enhance confidence in the measurement model. Although the authors report evidence of discriminant validity using the Fornell-Larcker criterion, I remain concerned about the very strong association observed between learning engagement and academic burnout (r = -.76). Such a correlation raises questions regarding the empirical distinctiveness of these constructs. Additional evidence, such as HTMT ratios, comparisons with alternative measurement models, or a more detailed discussion of potential conceptual overlap, would strengthen the validity argument.

Reply:

We sincerely appreciate this comprehensive methodological comment. We revised the measurement section to clarify the treatment of Likert-type data and to strengthen psychometric reporting. Cronbach’s alpha was retained for comparability with previous studies, while McDonald’s omega and ordinal alpha were added as supplementary reliability indicators in Table 1. We also clarified the CFA procedure and added an ordinal sensitivity check based on polychoric correlations and a robust weighted least squares estimator.

Regarding scale refinement, we added explicit deletion criteria, including corrected item-total correlation, standardized loading, cross-loading, and wording consistency in the Chinese undergraduate context. We also added Table S2 to report the deleted item codes, deletion criteria, loadings before deletion, corrected item-total correlations, and psychometric consequences. Appendix B now provides the full questionnaire items and item-retention status.

Regarding discriminant validity, we added HTMT ratios and alternative measurement model comparisons. We also expanded the theoretical discussion of the strong negative relationship between learning engagement and academic burnout to clarify why the two constructs are closely related but not empirically redundant.

Action:

Location in the revised manuscript: Section 3.2; Section 4; Appendix A, Tables S2, S3, and S4; Appendix B

“Because Likert-type responses are ordinal at the item level, reliability and validity were evaluated in a way that considered both the ordinal response format and the composite-score strategy used in the mediation analysis. Cronbach’s alpha was retained for comparability with previous studies, while McDonald’s omega (McDonald, 1999) and ordinal alpha were also considered as supplementary reliability indicators.”

“Items were considered for deletion when they met at least one of the following criteria: corrected item-total correlation below .40, standardized loading below .60, theoretically unjustified cross-loading above .30, or wording that weakened conceptual consistency in the Chinese undergraduate context.”

“Although learning engagement and academic burnout were strongly correlated (r = -0.76), the Fornell-Larcker results still supported satisfactory discriminant validity. Because this correlation was close to the conventional threshold for construct overlap, additional evidence was examined through HTMT ratios and alternative measurement models (Henseler et al., 2015).”

Comment 6:

While the findings are clearly summarized, the discussion would benefit from deeper engagement with previous literature, a more detailed interpretation of effect sizes, and greater consideration of the broader implications of the results. In particular, the strong relationship between engagement and burnout deserves further theoretical reflection.

Reply:

We agree that the original discussion did not sufficiently interpret the magnitude and theoretical meaning of the findings. In the revised manuscript, we added more detailed explanations of the path coefficients and indirect associations, especially the relatively large association between learning engagement and academic burnout.

We now interpret learning engagement as the most proximal academic-functioning correlate of burnout in the proposed model. We also explain why engagement and burnout are strongly related but distinct: engagement reflects energy, effort, attention, and emotional connection in learning, whereas burnout reflects exhaustion, cynicism, and reduced academic efficacy under sustained academic strain. These revisions deepen the theoretical and practical interpretation of the findings.

Action:

Location in the revised manuscript: Section 4; Section 5

“Among all direct relationships, learning engagement displayed the largest negative association with academic burnout (β = -0.50, p < .001), suggesting that learning engagement was the most proximal correlate of burnout in the proposed model.”

“The indirect association through learning engagement was the largest single indirect pathway and accounted for 29.13% of the total association, indicating that students’ active learning involvement may be a particularly important explanatory link between academic self-efficacy and burnout-related risk.”

“Learning engagement describes the presence of energy, effort, attention, and emotional connection in academic activities, whereas academic burnout describes exhaustion, cynicism, and reduced academic efficacy under sustained academic strain.”

Comment 7:

The assessment of common method bias relies exclusively on Harman’s single-factor test, a procedure that has well-known limitations in contemporary methodological literature. The authors may wish to justify this choice or consider complementary approaches.

Reply:

Thank you for highlighting this important issue. We revised the common method bias assessment to avoid relying only on Harman’s single-factor test. The revised manuscript now reports both procedural remedies and complementary statistical checks. Procedurally, we explain that the survey was anonymous and that items from different constructs were psychologically separated. Statistically, we report Harman’s single-factor test, a one-factor CFA model, and an unmeasured latent method factor check.

We also explicitly acknowledge that Harman’s test is limited and that future studies should use stronger designs such as temporal separation, marker variables, multi-source reports, and behavioral data.

Action:

Location in the revised manuscript: Section 4

“Since all principal variables were measured through self-reported questionnaires at a single time point, common method variance was assessed through both procedural and statistical approaches.”

“Because Harman’s test is only a preliminary diagnostic and has recognized limitations (Podsakoff et al., 2003), two complementary checks were also considered. A one-factor CFA model showed poor fit (χ²/df = 10.22, CFI = 0.638, TLI = 0.602, RMSEA = 0.081, SRMR = 0.096), and the inclusion of an unmeasured latent method factor did not materially change the substantive path coefficients.”

“Future studies should incorporate temporal separation between predictor, mediator, and outcome measures, marker-variable techniques, multi-source reports, or behavioral indicators such as attendance records and learning analytics.”

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Dear author(s),

This manuscript examines the relationship between academic self-efficacy and academic burnout among university students, with a particular focus on the serial mediating roles of self-control and learning engagement. The topic is relevant and potentially valuable for the fields of educational psychology and higher education research, and the effort to integrate self-control and learning engagement into a sequential mediation framework is commendable. The manuscript also has practical significance in that it attempts to clarify the psychological pathways through which academic self-efficacy may reduce burnout. At the same time, the current version has several conceptual, methodological, interpretive, and editorial weaknesses that need to be addressed before the manuscript can be considered for publication. My comments below are intended to help strengthen the paper.

First, the explanation of Social Cognitive Theory is superficial. The paper only mentions the theory name and basic ideas, but does not deeply analyze how the four core variables interact with each other. Please use the core ideas of Social Cognitive Theory to explain the whole logical path clearly. Specifically, illustrate how academic self-efficacy improves self-control, how self-control boosts learning engagement, and how this chain reduces academic burnout. This will strengthen the theoretical basis of the paper.

Second, the full items of questionnaires used in the survey are not attached. The description of item purification is also incomplete and non-standard. The authors only state that some items of the learning engagement scale and academic burnout scale were deleted, but do not list the deleted item numbers, selection rules and related statistical data. It is hard to check whether the item selection is reasonable. Please add all questionnaire items in the appendix. List deleted items and related statistical indicators, and clearly show the whole process of item purification.

Third, basic information of research samples is incomplete. The paper only introduces the sample source and gender ratio, but lacks key demographic information such as university types, student grades and majors. This will affect readers’ judgment on sample representativeness and external validity of research findings. Please supplement all above demographic data and briefly describe the distribution of different groups.

Fourth, the theoretical reasoning for the proposed research hypotheses lacks detailed and rigorous elaboration. In particular, the demonstration and interpretation of the chain mediation hypotheses are excessively succinct, resulting in insufficient theoretical grounding and weak argument persuasiveness. Comprehensive reference to authoritative domestic and foreign empirical research findings is required to systematically unpack the intrinsic logical association between self-control and learning engagement step by step. Sufficient classical theoretical foundations and latest literature evidence should be supplemented to support every proposed hypothesis, which can significantly enhance the scientificity, rationality and academic persuasiveness of the overall hypothesis system of the paper.

Fifth, the results section is poorly connected with the discussion section. There is no in-depth analysis of statistical results. The paper only reports correlations, path coefficients and significant mediation effects, but does not discuss the results based on relevant theories and academic backgrounds. The theoretical and practical value of this study cannot be fully reflected. Please add targeted analysis and discussion after presenting data results, and explain the practical meaning of these findings.

Sixth, the discussion about research limitations and future research directions is too simple. The authors only mention common problems such as cross-sectional design, self-report bias and limited sample coverage, but do not analyze how these weaknesses affect the research conclusions. Besides, the future research suggestions are too general and not combined with the chain mediation model in this study. Please make a deeper analysis of research limitations and put forward specific and practical suggestions for future research based on the findings.

Seventh, the placement of Figure 1 requires reasonable adjustment. Since this figure displays the overall research conceptual model, readers need to constantly refer to it when reading the theoretical elaboration and research hypothesis sections. It is suggested to move Figure 1 to the end of the second section, immediately after the presentation of all research hypotheses from H1 to H9. This layout can closely combine textual descriptions with the model diagram, smooth the overall logical progression of the paper, enhance content coherence, and effectively improve readability for all readers.

Eighth, there are English grammar mistakes in the manuscript. For example, the sentence “Such an approach provide a more effective basis” in the fourth paragraph of the introduction has a subject-verb agreement error. The singular subject “an approach” should be followed by the verb “provides”. Please read through the whole paper carefully, correct all grammar, word and sentence errors, polish the language and keep a unified academic writing style.

Overall, I believe the manuscript addresses a meaningful research question and has the potential to make a useful contribution. However, the current version requires substantial revision in order to clarify the conceptual model, improve methodological transparency, align the conclusions more closely with the results, and raise the overall presentation quality of the paper.

Best regards.

Author Response

Comment 1:

Reply:

We appreciate this important theoretical suggestion. We substantially strengthened the explanation of Social Cognitive Theory and used it as the central logic connecting the four constructs. The revised manuscript now explains how academic self-efficacy functions as a personal belief, how self-control represents behavioral self-regulation, how learning engagement reflects observable learning participation, and how academic burnout reflects a negative academic state under sustained strain.

We also clarified the full pathway: efficacy beliefs help students interpret academic demands as manageable and support planning, monitoring, persistence, and adjustment; these self-control processes then enable sustained behavioral, cognitive, and emotional engagement; stronger engagement is expected to be inversely associated with burnout symptoms.

Action:

Location in the revised manuscript: Introduction; Section 2

“Social Cognitive Theory provides the central explanatory logic of the present model. The theory views student functioning as a reciprocal system in which personal beliefs, self-regulatory behavior, learning actions, and academic environments influence one another (Bandura, 1997; Zimmerman, 2000).”

“Self-control represents the behavioral self-regulatory component of the system: it transforms efficacy beliefs into planning, monitoring, inhibition of distraction, persistence, and adjustment of learning behavior.”

“Learning engagement then represents the observable learning-participation state that emerges when students are able to sustain behavioral involvement, cognitive investment, and emotional attachment to study.”

Comment 2:

Reply:

Thank you for this valuable recommendation. We have added Appendix B to provide the full questionnaire items used in the study, including item codes, dimensions, item wording, and retention status. This allows readers to examine the measurement content directly.

We also expanded the description of item purification and added Table S2, which reports the deleted item codes, deletion criteria, standardized loading before deletion, corrected item-total correlation, and the psychometric consequence of each deletion. These revisions improve the transparency and reproducibility of the measurement process.

Action:

Location in the revised manuscript: Section 3.2; Appendix A, Table S2; Appendix B, Tables A1–A4

“The deleted item codes, deletion criteria, and psychometric consequences are reported in Table S2, and the full item list is provided in Appendix B to improve transparency and reproducibility.”

“The retained and deleted item wordings are reported in Appendix B so that the refinement process can be reproduced and evaluated by readers.”

“Appendix B. Questionnaire Items Used in the Study.”

Comment 3:

Reply:

We agree with this comment. We supplemented the sample description by adding university type, grade level, disciplinary background, place of origin, and student leadership experience. These variables are now summarized in Table S1.

We also revised the sample description to explain that these classification variables provide contextual information for evaluating sample diversity and external applicability, even though they are not the core theoretical variables in the serial mediation model.

Action:

Location in the revised manuscript: Section 3.1; Appendix A, Table S1

“The retained sample showed a relatively balanced demographic structure. It included 689 male students (48.15%) and 742 female students (51.85%).”

“Respondents were also classified by university type, grade level, disciplinary background, place of origin, and student leadership experience.”

“These variables were not the main theoretical focus of the serial mediation model, but they provide important contextual information for evaluating sample diversity and for positioning the present study within more differentiated research on student functioning and academic burnout.”

Comment 4:

Reply:

Thank you for this detailed recommendation. We expanded the theoretical reasoning for each hypothesis and supplemented the discussion with both classical theoretical foundations and recent empirical literature. In particular, we strengthened the rationale for the chain mediation hypothesis by explaining why self-control should precede learning engagement rather than merely operate as a parallel mediator.

The revised manuscript now states that efficacy beliefs create motivational and cognitive readiness, but they do not automatically produce stable engagement unless students can regulate their behavior. Self-control therefore performs a bridging function through inhibition of distractions, time management, monitoring of goals, and correction of ineffective learning strategies. These self-regulatory processes then support behavioral, cognitive, and emotional engagement.

Action:

Location in the revised manuscript: Section 2

“The key theoretical issue in this study is not whether self-control and learning engagement are beneficial, but why self-control should precede learning engagement in the explanatory sequence.”

“In Social Cognitive Theory, efficacy beliefs provide motivational and cognitive readiness, but they do not automatically become stable learning participation unless students can regulate their behavior.”

“Self-control performs this bridging function by helping students inhibit distractions, manage time, monitor goal progress, and correct ineffective learning strategies.”

Comment 5:

Reply:

We appreciate this suggestion. We revised the Results and Discussion section so that statistical findings are followed by theoretical interpretation and practical implications. In particular, we added explanations of the meaning of the direct paths, the relative magnitude of the indirect associations, and the practical implications for burnout prevention.

The revised discussion now emphasizes that the largest single indirect pathway operated through learning engagement and that the serial pathway was also significant. We therefore recommend that universities not only strengthen academic confidence, but also translate confidence into self-regulatory capacity and sustained learning engagement.

Action:

Location in the revised manuscript: Section 4; Section 5

“The results for the direct paths were consistent with the proposed theoretical expectations. Academic self-efficacy not only showed a direct negative association with academic burnout, but was also positively associated with both self-control and learning engagement.”

“The indirect association through learning engagement was the largest single indirect pathway and accounted for 29.13% of the total association.”

“Practically, the findings suggest that burnout prevention should not focus only on burnout symptoms; it should also strengthen students’ confidence in their academic abilities, reinforce their capacity for self-regulation, and encourage active engagement in learning.”

Comment 6:

Reply:

Thank you for this helpful comment. We substantially expanded the limitations and future research section. The revised manuscript now explains how the cross-sectional design limits causal and temporal interpretation, how self-report data may inflate associations because of shared method variance, how the use of composite scores may conceal subgroup and dimension-level heterogeneity, and how the strong correlation between learning engagement and burnout requires further validation.

We also added more specific future research directions directly linked to the present model, including longitudinal, cross-lagged, and experimental designs; multi-source and behavioral indicators; subgroup and latent-profile analyses; and further validation of the learning engagement-burnout distinction.

Action:

Location in the revised manuscript: Section 5

“Several limitations should be noted. First, the cross-sectional design prevents causal and temporal conclusions. Although the model is theoretically grounded in Social Cognitive Theory, the empirical evidence supports associations among variables rather than confirmed causal mechanisms.”

“Second, all variables were measured through self-report questionnaires, which may introduce shared method variance and social desirability bias.”

“Future studies could address these limitations by using longitudinal, cross-lagged, or experimental designs to evaluate temporal ordering more rigorously.”

Comment 7:

Reply:

We agree with this layout suggestion. Figure 1 has been moved to the end of Section 2, immediately after the presentation of H1 to H9 and before the Research Design section. This placement allows readers to connect the theoretical arguments and hypotheses directly with the conceptual framework before entering the methodological section.

Action:

Location in the revised manuscript: End of Section 2, immediately before Section 3

“Figure 1 illustrates the proposed analytical framework and shows the direct pathway, the two single-mediator pathways, and the sequential pathway through self-control and learning engagement.”

“Figure 1. Conceptual framework of the study.”

“Specifically, the model proposes that academic self-efficacy is positively associated with self-control and learning engagement and negatively associated with academic burnout.”

Comment 8:

Reply:

Thank you for identifying this language issue. We corrected the specific subject-verb agreement error and conducted a full language revision of the manuscript. We also unified the academic style by replacing causal or overly strong expressions with cautious association-based wording, which is more appropriate for the cross-sectional design.

Action:

Location in the revised manuscript: Throughout the manuscript; Introduction

“Such an approach provides a more comprehensive basis for burnout prevention in higher education.”

“Using cross-sectional questionnaire data from 1,431 full-time undergraduates, this study examined the association between academic self-efficacy and academic burnout and tested whether self-control and learning engagement statistically mediated this relationship in sequence.”

“The model examines whether academic self-efficacy is linked to academic burnout directly and indirectly through the sequential mechanism of self-control and learning engagement.”

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

Feedback on Manuscript: Academic Self-Efficacy and Academic Burnout among College Students

Overall, this is a well-written paper addressing an important issue in higher education. The sample size is strong, the manuscript is clearly structured, and the findings are likely to be of interest to researchers and practitioners working in student wellbeing and academic success. The proposed model is logical and the analyses are generally appropriate. I have, however, identified several areas that would strengthen the manuscript further.

1. Strengthening the Contribution and Novelty

The paper positions the inclusion of self-control and learning engagement as a novel chain mediation model. While this is potentially valuable, there is now a substantial body of literature examining self-efficacy, engagement, self-regulation, and burnout in various combinations. It would strengthen the manuscript to more clearly articulate how this model advances existing knowledge beyond previous mediation studies and why the proposed sequence of self-control leading to learning engagement is theoretically important.

2. Caution Around Causal Language

The manuscript frequently uses language suggesting causal relationships (e.g., academic self-efficacy reduces burnout). Given that the study is based on cross-sectional survey data, causal conclusions cannot be established. I recommend adopting more cautious language throughout the manuscript, referring to associations, relationships, or predictive pathways rather than causal effects. This limitation should also be discussed more explicitly.

3. Common Method Bias

As all variables were collected using self-report questionnaires at a single point in time, the possibility of common method variance warrants careful consideration. While Harman's single-factor test was reported, many contemporary scholars view this test as insufficient on its own. It would strengthen the paper to acknowledge the limitations of this approach and discuss additional procedural or statistical methods that could be employed in future research.

4. Greater Transparency Regarding Scale Refinement

The manuscript indicates that items were removed from the learning engagement and academic burnout measures during the purification process. Additional detail would be helpful regarding which items were removed, the criteria used for removal, and how these decisions affected the psychometric properties of the scales. Providing this information would improve transparency and reproducibility.

5. Discriminant Validity Between Learning Engagement and Burnout

The reported correlation between learning engagement and academic burnout is relatively strong. Although the Fornell-Larcker criterion suggests acceptable discriminant validity, it would be useful to provide additional discussion regarding the conceptual distinctiveness of these constructs. Consideration of more contemporary approaches to discriminant validity assessment may further strengthen confidence in the findings.

6. Use of Demographic Variables

The study collected a range of demographic variables, including gender, grade level, major, place of origin, and leadership experience. However, it is unclear whether these variables were considered as controls in the mediation analyses. The authors may wish to justify their exclusion or conduct supplementary analyses to determine whether the reported relationships remain consistent when demographic characteristics are taken into account.

Overall, I believe these revisions would strengthen the theoretical contribution, methodological transparency, and interpretive rigour of the manuscript.

Author Response

Comment 1:

Strengthening the Contribution and Novelty: The paper positions the inclusion of self-control and learning engagement as a novel chain mediation model. While this is potentially valuable, there is now a substantial body of literature examining self-efficacy, engagement, self-regulation, and burnout in various combinations. It would strengthen the manuscript to more clearly articulate how this model advances existing knowledge beyond previous mediation studies and why the proposed sequence of self-control leading to learning engagement is theoretically important.

Reply:

We sincerely thank the reviewer for this constructive suggestion. We revised the manuscript to avoid overstating novelty and to clarify the specific contribution of the model. The revised Introduction now explains that the contribution does not lie in introducing entirely new variables, but in specifying the theoretically ordered pathway from academic self-efficacy to self-control, from self-control to learning engagement, and from learning engagement to burnout-related risk.

We also added a detailed theoretical explanation of why self-control should precede learning engagement. The revised argument is that efficacy beliefs provide motivational readiness, but self-control is needed to convert this readiness into stable learning participation.

Action:

Location in the revised manuscript: Introduction; Section 2

“The contribution lies in specifying a theoretically ordered process in which academic self-efficacy is linked to burnout-related risk through a staged pathway from cognitive belief to regulatory capacity and then to learning participation.”

“Therefore, the proposed sequence of self-control leading to learning engagement is not a mechanical ordering of variables; it reflects a theoretically meaningful transition from self-regulatory capacity to actual academic participation.”

“Existing evidence supports the separate links among self-control, adaptive academic functioning, engagement, and burnout, but the sequential ordering among these variables still requires explicit empirical examination.”

Comment 2:

Caution Around Causal Language: The manuscript frequently uses language suggesting causal relationships (e.g., academic self-efficacy reduces burnout). Given that the study is based on cross-sectional survey data, causal conclusions cannot be established. I recommend adopting more cautious language throughout the manuscript, referring to associations, relationships, or predictive pathways rather than causal effects. This limitation should also be discussed more explicitly.

Reply:

We fully agree with this comment. We carefully revised the manuscript to avoid causal language. The terms “effect,” “reduce,” and “influence” were replaced where appropriate with “association,” “relationship,” “linked to,” and “statistically mediated.” We also added explicit limitations explaining that the empirical results do not establish temporal or causal mediation.

These revisions ensure that the language of the paper is aligned with the cross-sectional and self-report nature of the data.

Action:

Location in the revised manuscript: Abstract; Section 3.3; Section 5

“Given the cross-sectional and self-report design, the results should be interpreted as evidence of theoretically informed associations rather than causal effects.”

“Because the data were cross-sectional, the mediation model is interpreted as a theoretically informed statistical mediation model rather than as evidence of temporal or causal mediation.”

“This means that the proposed order from academic self-efficacy to self-control, learning engagement, and academic burnout should be understood as theoretically plausible rather than empirically proven over time.”

Comment 3:

Common Method Bias: As all variables were collected using self-report questionnaires at a single point in time, the possibility of common method variance warrants careful consideration. While Harman's single-factor test was reported, many contemporary scholars view this test as insufficient on its own. It would strengthen the paper to acknowledge the limitations of this approach and discuss additional procedural or statistical methods that could be employed in future research.

Reply:

Thank you for this important methodological suggestion. We revised the common method bias section by adding procedural and statistical checks beyond Harman’s single-factor test. In addition to reporting the first unrotated factor, we added a one-factor CFA and an unmeasured latent method factor check.

We also explicitly acknowledged that Harman’s test is not sufficient as a definitive diagnostic. The revised manuscript now recommends temporal separation, marker-variable techniques, multi-source reports, and behavioral indicators in future research.

Action:

Location in the revised manuscript: Section 4

“Since all principal variables were measured through self-reported questionnaires at a single time point, common method variance was assessed through both procedural and statistical approaches.”

“Because Harman’s test is only a preliminary diagnostic and has recognized limitations (Podsakoff et al., 2003), two complementary checks were also considered.”

Comment 4:

Greater Transparency Regarding Scale Refinement: The manuscript indicates that items were removed from the learning engagement and academic burnout measures during the purification process. Additional detail would be helpful regarding which items were removed, the criteria used for removal, and how these decisions affected the psychometric properties of the scales. Providing this information would improve transparency and reproducibility.

Reply:

We appreciate this comment. We expanded the measurement section and added supplementary tables to make the item refinement procedure transparent. The revised manuscript now identifies the deleted items, explains the deletion criteria, and reports the psychometric consequences.

Specifically, USEI-CE5 and USEI-EE5 were removed from the learning engagement measure, while MBI-SS EX5, MBI-SS RE5, and MBI-SS RE6 were removed from the academic burnout measure. Table S2 reports the statistical indicators, and Appendix B provides both retained and deleted items.

Action:

Location in the revised manuscript: Section 3.2; Appendix A, Table S2; Appendix B

“For the learning engagement measure, deleting USEI-CE5 and USEI-EE5 improved the clarity of the three-dimensional structure and increased the reliability of the cognitive and emotional engagement dimensions without changing the conceptual scope of behavioral, cognitive, and emotional engagement.”

“For the academic burnout measure, deleting MBI-SS EX5, MBI-SS RE5, and MBI-SS RE6 produced a more balanced structure with four retained items per dimension and improved the stability of the exhaustion and reduced-efficacy dimensions after reverse coding.”

“These decisions were therefore made for psychometric transparency and parsimony, not to alter the theoretical meaning of either scale.”

Comment 5:

Discriminant Validity Between Learning Engagement and Burnout: The reported correlation between learning engagement and academic burnout is relatively strong. Although the Fornell-Larcker criterion suggests acceptable discriminant validity, it would be useful to provide additional discussion regarding the conceptual distinctiveness of these constructs. Consideration of more contemporary approaches to discriminant validity assessment may further strengthen confidence in the findings.

Reply:

Thank you for this insightful comment. We added additional discriminant validity evidence using HTMT ratios and alternative measurement model comparisons. The revised manuscript reports that the highest HTMT value was between learning engagement and academic burnout but remained below the conservative .85 criterion. We also added model comparisons showing that the four-factor model fit the data substantially better than a model combining learning engagement and burnout.

We further added a conceptual explanation distinguishing the two constructs. Learning engagement represents energy, effort, attention, and emotional connection in academic activities, while academic burnout represents exhaustion, cynicism, and reduced efficacy under sustained academic strain.

Action:

Location in the revised manuscript: Section 4; Appendix A, Tables S3 and S4

“The highest HTMT value was observed between learning engagement and academic burnout, but it remained below the conservative .85 criterion.”

“Moreover, the four-factor model fit substantially better than models in which learning engagement and academic burnout were collapsed into a single factor.”

“The two constructs are negatively related because they represent contrasting aspects of student functioning, but they are not simple mirror images.”

Comment 6:

Use of Demographic Variables: The study collected a range of demographic variables, including gender, grade level, major, place of origin, and leadership experience. However, it is unclear whether these variables were considered as controls in the mediation analyses. The authors may wish to justify their exclusion or conduct supplementary analyses to determine whether the reported relationships remain consistent when demographic characteristics are taken into account.

Reply:

We agree that the treatment of demographic variables needed clarification. In the revised manuscript, we explained that the primary model was estimated without controls to test the theoretical pathway parsimoniously. At the same time, we conducted a supplementary robustness analysis by entering gender, grade level, university type, disciplinary background, place of origin, and student leadership experience as controls.

The results remained consistent after including these controls, indicating that the main direct and indirect associations were robust to observed demographic differences. The supplementary results are now reported in Table S5.

Action:

Location in the revised manuscript: Section 3.3; Appendix A, Table S5

“Demographic variables were handled as supplementary controls rather than as core theoretical variables.”

“However, because gender, grade level, university type, disciplinary background, place of origin, and student leadership experience may be related to student functioning, a robustness analysis was also conducted with these variables entered as controls.”

“This supplementary analysis was used to evaluate whether the main direct and indirect associations remained consistent after accounting for observed demographic differences.”

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Dear Authors,

Thank you for submitting the revised version of your manuscript. The theoretical rationale is now much clearer, the novelty of the proposed serial mediation model is better justified, and the methodological reporting has been significantly strengthened. I particularly appreciate the inclusion of additional psychometric evidence, the clearer justification for the analytical strategy, the more comprehensive assessment of common method bias, and the more cautious interpretation of the findings given the cross-sectional design.

You have also responded appropriately to the previous concerns regarding the theoretical contribution of the study and have provided a more balanced discussion of the findings and their limitations.

Only a few minor issues remain. The Introduction could be slightly streamlined to reduce some redundancy, and a careful proofreading would help correct a few minor grammatical and editorial errors.

Thanks

Author Response

Reply: We sincerely thank the reviewer for this positive evaluation and helpful final suggestion. We agree that the Introduction could be made more concise. Accordingly, we revised the Introduction by removing repeated explanations, tightening the statement of the research gaps, and presenting the theoretical logic and contributions in a more direct manner. At the same time, we retained the manuscript’s existing cautious wording, such as “associated with,” “linked to,” and “statistical mediation,” to avoid causal overstatement given the cross-sectional design.

Action: The Introduction has been streamlined and proofread. The revised section now reads as follows:

“Academic burnout remains a persistent concern in higher education because it is closely related to students’ study adjustment, emotional health, and academic development (Abraham et al., 2024; Liu et al., 2024; Wang et al., 2024). Rather than a brief response to temporary academic pressure, it refers to a sustained dysfunctional learning state marked by emotional exhaustion, detached or negative attitudes toward study, and a weakened sense of academic capability (Jagodics & Szabó, 2023). Its consequences may extend beyond emotional discomfort by undermining learning motivation, classroom involvement, academic achievement, and continued participation in university life (Olson et al., 2025; Wang et al., 2021). Clarifying the psychological processes associated with academic burnout therefore has both theoretical and practical significance.”

“Although academic self-efficacy, self-control, learning engagement, and academic burnout have each been widely examined (Reyes-de-Cózar et al., 2023; Tang & He, 2025; Yang et al., 2024), several gaps remain. First, many studies have tested pairwise relationships or parallel mediation models, which are useful for identifying associations but provide limited explanation of the order in which cognitive beliefs, regulatory capacity, and learning participation may operate. Second, engagement and self-regulation have often been treated as separate explanatory variables, leaving insufficient attention to the theoretically meaningful sequence from efficacy beliefs to self-control and then to learning engagement. Third, most existing models emphasize average relationships among composite scores, whereas student functioning may differ across demographic and academic subgroups. The present study therefore examines an average-process model while acknowledging the need for more differentiated subgroup, moderation, latent-profile, and dimension-level analyses in future research.”

“To address these gaps, this study proposes a serial mediation framework in which academic self-efficacy is associated with academic burnout through self-control and learning engagement. Social Cognitive Theory suggests that efficacy beliefs shape how students appraise academic demands, set goals, mobilize effort, and regulate behavior (Bandura, 1997; Zimmerman, 2000). Students with stronger efficacy beliefs may be more likely to interpret academic difficulties as manageable and to maintain planning, monitoring, persistence, and behavioral adjustment. These regulatory behaviors are reflected in self-control, which may help translate efficacy beliefs into sustained learning involvement. Learning engagement, as a more proximal form of behavioral, cognitive, and emotional participation in study, is expected to be inversely associated with burnout symptoms. Thus, the proposed model clarifies an ordered pathway from cognitive belief to regulatory capacity, from regulatory capacity to learning involvement, and from learning involvement to burnout-related risk.”

“This study makes three main contributions. First, it specifies a staged process in which academic self-efficacy is linked to burnout-related risk through regulatory capacity and learning participation, rather than presenting the four variables as isolated predictors. Second, it compares the relative magnitudes of the direct association, the two single-mediator pathways, and the serial pathway, thereby clarifying whether learning engagement is a more proximal correlate of academic burnout than self-control in the proposed framework. Third, it documents sample classification variables and adds supplementary analyses to examine whether the main pathways remain stable after demographic characteristics are considered. Practically, the findings suggest that burnout-related interventions should not focus only on symptoms, but should also strengthen students’ academic confidence, self-regulatory capacity, and active learning engagement.”

Additional proofreading action: We also corrected minor grammatical and editorial issues identified during proofreading, including the sentence in the Results section describing the direct associations: “academic self-efficacy was significantly and positively associated with self-control ... and was significantly and negatively associated with academic burnout ...”.

Reviewer 3 Report

Comments and Suggestions for Authors

Recommendation: Accept. The authors have responded comprehensively to the previous review comments and have substantially strengthened the manuscript. The theoretical rationale, methodological transparency, and interpretation of the findings have all been improved, resulting in a much stronger and more rigorous paper.

Author Response

Reply: We are grateful to the reviewer for the encouraging assessment and for recognizing the improvements made in the revised manuscript. No further substantive revision was requested by this reviewer. In the present revision, we have nevertheless conducted an additional careful proofreading and ensured that the Introduction remains concise and consistent with the overall manuscript.

Action: No additional theoretical or methodological changes were required in response to this comment. The final proofreading and Introduction streamlining have been incorporated into the revised manuscript.

Once again, we sincerely thank the Editor and Reviewers for their valuable comments and constructive guidance.