Physical Activity and Mental Well-Being Among University Students: The Role of Beliefs in the Mental Health Benefits of Physical Activity

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Dear Authors,

Thank you for the opportunity to review your manuscript. Your study addresses a timely and valuable topic regarding beliefs about physical activity and their role in mental well-being. The subject is relevant, contemporary, and theoretically interesting. However, several revisions are needed to strengthen the structure, clarity, and methodological transparency of the manuscript. My comments are outlined below.

 

Introduction

1. The Introduction is currently too long and resembles the literature review section of a thesis. A standard journal Introduction is typically between 400 and 500 words and is ideally structured into approximately four main paragraphs.

2. If you feel that the topic may not be familiar to a general scientific audience and that the necessary background cannot be adequately covered within this structure, I suggest tailoring the Introduction to readers who are already knowledgeable in the field. Additional background information intended for non-expert readers could be transferred to the Supplementary Materials.

3. Streamlining the Introduction will improve readability and ensure that the research gap and study objective are presented more directly and concisely.

 

Methods

1. In the Instruments section, the newly developed belief scale should be included as an appendix or supplementary file.

2. Have you published another study specifically focused on the development of this instrument? If so, please clearly reference it.

3. If you do not intend to publish a separate development study, the current manuscript should include a more detailed description of the scale development process (e.g., item generation, expert review, pilot testing). This section requires clearer structure and more comprehensive explanation.

4. The belief scale shows very high internal consistency (α = .97) and high factor loadings. While this is encouraging, such high reliability may also indicate potential item redundancy. Further clarification would strengthen the validation section.

 

Results

1. Please add a table or a subheading presenting baseline characteristics of the sample.

2. I recommend adding a dedicated subsection titled “Psychometric Properties.” Information regarding the psychometric characteristics of translated and newly developed instruments should be consolidated under this heading. This will improve clarity and organization.

 

Discussion

1. The study employs a cross-sectional design with self-report measures for all variables. Under such conditions, mediation analysis demonstrates statistical associations consistent with a mediation model, but it does not establish causal direction or mechanism. In several places, the manuscript states that beliefs “fully mediate,” “fully explain,” or “account for” the relationship between physical activity and mental well-being. These formulations risk implying causal inference beyond what the study design permits. It would be better to replace causal language with more cautious phrasing, acknowledge alternative plausible directional models (for example, mental well-being influencing beliefs and physical activity), and clarify that “complete mediation” reflects a statistical pattern rather than evidence of a causal psychological mechanism.

2. The manuscript draws on Bandura’s and Kirsch’s theories, activity adequacy mindset research, and self-efficacy literature. While this theoretical integration is ambitious and interesting, the distinctions among these constructs are not sufficiently clarified. Please more clearly explain how your belief construct differs from or overlaps with these related concepts.

3. Given that 80% of the sample is female, you should either: Test gender differences, or explicitly discuss sample imbalance as a limitation.

 Limitations

All variables were assessed via self-report at a single time point. This raises concerns about shared method variance and consistency bias.

 

Overall, your study addresses an important and relevant topic. With structural refinement, clearer theoretical positioning, and more cautious interpretation of findings, the manuscript will be significantly strengthened.

I hope these comments are helpful in revising your valuavle work.

With best regards

 

Comments on the Quality of English Language

There is some typo and grammatical issues.

Author Response

Dear Authors,

Thank you for the opportunity to review your manuscript. Your study addresses a timely and valuable topic regarding beliefs about physical activity and their role in mental well-being. The subject is relevant, contemporary, and theoretically interesting. However, several revisions are needed to strengthen the structure, clarity, and methodological transparency of the manuscript.

 

General Response

 

Dear reviewer, we sincerely thank you for your careful reading of our manuscript and for the constructive and thoughtful comments. We particularly appreciate detailed suggestions regarding the structure of the Introduction, the reporting of the scale development procedure, and the clarification of the psychometric evaluation of the newly developed belief scale. These comments were very helpful in identifying places where the manuscript could be made clearer and more focused.

In response to the comments and suggestions, several parts of the manuscript have been revised. The Introduction has been substantially shortened and reorganized in order to present the research background, study objectives, and hypothesis more directly. The description of the scale development procedure in the Methods section has been expanded and structured more clearly, and the full wording of the belief scale items together with expert content validity ratings and pilot clarity results is now provided in the Supplementary Materials. In addition, the Results section has been reorganized by introducing a dedicated subsection reporting the psychometric properties of the scale, and the Discussion has been revised to ensure a more cautious interpretation of the mediation analysis and to clarify the conceptual distinctions between the belief construct and related theoretical concepts.

We are grateful for these comments, which we believe have helped improve the clarity, transparency, and overall structure of the manuscript. We hope that the revisions adequately address the reviewer’s concerns.

 

Introduction

 

1. The Introduction is currently too long and resembles the literature review section of a thesis. A standard journal Introduction is typically between 400 and 500 words and is ideally structured into approximately four main paragraphs.

 

2. If you feel that the topic may not be familiar to a general scientific audience and that the necessary background cannot be adequately covered within this structure, I suggest tailoring the Introduction to readers who are already knowledgeable in the field. Additional background information intended for non-expert readers could be transferred to the Supplementary Materials.

 

3. Streamlining the Introduction will improve readability and ensure that the research gap and study objective are presented more directly and concisely.

 

Response:

 

These comments and suggestions were very helpful. In response to these suggestions, the Introduction has been substantially shortened and streamlined. We removed the broader historical overview of research on the determinants and outcomes of PA and reduced the discussion of theoretical models and previous studies. The revised Introduction is now streamlined and structured into four main paragraphs, describing the empirical and theoretical foundations on which the study is based, the research gaps it seeks to address, the objectives, the hypothesis, and the study's contributions. We hope that these revisions have made the Introduction more focused and tailored to readers knowledgeable in this specific field, while still leaving sufficient information for a more general scientific audience, making additional supplementary material related to the Introduction unnecessary.

 

Methods

 

1. In the Instruments section, the newly developed belief scale should be included as an appendix or supplementary file.

 

2. Have you published another study specifically focused on the development of this instrument? If so, please clearly reference it.

 

3. If you do not intend to publish a separate development study, the current manuscript should include a more detailed description of the scale development process (e.g., item generation, expert review, pilot testing). This section requires clearer structure and more comprehensive explanation.

 

4. The belief scale shows very high internal consistency (α = .97) and high factor loadings. While this is encouraging, such high reliability may also indicate potential item redundancy. Further clarification would strengthen the validation section.

 

Response:

 

We thank the reviewer for these helpful comments, and as they address closely related aspects of the development, documentation, and psychometric evaluation of the instrument, we provide a joint response summarizing the revisions made to the manuscript.

 

First, following the reviewer’s suggestion, the full wording of the scale items is now provided in Supplementary File 1. This table presents the initial 11-item version of the Belief in the Mental Health Benefits of Physical Activity Scale, together with item-level expert content validity ratings (I-CVI) and pilot clarity evaluation results. Including these materials improves transparency regarding the content of the scale and the evaluation of individual items.

 

Second, we appreciate the reviewer’s suggestion to clarify the reporting of the instrument development. As the development of this scale has not been described in a separate publication, we expanded the description of the scale development process in the Methods section to provide a clearer and more comprehensive account of how the instrument was constructed and evaluated in the present study.

Specifically, the revised Methods section now provides a clearer and more structured description of the scale development procedure, including:

(1) the conceptual definition of the construct and item generation based on empirical findings and relevant theoretical perspectives,

(2) expert content validation using item-level content validity indices (I-CVI),

(3) pilot testing assessing item clarity in an independent sample of adult participants, and

(4) subsequent item reduction based on psychometric analyses.

 

Finally, we appreciate the reviewer’s observation that the very high internal consistency of the initial scale version (α = .97) may indicate potential item redundancy. To address this concern, we expanded the psychometric evaluation and reported additional indices beyond Cronbach’s alpha, including the average inter-item correlation (AIC), corrected item–total correlations, and a SEM-based omega reliability estimate derived from the CFA model. These diagnostics indicated very high item homogeneity in the original 11-item scale (α = .971; AIC = .755; corrected item–total correlations = .799–.921; SEM-based ω = .968), which is consistent with potential conceptual overlap among items.

Based on these findings, we derived and evaluated a shortened 6-item version of the scale designed to preserve conceptual coverage while reducing redundant content. The shortened scale retained a clear one-factor structure and strong psychometric performance (EFA and CFA results reported in the revised Results section), with high reliability (α = .939; SEM-based ω = .939) and slightly lower inter-item overlap (AIC = .722; corrected item–total correlations = .768–.845). The revised manuscript now provides an explicit rationale for item reduction and reports the psychometric results for both the original and shortened versions in the Psychometric Properties section of the Results.

 

We hope that these revisions clarify the development and evaluation of the scale and adequately address the reviewer’s comments regarding the reporting of the instrument and its psychometric properties.

 

Results

 

1. Please add a table or a subheading presenting baseline characteristics of the sample.

 

Response:

We completely understand the value of this comment and suggestion. However, the main baseline characteristics of the sample (age, gender distribution, and study level) are described in the Participants subsection of the Methods section, where the composition of the study sample is presented. The study focused on a relatively homogeneous sample of university students capable of performing typical physical activities, and therefore, only basic sample characteristics were recorded. We are aware that additional background information (e.g., socioeconomic or contextual variables) could provide further insight into the sample composition, but unfortunately, we did not include them in the present data collection. Given the limited number of available background variables, we retained the current text-based description of the sample, as presenting these characteristics in a separate table would provide little additional information – a separate table would appear overly simplistic, or add little additional information. We also acknowledge that the absence of additional background characteristics may limit the contextual interpretation of the findings, and we have now explicitly noted this point in the Limitations subsection of the Discussion.

 

 

2. I recommend adding a dedicated subsection titled “Psychometric Properties.” Information regarding the psychometric characteristics of translated and newly developed instruments should be consolidated under this heading. This will improve clarity and organization.

 

Response:

We agree that presenting the instrument's psychometric characteristics in a subsection improves the clarity and organisation of the Results section.

Following this recommendation, we introduced a new subsection titled “Psychometric Properties of the Belief in the Mental Health Benefits of PA Scale". In this section, we consolidated the analyses of the scale’s psychometric evaluation. Specifically, we now report the exploratory and confirmatory factor analyses for the original 11-item version, additional diagnostics addressing item redundancy and internal consistency, the rationale and procedure for scale reduction, and the psychometric evaluation of the resulting 6-item short form.

We hope that this reorganisation makes the presentation of the scale’s psychometric characteristics clearer and easier to follow.

 

Discussion

 

1. The study employs a cross-sectional design with self-report measures for all variables. Under such conditions, mediation analysis demonstrates statistical associations consistent with a mediation model, but it does not establish causal direction or mechanism. In several places, the manuscript states that beliefs “fully mediate,” “fully explain,” or “account for” the relationship between physical activity and mental well-being. These formulations risk implying causal inference beyond what the study design permits. It would be better to replace causal language with more cautious phrasing, acknowledge alternative plausible directional models (for example, mental well-being influencing beliefs and physical activity), and clarify that “complete mediation” reflects a statistical pattern rather than evidence of a causal psychological mechanism.

 

Response:

The description of the mediation analysis findings has been revised throughout the manuscript to avoid implications of causal relations.  Specifically, expressions implying causal mediation (e.g., “fully mediates” or “explains”) were replaced with more cautious formulations (e.g., “indirect-only mediation pattern” or “association conveyed through beliefs”), and the possibility of alternative directional relationships has been acknowledged.   

 

2. The manuscript draws on Bandura’s and Kirsch’s theories, activity adequacy mindset research, and self-efficacy literature. While this theoretical integration is ambitious and interesting, the distinctions among these constructs are not sufficiently clarified. Please more clearly explain how your belief construct differs from or overlaps with these related concepts.

 

Response:

The description of similarities and differences between outcome expectations, response expectancies, and beliefs in the mental health benefits of PA has been added in the Discussion. In particular, the revised Discussion clarifies how the present construct reflects domain-specific beliefs about the mental health consequences of one’s own PA, which overlap with but are conceptually distinct from broader outcome expectations and response expectancies.     

 

3. Given that 80% of the sample is female, you should either: Test gender differences, or explicitly discuss sample imbalance as a limitation.

 

Response:

A reference to sample imbalance has been added in the description of study limitations in the Discussion.     

 

Limitations

 

All variables were assessed via self-report at a single time point. This raises concerns about shared method variance and consistency bias.

 

Response:

This issue has now been addressed in the description of study limitations in the Discusion, where we acknowledge the potential influence of shared method variance and consistency bias due to the exclusive use of self-report measures.

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript addresses a relevant question on how beliefs about PA’s mental-health benefits relate to well-being in university students. The topic is timely, and the attempt to develop a multi-item belief scale is promising. However, it is needed due to design interpretation, scale validation reporting, and mediation-analysis transparency.

1. Soften casual language (e.g., “full-complete mediation”). With cross-sectional data, directionality and confounding remain plausible. Reframe as an associational indirect-only pattern and expand limitations.
2. Provide fuller-scale development details (construct definition, item generation, expert-content review, pilot). Avoid relying only on Kaiser-Guttman; add parallel analysis. Report omega and item-total statistics, and discuss potential redundancy (very high internal consistency may indicate overlap). If EFA and CFA used the same sample, acknowledge it as preliminary or cross-validate if possible.
3. Correlated residuals between similar items should be justified and discussed (local dependence-redundancy). Include full item wording and report RMSEA CI and fit criteria used.
4. Specify the PROCESS model, bootstrap resamples-CI type, missing-data handling, and covariates. Report unstandardised + standardised paths and an indirect-effect effect size. Consider SEM mediation with latent variables if feasible.
5. All measures are self-report and concurrent; discuss common-method variance and construct overlap as a limitation.
6. Clarify Godin–Shephard as a PA score (not objective “intensity”).
7. Briefly justify SWEMWBS scoring approach.
8. Correct abstract phrasing (avoid “serial” if only one mediator).
9. Note generalizability limits (predominantly female sample) and interpret practical significance cautiously.

Author Response

REVIEWER 2

 

The manuscript addresses a relevant question on how beliefs about PA’s mental-health benefits relate to well-being in university students. The topic is timely, and the attempt to develop a multi-item belief scale is promising. However, it is needed due to design interpretation, scale validation reporting, and mediation-analysis transparency.

 

General response

We would like to thank the reviewer for carefully reading our manuscript and for the detailed, constructive comments. We particularly appreciate the reviewer’s thoughtful suggestions regarding the reporting of the scale development procedure, the clarification of the psychometric analyses, and the transparency of the mediation modeling approach. These comments were very helpful in identifying several aspects of the manuscript where the methodological description and interpretation of the results could be improved.

In response to these suggestions, we revised multiple sections of the manuscript. The description of the scale development process in the Methods section has been expanded to provide a clearer account of construct definition, item generation, expert content validation, and pilot testing. The reporting of the psychometric analyses has also been strengthened by including additional reliability and item-level statistics, as well as parallel analysis for factor retention and a more explicit discussion of potential item redundancy.

Following the reviewer’s recommendation regarding mediation analysis, we also revised the analytic approach by estimating the mediation model using structural equation modeling (SEM). This allowed us to specify beliefs about the mental health benefits of PA and mental well-being as latent variables and to report both standardized and unstandardized path estimates together with bootstrap confidence intervals for indirect effects. In addition, the manuscript now provides clearer reporting of estimation procedures, model fit indices, and limitations related to the cross-sectional design and the exclusive use of self-report measures.

We are grateful for these comments, which helped us improve the methodological clarity and transparency of the manuscript. We hope that the revisions adequately address the reviewer’s concerns.

 

1. Soften casual language (e.g., “full-complete mediation”). With cross-sectional data, directionality and confounding remain plausible. Reframe as an associational indirect-only pattern and expand limitations.

 

Response:

 

The description of the mediation analysis findings has been revised throughout the manuscript to avoid implications of causal relations. In particular, expressions suggesting causal mediation (e.g., “fully mediates” or “explains”) were replaced with more cautious formulations (e.g., “indirect-only mediation pattern” or “association conveyed through beliefs”), and the possibility of alternative directional relationships is now acknowledged in the Discussion.

    

2. Provide fuller-scale development details (construct definition, item generation, expert-content review, pilot). Avoid relying only on Kaiser-Guttman; add parallel analysis. Report omega and item-total statistics, and discuss potential redundancy (very high internal consistency may indicate overlap). If EFA and CFA used the same sample, acknowledge it as preliminary or cross-validate if possible.

 

Response:

 

We would like to thank the reviewer for this constructive comment and for highlighting several important aspects of the scale's development and psychometric evaluation.

 

In response to the first part of this comment, we substantially expanded the description of the scale development process in the Methods section. The revised manuscript now provides a clearer and more structured account of the conceptual definition of the construct, the item generation procedure, expert content validation, and pilot testing used to evaluate item clarity.

 

Specifically, the Methods section now includes a detailed description of:

(1) the conceptual definition of the construct and item generation informed by empirical findings and relevant theoretical perspectives,

(2) expert content validation using item-level content validity indices (I-CVI), and

(3) a pilot study assessing the clarity of the items in an independent sample of adult participants, and

(4) subsequent item reduction informed by psychometric analyses.

 

Detailed item-level results for the expert evaluation and pilot clarity assessment are now provided in Supplementary File 1, which presents the full wording of the initial 11-item version of the scale together with expert relevance ratings and pilot clarity results.

 

In response to the reviewer’s suggestions regarding the psychometric evaluation of the scale, we expanded the reporting of reliability and item-level statistics in the Psychometric Properties section of the Results. In addition to Cronbach’s alpha, we now report corrected item–total correlations, the average inter-item correlation (AIC), and an omega reliability estimate computed as an SEM-based omega derived from the CFA model. These indices indicated very high item homogeneity in the original 11-item scale (α = .971; AIC = .755; corrected item–total correlations = .799–.921; SEM-based ω = .968), suggesting potential conceptual overlap among items.

Based on these findings, we evaluated a shortened 6-item version of the scale derived from the original item pool to preserve conceptual coverage while reducing redundant content. The revised manuscript now provides an explicit rationale for item reduction and reports the psychometric results for both the original and shortened versions of the scale.

Regarding factor retention, we complemented the Kaiser–Guttman criterion with parallel analysis, which also supported a one-factor solution. Finally, because cross-validation was not feasible with the current dataset, we explicitly acknowledge in the manuscript that both exploratory and confirmatory factor analyses were conducted on the same sample and present these findings as preliminary evidence for the measurement structure, recommending cross-validation in future studies.

 

We hope that these revisions adequately address the reviewer’s comments and improve the clarity and transparency of the scale development and psychometric evaluation reported in the manuscript.

 

3. Correlated residuals between similar items should be justified and discussed (local dependence-redundancy). Include full item wording and report RMSEA CI and fit criteria used.

 

Response:

 

We have clarified the rationale for allowing correlated residuals between the two conceptually similar items (stress and worry), noting that their highly similar wording and closely related affective content may introduce local item dependence. As the final version of the scale was subsequently reduced to a six-item form to minimize redundancy, the items involved in the correlated residual specification are no longer part of the retained item set. We have also included the full wording of all items in Supplementary Table S1 and expanded the reporting of model fit by providing the RMSEA 90% confidence interval and specifying the fit criteria used to evaluate model adequacy.

 

4. Specify the PROCESS model, bootstrap resamples-CI type, missing-data handling, and covariates. Report unstandardised + standardised paths and an indirect-effect effect size. Consider SEM mediation with latent variables if feasible.

 

Response:

 

We consider this suggestion helpful and very constructive. In response, we revised the mediation analysis to improve methodological transparency and robustness and to address the reviewer’s recommendation to consider SEM-based mediation.

Rather than retaining the original PROCESS-based mediation approach, we estimated the mediation model using structural equation modeling (SEM). This allowed us to specify beliefs about the mental health benefits of PA and mental well-being as latent variables, while PA intensity was retained as an observed predictor. The revised SEM demonstrated acceptable model fit and supported a significant indirect effect of PA intensity on mental well-being via beliefs, whereas the direct effect was not significant.

We also clarified the analytic procedure in the Methods and Results sections. Model parameters were estimated using maximum likelihood (ML), and missing-data handling was specified using full information maximum likelihood (FIML). In the present dataset, all analyses were conducted on complete cases (fully completed questionnaires; N = 339), meaning that no missing responses were present; therefore, FIML estimation did not alter parameter estimation but is reported for transparency of the estimation procedure. Standard errors and confidence intervals were obtained using nonparametric bootstrap procedures with bias-corrected and accelerated (BCa) 95% confidence intervals based on 2,000 resamples. No covariates were included in the mediation model, as the analysis was designed to examine the primary associations among PA intensity, beliefs about the mental health benefits of PA, and mental well-being. This modeling choice is acknowledged as a limitation of the study and discussed in the Limitations section of the manuscript.

In addition, the revised Results section now reports both unstandardized and standardized coefficients for all structural paths, as well as the unstandardized and standardized indirect and total effects. Standardized estimates are also presented in Figure 1 to facilitate comparison of path magnitudes. To address the reviewer’s request for an indirect-effect effect size, we report the standardized indirect effect (β), which provides an interpretable index of the magnitude of the mediated pathway (see Lachowicz, Preacher, & Kelley, 2018; we realize that providing the link may not be strictly necessary, but we include it here for completeness: https://pubmed.ncbi.nlm.nih.gov/29172614/ )

Finally, we explored an alternative SEM specification in which PA intensity was modeled as a latent factor indicated by the three GSLTPAQ activity-frequency items (mild, moderate, and strenuous activity frequency). However, because the GSLTPAQ score reflects a weighted composite based on a formative scoring logic rather than a reflective latent construct, this specification produced unstable estimation and non-uniform indicator behavior. Accordingly, the SEM with observed PA intensity, latent beliefs, and latent mental well-being was retained as the primary model reported in the manuscript.

 

5. All measures are self-report and concurrent; discuss common-method variance and construct overlap as a limitation.

 

Response:

 

We consider this comment very important. Because all variables in the study were assessed using self-report measures at a single time point, we agree that the potential influence of common-method variance should be explicitly addressed.

First, following the reviewer’s suggestion, we examined the potential influence of common-method variance using Harman’s single-factor test. An unrotated principal component analysis including all measurement items (PA, mental well-being, and belief scale items) indicated that the first factor accounted for 43.53% of the variance in the initial 11-item version of the belief scale and 34.4% in the final six-item version. As no single factor accounted for most of the variance, these results suggest that common-method variance is unlikely to substantially bias the observed relationships. A description of this analysis has been added to the Statistical Analysis subsection of the Methods.

Second, we acknowledge that the concurrent self-report assessment of conceptually related constructs may still introduce shared method variance or partial construct overlap. This issue is now explicitly discussed in the Limitations subsection of the Discussion, where we note that the cross-sectional and self-report nature of the data may have inflated the observed associations between variables.

These additions improve the methodological transparency of the manuscript and acknowledge the potential limitations associated with the measurement approach used in the study.

 

 

6. Clarify Godin–Shephard as a PA score (not objective “intensity”).

 

Response:

 

We thank the reviewer for this helpful clarification. The description of the Godin–Shephard Leisure-Time PA Questionnaire in the Methods section has been revised to clarify that the instrument provides a composite leisure-time PA score based on the reported frequency of activities of different intensities. The manuscript now explicitly states that this score represents a self-reported indicator of PA intensity.

 

7. Briefly justify SWEMWBS scoring approach.

 

Response:

We also appreciate this comment. We have now briefly clarified the SWEMWBS scoring approach in the Methods section. Specifically, we added a sentence noting that the total raw score was obtained by summing the item scores. Although the developers of SWEMWBS provide a Rasch-based conversion of raw scores to metric scores, we used raw sum scores in the present study, as this approach is also common in empirical research employing the scale. We hope this clarification improves the transparency of the scoring procedure

 

 

8. Correct abstract phrasing (avoid “serial” if only one mediator).

 

Response:

We appreciate pointing this out. The wording in the abstract has been corrected to remove the term “serial mediation,” as the model includes only a single mediator. The revised abstract now consistently refers to mediation and indirect effects.

 

9. Note generalizability limits (predominantly female sample) and interpret practical significance cautiously.

 

Response:

A reference to sample imbalance has been added in the description of study limitations in the Discussion. Practical implications have been intrepreted more cautiously, taking into account the cross-sectional design and the modest effect sizes observed in the analyses

Reviewer 3 Report

Comments and Suggestions for Authors

 Thank you for the opportunity to review your manuscript examining the mediating role of beliefs about the mental health benefits of physical activity in the relationship between physical activity intensity and mental well-being among university students.

Several methodological and interpretative aspects would benefit from clarification and strengthening to enhance the scientific rigor of the study.

 

Comment 1:

In the abstract, I encourage you to slightly temper the causal wording used to describe the mediation results. Phrases suggesting that beliefs “fully mediate” or “explain” the relationship between physical activity and mental well-being may overstate what can be concluded from cross-sectional data. Reframing the findings as “consistent with an indirect association” or “statistically compatible with mediation” would provide a more methodologically accurate summary. It may also be helpful to briefly indicate the modest proportion of explained variance in mental well-being to give readers a clearer sense of the practical magnitude of the effects. Finally, specifying that the design is cross-sectional directly in the abstract would improve transparency.

 

 

Comment 2

The introduction section feels, at times,  somewhat lengthy and could benefit from tighter focus on the specific research gap addressed by the present study. You might consider streamlining parts of the historical discussion and moving more quickly toward the rationale for developing the new belief scale and testing the mediation model in university students. Clarifying the novelty of the study more explicitly (for example, by emphasizing how this belief-based mechanism differs from existing self-efficacy or affect-based explanations) would strengthen the argument for originality. Additionally, clearly stating the hypotheses in a concise and testable form at the end of the introduction would help guide the reader.

 

Comment 3

In the Methods section, the recruitment procedure would benefit from additional clarification. Although you specify that 339 university students aged 18–28 participated and that inclusion required being a “healthy student capable of typical physical activities,” more detail regarding how participants were approached, the recruitment channels used, and the approximate response rate would help readers better evaluate representativeness. Given that 80.2% of the sample was female, it would also be important to acknowledge the potential implications of this imbalance for generalizability and to briefly discuss possible sampling bias.

Because all constructs ( PA intensity (GSLTPAQ), mental well-being (SWEMWBS), and beliefs ) were assessed via self-report at a single time point, the risk of common method variance should be addressed more explicitly. This could be done either by reporting a statistical check (e.g., a Harman single-factor test or similar approach) or by expanding the limitations section to acknowledge the potential inflation of associations due to shared method variance.

The newly developed Belief in the Mental Health Benefits of Physical Activity Scale is a clear contribution, and the reported EFA and CFA results (including strong loadings and acceptable fit indices after correlating Items 1 and 2) are presented clearly. However, the very high internal consistency (α = .97) raises the possibility of item redundancy rather than optimal construct coverage. To reinforce the psychometric argument, I recommend reporting additional indices such as composite reliability and average variance extracted, as well as examining discriminant validity against conceptually related constructs. It may also be worthwhile to explore whether a shorter version of the scale could retain similar explanatory power while reducing potential overlap. Testing measurement invariance across gender or academic level would further strengthen the validation evidence.

With regard to the mediation analysis, the explained variance in mental well-being (7.31%) is modest and would benefit from more explicit contextualization in terms of practical significance. The current model does not include covariates such as gender, age, or academic stress, which are established correlates of student mental health and could influence the observed relationships. Providing more information about assumption testing and specifying the number of bootstrap resamples used in the PROCESS analysis would improve methodological transparency.

 

Comment 4

In the Results section, providing slightly more guidance regarding the magnitude of the observed effects would help readers better interpret their practical meaning. For example, the mediation model accounts for 7.31% of the variance in mental well-being. While this effect is statistically significant, it is relatively small in magnitude, and briefly contextualizing this value, such as noting that beliefs explain only a modest proportion of variability in students’ well-being, would allow readers to better assess the practical importance of the findings.

When presenting the mediation model, it would also be helpful to distinguish more explicitly between standardized and unstandardized coefficients. In Figure 1 and in the text, please ensure that it is clearly specified whether the reported β values are standardized and that labeling is consistent throughout. This will make it easier for readers to compare the relative strength of the paths (e.g., PA → belief, belief → well-being, and the indirect effect).

While confidence intervals are reported for several paths, they should be presented consistently for all key effects, including the total, direct, and indirect effects, to provide a complete picture of precision. Clarifying whether the assumptions underlying the analyses were checked (such as normality of residuals, absence of multicollinearity, and suitability for mediation modeling) would further strengthen the transparency and robustness of the statistical reporting.

 

Comment 5

The Discussion section is conceptually strong; however, some interpretations appear stronger than warranted by the design. Statements suggesting that beliefs “fully account for” or “explain” the influence of physical activity on mental well-being should be moderated. Given the cross-sectional nature of the data and the modest effect sizes, it would be more appropriate to frame the results as preliminary evidence supporting the plausibility of belief-related pathways. Expanding the discussion of alternative mechanisms, such as affective responses, social support, or self-efficacy, would provide a more balanced perspective and situate beliefs as one of several potential contributors rather than the sole mechanism.

 

 

Comment 6

Although you acknowledge several limitations, this section could be expanded and made more explicit. In particular, the cross-sectional design, reliance on self-report measures, potential common method bias, lack of objective physical activity assessment, and limited generalizability due to the predominantly female convenience sample deserve clearer emphasis. Recognizing the modest explained variance and the absence of covariates would also present a more cautious and scientifically balanced interpretation

 

Comment 7

The conclusion would benefit from more conservative wording to avoid implying causal or definitive mechanisms. Framing the findings as suggesting that beliefs may represent a promising psychological pathway, rather than demonstrating that they are the central mechanism, would better align with the evidence. Emphasizing the need for longitudinal or experimental research to confirm temporal ordering would also provide a constructive forward-looking perspective.

 

Author Response

Thank you for the opportunity to review your manuscript examining the mediating role of beliefs about the mental health benefits of physical activity in the relationship between physical activity intensity and mental well-being among university students.

 

Several methodological and interpretative aspects would benefit from clarification and strengthening to enhance the scientific rigor of the study.

 

General response:

We thank the reviewer for carefully reading our manuscript and for the thoughtful, constructive comments. We appreciate the reviewer’s balanced evaluation of the study and the recognition of the potential contribution of the newly developed belief scale. The suggestions helped us improve several aspects of the manuscript, particularly the clarity of the introduction, the transparency of the methodological description, and the interpretation of the mediation results.

In response to these comments, we revised the manuscript to further temper causal wording in the abstract, discussion, and conclusion. We streamlined the introduction, clarified the study’s conceptual contribution, and expanded the methodological description, including the recruitment procedure and additional information on the mediation analysis and model diagnostics. We also strengthened the reporting of the psychometric evaluation of the belief scale, including additional reliability indices, redundancy diagnostics, and the evaluation of a shortened six-item version of the scale. Furthermore, we expanded the discussion of the modest explained variance and clarified that beliefs represent only one of several possible mechanisms linking physical activity (PA) and mental well-being.

We hope these revisions adequately address the reviewer’s comments and improve the manuscript's clarity and transparency.

 

 

Comment 1:

In the abstract, I encourage you to slightly temper the causal wording used to describe the mediation results. Phrases suggesting that beliefs “fully mediate” or “explain” the relationship between physical activity and mental well-being may overstate what can be concluded from cross-sectional data. Reframing the findings as “consistent with an indirect association” or “statistically compatible with mediation” would provide a more methodologically accurate summary. It may also be helpful to briefly indicate the modest proportion of explained variance in mental well-being to give readers a clearer sense of the practical magnitude of the effects. Finally, specifying that the design is cross-sectional directly in the abstract would improve transparency.

 

Response:

The description of the mediation analysis findings has been revised throughout the manuscript to avoid implications of causal relations. Expressions suggesting causal mediation were replaced with more cautious formulations referring to indirect associations. We also clarified in the abstract that the study used a cross-sectional design and briefly noted that the mediation model explains a modest proportion of variance in mental well-being.

 

Comment 2

The introduction section feels, at times, somewhat lengthy and could benefit from tighter focus on the specific research gap addressed by the present study. You might consider streamlining parts of the historical discussion and moving more quickly toward the rationale for developing the new belief scale and testing the mediation model in university students. Clarifying the novelty of the study more explicitly (for example, by emphasizing how this belief-based mechanism differs from existing self-efficacy or affect-based explanations) would strengthen the argument for originality. Additionally, clearly stating the hypotheses in a concise and testable form at the end of the introduction would help guide the reader.

 

Response:

As suggested, the Introduction is now shortened by leaving out the description and historical overview of the broader research on determinants and outcomes of PA, and by reducing the description of theoretical models and previous studies on which the present study is grounded. It has also been streamlined and structured into four main paragraphs, describing the empirical and theoretical foundations on which the study is based, research gaps it seeks to address, objectives, hypothesis and contributions of the study. The revised introduction also more clearly emphasizes the conceptual contribution of the study, particularly the distinction between beliefs about the mental health benefits of PA and related constructs such as self-efficacy or affect-based explanations.

 

Comment 3

In the Methods section, the recruitment procedure would benefit from additional clarification. Although you specify that 339 university students aged 18–28 participated and that inclusion required being a “healthy student capable of typical physical activities,” more detail regarding how participants were approached, the recruitment channels used, and the approximate response rate would help readers better evaluate representativeness. Given that 80.2% of the sample was female, it would also be important to acknowledge the potential implications of this imbalance for generalizability and to briefly discuss possible sampling bias.

 

Because all constructs (PA intensity (GSLTPAQ), mental well-being (SWEMWBS), and beliefs) were assessed via self-report at a single time point, the risk of common method variance should be addressed more explicitly. This could be done either by reporting a statistical check (e.g., a Harman single-factor test or similar approach) or by expanding the limitations section to acknowledge the potential inflation of associations due to shared method variance.

 

The newly developed Belief in the Mental Health Benefits of Physical Activity Scale is a clear contribution, and the reported EFA and CFA results (including strong loadings and acceptable fit indices after correlating Items 1 and 2) are presented clearly. However, the very high internal consistency (α = .97) raises the possibility of item redundancy rather than optimal construct coverage. To reinforce the psychometric argument, I recommend reporting additional indices such as composite reliability and average variance extracted, as well as examining discriminant validity against conceptually related constructs. It may also be worthwhile to explore whether a shorter version of the scale could retain similar explanatory power while reducing potential overlap. Testing measurement invariance across gender or academic level would further strengthen the validation evidence.

 

With regard to the mediation analysis, the explained variance in mental well-being (7.31%) is modest and would benefit from more explicit contextualization in terms of practical significance. The current model does not include covariates such as gender, age, or academic stress, which are established correlates of student mental health and could influence the observed relationships. Providing more information about assumption testing and specifying the number of bootstrap resamples used in the PROCESS analysis would improve methodological transparency.

 

Response:

We appreciate these thoughtful and constructive comments addressing several methodological aspects of the study. As the comment raises multiple related issues concerning the recruitment procedure, potential common method variance, the psychometric evaluation of the newly developed scale, and the mediation analysis, we address these points in turn.

 

First, in response to the reviewer’s suggestion, we clarified the participant recruitment procedure in the Methods subsection Procedure. The Procedure section has been revised to provide additional clarification regarding the recruitment process. Students were invited to participate through an invitation letter distributed via project collaborators and partner universities across four Croatian universities. We have also added information about the number of students who accessed the survey link (N = 522) and the number who completed the questionnaire (N = 339), corresponding to a completion rate of 64.9%. The revised text additionally clarifies that only fully completed questionnaires were retained for analysis and that the survey platform prevented multiple submissions while preserving participant anonymity. Finally, we acknowledge the gender imbalance in the sample (80.2% female) and discuss its potential implications for generalizability and possible sampling bias in the Discussion section.

 

Second, we appreciate the reviewer’s suggestion to address potential common method variance, given that all constructs were assessed using self-report measures at a single time point. To examine this possibility, we conducted a Harman single-factor check across all self-report items entering the substantive model (three PA intensity items, 11 belief items, and seven mental well-being items). The first unrotated component accounted for 43.51% of the variance (PCA), and the first unrotated factor accounted for 42.04% (PAF), indicating that a single general factor did not account for the majority of covariance across measures. We additionally repeated this analysis using the shortened six-item belief scale, which further reduced the proportion of variance explained by the first factor (PCA: 34.40%; PAF: 31.23%). These results suggest that common method variance is unlikely to substantially bias the observed relationships. We also expanded the limitations section to acknowledge that single-source self-report data collected at one time point may inflate observed associations, even when a single-factor solution does not dominate the covariance structure.

 

Third, we appreciate the reviewer’s positive assessment of the newly developed Belief in the Mental Health Benefits of Physical Activity Scale and the helpful suggestions for strengthening its psychometric evaluation. We agree that extremely high internal consistency may reflect item redundancy rather than optimal construct coverage. To address this concern, we expanded the psychometric reporting in the Psychometric Properties section of the Results.

We also agree that testing measurement invariance across groups would further strengthen the validation evidence. However, given the scope of the present study and the available sample size, such analyses were beyond the objectives of the current manuscript and are recommended for future validation studies.

 

In addition to Cronbach’s alpha, we now report redundancy-sensitive indices including the average inter-item correlation (AIC) and corrected item–total correlations, as well as a SEM-based omega reliability estimate derived from the CFA model. For the full 11-item version, these indices indicated very high item homogeneity (α = .971; AIC = .755; corrected item–total correlations = .799–.921; SEM-based ω = .968). In addition, we report composite reliability and average variance extracted from the CFA solution (CR = .971; AVE = .754), providing further model-based evidence of reliability and convergent validity.

 

Following a content-based item review aimed at reducing conceptual overlap, we derived and evaluated a shortened six-item version of the scale. The short form retained a clear one-factor structure and strong psychometric performance (with high loadings and acceptable model fit indices reported), while showing slightly reduced item overlap (AIC = .722; corrected item–total correlations = .768–.845). Reliability remained high (α = .939; SEM-based ω = .939), and composite reliability and AVE were also strong for the shortened scale (CR = .940; AVE = .723). The revised manuscript now provides an explicit rationale for item reduction and reports psychometric results for both the original and shortened versions of the scale.

 

With respect to discriminant validity, we examined the empirical relationship between the belief scale and mental well-being. The correlation between the belief scale and mental well-being was moderate (r = .27), indicating that the constructs are related but not redundant, which provides preliminary support for their empirical distinctiveness.

At the same time, we acknowledge that discriminant validity could not be formally evaluated with the current set of constructs, as the study did not include additional conceptually related measures. We therefore note this as a limitation and suggest that future validation studies include related constructs in order to more directly examine discriminant validity.

 

Finally, we appreciate the reviewer’s suggestions regarding the mediation analysis. In response, we clarified the specification and reporting of the mediation model in the Results section. The revised manuscript now presents a structural equation model in which PA intensity was included as an observed predictor, beliefs were specified as a latent mediator, and mental well-being as a latent outcome. Model fit indices, standardized factor loadings, and structural path coefficients are now reported to improve methodological transparency. We also expanded the interpretation of the explained variance in mental well-being to better contextualize the practical significance of the findings.

 

We hope that these revisions adequately address the reviewer’s comments and improve the clarity and transparency of the methodological reporting in the manuscript.

 

 

Comment 4

In the Results section, providing slightly more guidance regarding the magnitude of the observed effects would help readers better interpret their practical meaning. For example, the mediation model accounts for 7.31% of the variance in mental well-being. While this effect is statistically significant, it is relatively small in magnitude, and briefly contextualizing this value, such as noting that beliefs explain only a modest proportion of variability in students’ well-being, would allow readers to better assess the practical importance of the findings.

 

When presenting the mediation model, it would also be helpful to distinguish more explicitly between standardized and unstandardized coefficients. In Figure 1 and in the text, please ensure that it is clearly specified whether the reported β values are standardized and that labeling is consistent throughout. This will make it easier for readers to compare the relative strength of the paths (e.g., PA → belief, belief → well-being, and the indirect effect).

 

While confidence intervals are reported for several paths, they should be presented consistently for all key effects, including the total, direct, and indirect effects, to provide a complete picture of precision. Clarifying whether the assumptions underlying the analyses were checked (such as normality of residuals, absence of multicollinearity, and suitability for mediation modeling) would further strengthen the transparency and robustness of the statistical reporting.

 

Response

We see this comment very constructive. We agree that providing clearer guidance regarding the magnitude and interpretation of the observed effects can help readers better assess their practical significance. In response to this suggestion, we revised the Results section to more explicitly contextualize the magnitude of the observed effects. In particular, we now note that the mediation model explains a relatively modest proportion of variance in mental well-being, indicating that beliefs about the mental health benefits of PA represent only one of several factors contributing to students’ well-being.

We also clarified the reporting of standardized and unstandardized coefficients throughout the Results section and in Figure 1. The manuscript now consistently distinguishes between unstandardized coefficients (with standard errors and p-values) and standardized coefficients (β), which are presented to facilitate comparison of the relative strength of the structural paths in the model.

In addition, confidence intervals are now reported consistently for all key effects in the mediation model, including the direct, indirect, and total effects. As recommended, these intervals are based on bias-corrected and accelerated (BCa) bootstrap estimates.

Finally, we expanded the description of the analytic procedure to clarify the estimation approach and model evaluation criteria used in the SEM mediation analysis. The manuscript now specifies the estimation approach and model diagnostics used in evaluating the SEM mediation model, thereby improving the transparency of the statistical reporting. In addition, standard SEM diagnostics were examined to ensure the suitability of the model estimation.

We hope these revisions make the interpretation of the results clearer and provide a more complete presentation of the statistical evidence.

 

 

Comment 5

The Discussion section is conceptually strong; however, some interpretations appear stronger than warranted by the design. Statements suggesting that beliefs “fully account for” or “explain” the influence of physical activity on mental well-being should be moderated. Given the cross-sectional nature of the data and the modest effect sizes, it would be more appropriate to frame the results as preliminary evidence supporting the plausibility of belief-related pathways. Expanding the discussion of alternative mechanisms, such as affective responses, social support, or self-efficacy, would provide a more balanced perspective and situate beliefs as one of several potential contributors rather than the sole mechanism.

 

Response

The description of the mediation analysis findings has been revised throughout the manuscript to avoid implications of causal relations. Expressions suggesting that beliefs “fully explain” or “account for” the relationship between PA and mental well-being were replaced with more cautious formulations referring to an indirect association compatible with mediation. In addition, the Discussion now acknowledges that beliefs about the mental health benefits of physical activity may represent only one possible pathway linking PA and mental well-being. The interpretation of the findings is therefore situated within broader theoretical perspectives that emphasize affective responses and cognitive interpretations of PA experiences, and future studies are encouraged to examine these and other potential mechanisms more directly.

 

Comment 6

Although you acknowledge several limitations, this section could be expanded and made more explicit. In particular, the cross-sectional design, reliance on self-report measures, potential common method bias, lack of objective physical activity assessment, and limited generalizability due to the predominantly female convenience sample deserve clearer emphasis. Recognizing the modest explained variance and the absence of covariates would also present a more cautious and scientifically balanced interpretation

 

Response

The Limitations subsection of the Discussion has been expanded to more explicitly acknowledge the cross-sectional design, reliance on self-report measures, the potential influence of common method variance, the absence of objective PA measures, the predominatnly female sample, the modest explained variance, and the absence of covariates in the mediation model 

 

Comment 7

The conclusion would benefit from more conservative wording to avoid implying causal or definitive mechanisms. Framing the findings as suggesting that beliefs may represent a promising psychological pathway, rather than demonstrating that they are the central mechanism, would better align with the evidence. Emphasizing the need for longitudinal or experimental research to confirm temporal ordering would also provide a constructive forward-looking perspective.

 

Response

The conclusion has been rephrased to avoid implications of causal relations. The need for longitudinal studies to confirm temporal ordering has been described in suggestions for future studies in the Discussion.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have responded constructively to the first-round critique, and several concerns have been adequately resolved. However, substantive issues remain insufficiently addressed and must be resolved before the manuscript can be accepted.

1. The acknowledgment of this limitation remains confined to the limitations section. For a scale being introduced to the literature, this caveat must appear prominently in 2.3.3 and 3.2, with all psychometric conclusions explicitly qualified as preliminary pending independent cross-validation. Readers consulting the psychometric results in isolation will otherwise overestimate the stability of the reported factor structure and fit indices.
2. Harman's single-factor test is reported, but its well-documented insensitivity is not acknowledged. The authors should either apply a more defensible diagnostic within the existing SEM framework, such as an unmeasured latent method factor, or provide an analytically grounded discussion of which specific associations are most vulnerable to CMV inflation, particularly the belief-well-being path on which the mediation conclusion depends. 
3. The standardised indirect effect remains small, and the practical implications section continues to recommend specific intervention modalities, workshops, digital tools, self-reflection modules, with a degree of specificity that exceeds what the evidence supports. Recommendations should be explicitly framed as tentative and contingent on experimental and longitudinal replication.
4. The term “PA intensity” persists in several locations where “leisure-time PA score” or “self-reported PA level” would be more precise given the instrument's formative composite nature. A global search-and-replace is recommended.
5. The justification for using raw rather than Rasch-transformed SWEMWBS scores remains minimal. One additional sentence clarifying that transformed scores were not required for the present analytical purpose would suffice.

Author Response

REVIEWER 2

 

The authors have responded constructively to the first-round critique, and several concerns have been adequately resolved. However, substantive issues remain insufficiently addressed and must be resolved before the manuscript can be accepted.

 

General response:

 

We sincerely thank the reviewer for the careful and constructive round 2 evaluation of our manuscript. We particularly appreciate the emphasis on methodological clarity, interpretative caution, and alignment between empirical findings and their presentation, which has helped us to further strengthen the manuscript in several important ways. We hope that the revisions have adequately addressed the reviewer’s concerns and improved the manuscript's clarity and robustness. Of course, we remain available for any further clarification if needed.

 

1. The acknowledgment of this limitation remains confined to the limitations section. For a scale being introduced to the literature, this caveat must appear prominently in 2.3.3 and 3.2, with all psychometric conclusions explicitly qualified as preliminary pending independent cross-validation. Readers consulting the psychometric results in isolation will otherwise overestimate the stability of the reported factor structure and fit indices.

 

Response:

Appreciating the emphasis on the importance of methodological clarity in reporting newly developed scales, we fully agree that, given that the scale was developed and evaluated within a single sample, the psychometric findings should be clearly qualified as preliminary pending independent cross-validation. In response, we have revised Sections 2.3.3 and 3.2 to explicitly acknowledge this limitation and to frame the reported psychometric results as preliminary. Specifically, in Section 2.3.3.1, we added a statement at the beginning of the scale development description indicating that the findings should be considered preliminary and require independent cross-validation, and we also retained and clarified this qualification at the end of the section. In addition, in Section 3.2, we revised the opening paragraph to explicitly state that all reported psychometric findings and conclusions are preliminary, should be interpreted with caution, and require independent cross-validation before firm conclusions about the scale’s structure and validity can be drawn. Furthermore, these statements are positioned prominently at the beginning of each section to ensure that readers interpreting the psychometric findings in isolation are clearly informed of their preliminary nature.

 

 

2. Harman's single-factor test is reported, but its well-documented insensitivity is not acknowledged. The authors should either apply a more defensible diagnostic within the existing SEM framework, such as an unmeasured latent method factor, or provide an analytically grounded discussion of which specific associations are most vulnerable to CMV inflation, particularly the belief-well-being path on which the mediation conclusion depends. 

 

Response:

We appreciate the reviewer’s emphasis on this issue and fully recognise its importance. We acknowledge the limited sensitivity of Harman’s single-factor test and have clarified this in the manuscript. We addressed this concern by complementing the initial diagnostic with an analytically grounded interpretation of the results. Specifically, we added a brief note in the Results section indicating that the observed associations may be partly influenced by shared method variance, and expanded the Discussion to consider its potential impact on the key mediation pathway between belief and mental well-being. Importantly, the modest magnitude of this association and the observed indirect-only mediation pattern, including the absence of a direct association between PA level and mental well-being, are less consistent with a generalised inflation effect that would be expected under strong common method bias. Taken together, these considerations suggest that, although common method variance cannot be fully ruled out, it is unlikely to account for the observed pattern of relationships.

 

 

 

3. The standardised indirect effect remains small, and the practical implications section continues to recommend specific intervention modalities, workshops, digital tools, self-reflection modules, with a degree of specificity that exceeds what the evidence supports. Recommendations should be explicitly framed as tentative and contingent on experimental and longitudinal replication.

 

Response:

The practical implications section has been revised to describe beliefs about the mental health benefits of PA as one possible avenue for enhancing PA's effects on students’ mental well-being. Only a broad suggestion on how these beliefs might be strengthened is offered, and recommendations of specific intervention modalities are removed.   Additionally, to further ensure an appropriately cautious interpretation of the findings, we have revised the structure of the Discussion by presenting the study limitations prior to the practical implications. This allows the implications to be more explicitly interpreted in light of the study’s methodological constraints. Correspondingly, the opening of the Practical Implications section was revised to explicitly acknowledge these limitations.

 

4. The term “PA intensity” persists in several locations where “leisure-time PA score” or “self-reported PA level” would be more precise given the instrument's formative composite nature. A global search-and-replace is recommended.

 

Response:

The term “PA intensity” has been replaced with the term “PA level” throughout the manuscript, and the term “PA level” is explained in more detail in the Method section. Specifically, a clarifying sentence was added in the Methods section indicating that the LSI reflects self-reported leisure-time PA and represents overall PA level, rather than a direct measure of PA intensity.

 

5. The justification for using raw rather than Rasch-transformed SWEMWBS scores remains minimal. One additional sentence clarifying that transformed scores were not required for the present analytical purpose would suffice.

 

Response:

Following the reviewer’s earlier comment, we now better understand the intention behind clarifying the scoring approach for SWEMWBS and have revised the Methods section accordingly. We believe this clarification improves transparency and provides a more informative description of the scoring approach. We therefore appreciate this helpful suggestion. We specify that although a Rasch-based transformation of total scores is available, it was not required for the present analytical purpose. Given that the primary analyses were conducted using latent variable modelling, which focuses on associations among constructs rather than on metric scaling of observed total scores, the use of raw scores was not expected to meaningfully affect the estimation or interpretation of the results.

 

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The revised manuscript shows clear improvements and the authors have addressed several of my previous comments, particularly in moderating causal language and clarifying that the study uses a cross-sectional design. However, some important issues appear to have been addressed only partially and further minor revisions are still required to strengthen the clarity and methodological transparency of the manuscript.

 

First, the introduction would benefit from a clearer and more explicit formulation of the study hypotheses. Although the theoretical background is presented, the hypotheses are not sufficiently articulated at the end of the introduction. To improve clarity and guide the reader, the authors should explicitly state their hypotheses in a concise form. For example, the section could conclude with a statement such as: “We hypothesized that: (H1) physical activity intensity would be associated with stronger beliefs regarding the mental health benefits of physical activity, and (H2) these beliefs would mediate the association between physical activity intensity and mental well-being.” This type of explicit formulation would make the conceptual model and analytical strategy easier to understand.

 

Second, the methodology section still requires additional clarification regarding participant recruitment and potential sampling bias. The manuscript reports a sample with a strong gender imbalance (approximately 80% female participants), yet the implications of this imbalance are not sufficiently discussed. The authors should describe more clearly how participants were recruited (e.g., online survey dissemination, institutional mailing lists, convenience sampling, or voluntary participation through social media). In addition, they should acknowledge the potential bias associated with the gender distribution and explain how it may affect the generalizability of the findings. A concrete improvement would be to include gender and age as covariates in the structural model or to perform sensitivity analyses controlling for these demographic variables. Alternatively, the authors could explicitly discuss this limitation in the discussion section and clarify that the findings may be more representative of female respondents.

 

Furthermore, the statistical reporting could be strengthened by providing a clearer interpretation of the model’s explanatory power. The reported coefficient of determination (R² = 7.31%) suggests that the model explains only a modest proportion of the variance in mental well-being. While statistically significant relationships may still be meaningful, this relatively small effect size should be explicitly acknowledged and interpreted in the discussion section. For instance, the authors could note that the results indicate a statistically significant but limited explanatory contribution of physical activity intensity and related beliefs, suggesting that other psychological, social, or contextual factors likely play an important role in explaining mental well-being. Including such an interpretation would provide a more balanced and transparent discussion of the findings.

Author Response

REVIEWER 3

 

The revised manuscript shows clear improvements and the authors have addressed several of my previous comments, particularly in moderating causal language and clarifying that the study uses a cross-sectional design. However, some important issues appear to have been addressed only partially and further minor revisions are still required to strengthen the clarity and methodological transparency of the manuscript.

 

General response:

 

We sincerely thank the reviewer for the constructive second-round evaluation and for recognising the improvements made to the manuscript, particularly regarding the moderation of causal language and clarification of the study design. We also appreciate the additional suggestions aimed at improving the manuscript's clarity and methodological transparency.

We have carefully addressed all remaining comments by further clarifying the formulation of study hypotheses, expanding the description of participant recruitment and sampling considerations, and strengthening the interpretation of the model’s explanatory power within the Discussion. We have also ensured that study limitations, particularly those related to sample characteristics, are more clearly reflected in the Discussion's interpretation of the findings.

We hope these revisions have adequately addressed the reviewer’s concerns and improved the manuscript's overall clarity and transparency. We certainly remain available for any further clarification if needed.

 

First, the introduction would benefit from a clearer and more explicit formulation of the study hypotheses. Although the theoretical background is presented, the hypotheses are not sufficiently articulated at the end of the introduction. To improve clarity and guide the reader, the authors should explicitly state their hypotheses in a concise form. For example, the section could conclude with a statement such as: “We hypothesized that: (H1) physical activity intensity would be associated with stronger beliefs regarding the mental health benefits of physical activity, and (H2) these beliefs would mediate the association between physical activity intensity and mental well-being.” This type of explicit formulation would make the conceptual model and analytical strategy easier to understand.

 

Response:

Study hypotheses are now clearly stated at the end of the Introduction.

 

Second, the methodology section still requires additional clarification regarding participant recruitment and potential sampling bias. The manuscript reports a sample with a strong gender imbalance (approximately 80% female participants), yet the implications of this imbalance are not sufficiently discussed. The authors should describe more clearly how participants were recruited (e.g., online survey dissemination, institutional mailing lists, convenience sampling, or voluntary participation through social media). In addition, they should acknowledge the potential bias associated with the gender distribution and explain how it may affect the generalizability of the findings. A concrete improvement would be to include gender and age as covariates in the structural model or to perform sensitivity analyses controlling for these demographic variables. Alternatively, the authors could explicitly discuss this limitation in the discussion section and clarify that the findings may be more representative of female respondents.

 

Response:

Acknowledging the value of this comment, we fully agree that clear reporting of participant recruitment, potential sampling bias, and the role of demographic variables is essential for maintaining methodological rigour and for ensuring an appropriate interpretation of the study’s theoretical contribution, particularly when introducing a newly developed scale. In response, we have revised the Methods section (Procedure) to provide a more detailed and explicit description of the recruitment procedure. Specifically, we clarified that participants were recruited via an online survey distributed through project collaborators and partner universities across four institutions, primarily via institutional mailing lists and student networks associated with the project, reflecting a voluntary, convenience-based sampling approach. We also explicitly acknowledge the potential for self-selection bias inherent in this recruitment strategy.

We also considered including demographic covariates. We agree that including variables such as gender could provide additional insight into the robustness of the observed relationships. In the present sample, the substantial gender imbalance limits the interpretability of such adjustments and may affect the stability of model estimates. Age was also not included as a covariate, as the sample is relatively homogeneous. Nevertheless, examining its potential role would benefit from future studies using larger and more diverse samples. Accordingly, we addressed this issue by explicitly acknowledging the gender imbalance and its implications in the Discussion section. Specifically, we now clarify that the findings may be more representative of female respondents and should be interpreted with caution when generalizing to male student populations, and we briefly discuss the sampling context that may have contributed to this distribution.

Finally, we note that future research should examine the robustness of the observed relationships across demographic groups, ideally using more balanced samples and including relevant covariates or conducting sensitivity analyses.

 

Furthermore, the statistical reporting could be strengthened by providing a clearer interpretation of the model’s explanatory power. The reported coefficient of determination (R² = 7.31%) suggests that the model explains only a modest proportion of the variance in mental well-being. While statistically significant relationships may still be meaningful, this relatively small effect size should be explicitly acknowledged and interpreted in the discussion section. For instance, the authors could note that the results indicate a statistically significant but limited explanatory contribution of physical activity intensity and related beliefs, suggesting that other psychological, social, or contextual factors likely play an important role in explaining mental well-being. Including such an interpretation would provide a more balanced and transparent discussion of the findings.

 

Response:

The coefficients of determination are now explicitly repeated in the Discussion, under the description of study limitations, and the limited explanatory power of the model is additionally stressed. Also, the incorporation of other psychological and contextual variables in the model is added as a suggestion for future studies.

Author Response File: Author Response.pdf