Relative Advantage and Compatibility as Drivers of ChatGPT Adoption in Latin American Higher Education: A PLS SEM Study Towards Sustainable Digital Education

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This study investigates the factors influencing the adoption of ChatGPT among higher education students in Latin America, framed within the Diffusion of Innovations (DOI) theory. Drawing on a survey of 792 undergraduates from Peru, Chile, Bolivia, Argentina, and Colombia, the research examines the roles of relative advantage, compatibility, complexity, trialability, and observability in shaping students’ attitudes toward ChatGPT, and how these attitudes, in turn, affect their intention to use the tool. Using a cross-sectional quantitative design and Partial Least Squares Structural Equation Modeling (PLS-SEM), the findings reveal that relative advantage (β = 0.247) and compatibility (β = 0.246) exert the strongest positive effects on attitude, followed by trialability, observability, and complexity. Attitude emerges as a strong predictor of usage intention (β = 0.777), explaining 60.4% of its variance. The results highlight that adoption is driven more by perceived academic benefits and alignment with existing learning practices than by technical ease, underscoring the need for institutions to provide structured opportunities for experimentation, clear ethical guidelines, and integration of AI competencies into curricula to foster sustainable digital education in the region.

However, I have the following suggestions for improving the manuscript:

1. In the abstract: include explicit mention of the sample size and country coverage earlier in the abstract for context.

In the Introduction 

1. Explicitly articulate what is missing in existing ChatGPT adoption research in Latin America (most of your justification is global or generic).
2. Explain more clearly how ChatGPT adoption aligns with SDG 4 (Quality Education) and what “sustainable digital education” means in measurable or conceptual terms.

In LR

1. Some DOI construct descriptions repeat information from Table 1—condense to avoid redundancy.
2. Strengthen the contrast between Latin America and other regions in terms of technological infrastructure, policy, and student readiness.
3. Strengthen the LR with recent studies in the field and ethics of ChatGpt such as:
   1. George Reyes, C.E., Avello Martínez, R. and Buenestado Fernández, M., 2025. Perceptions of ChatGPT and the Complexity of Its Impact Among Higher Education Students: Evidence Across Ten Countries of Latin America and Europe.
   2. Al-Kfairy, M., Mustafa, D., Kshetri, N., Insiew, M. and Alfandi, O., 2024, September. Ethical challenges and solutions of generative AI: An interdisciplinary perspective. In Informatics (Vol. 11, No. 3, p. 58). Multidisciplinary Digital Publishing Institute.
   3. and many others
4. Explicitly number and bold hypotheses (H1–H6) for quick scanning.

Methods

1. Simple random sampling is claimed, but recruitment via institutional channels and WhatsApp groups is more like convenience sampling with random elements. This needs clarification to avoid misrepresentation.

Discussion

1. Move beyond confirming prior findings—highlight where your results diverge from other regions or contradict earlier studies.
2. Explicitly address why complexity had a positive rather than negative effect (interesting theoretical implication).

Good Luck

Author Response

We have taken on board all the suggestions and made the changes indicated, ensuring that the changes reflect the comments made. We thank the reviewer–editor for his comments, which have helped to improve the quality and clarity of the work.

"We used the journal’s English editing service to enhance the quality of this manuscript."

1. Point-by-Point Response to Author Comments and Suggestions

Comments 1: [1. In the abstract: include explicit mention of the sample size and country coverage earlier in the abstract for

Response 1. We appreciate the reviewer’s helpful comment. We have incorporated the requested information into the first sentence of the Abstract so that readers immediately see the sample size (n = 792) and country coverage (Peru, Chile, Bolivia, Argentina, and Colombia) while preserving the Abstract’s concision and flow [16,17].

2. Point-by-Point Response to Author Comments and Suggestions

Comment 2: [What 1. Explicitly articulate what is missing in existing ChatGPT adoption research in Latin America (most of your justification is global or generic)].

Response 2: We appreciate the reviewer’s observation since it enabled us to more clearly articulate the regional gaps in Latin America regarding ChatGPT adoption; justify the exploratory use of the diffusion of innovations framework; and specify the study’s empirical, context-grounded contribution. These improvements have been incorporated into the Introduction [124-142].

3. Point-by-Point Response to Author Comments and Suggestions

Comment 3: [3. Explain more clearly how ChatGPT adoption aligns with SDG 4 (Quality Education) and what “sustainable digital education” means in measurable or conceptual terms.

Response 3 : Thank you for the comment. We addressed it by concisely defining sustainable digital education as technology-mediated learning (including generative AI) that improves quality with inclusion, efficiency, and integrity, and by clarifying its general alignment with SDG 4 through monitoring and real-time feedback, thereby supporting the relevance of adopting ChatGPT  [104-117].

4. Point-by-Point Response to Author Comments and Suggestions

Comments 4: [Some DOI construct descriptions repeat information from Table 1—condense to avoid redundancy.]

Response 4: Thank you for the observation. We reviewed it carefully and agree with the identified redundancy. We have already implemented the changes: Table 1 now presents concise, single-line definitions, and redundant definitions have been removed from the text. Accordingly, the observation has been resolved [180-181].

5. Point-by-Point Response to Author Comments and Suggestions

Comments 5: [Strengthen the contrast between Latin America and other regions in terms of technological infrastructure, policy, and student readiness.]

Response 5: Thank you for your comment on “strengthening the contrast between Latin America and other regions in terms of technological infrastructure, policies, and student preparedness.” We confirm that the correction has been made exactly as indicated [124,150].

6. Point-by-Point Response to Author Comments and Suggestions

Comments 6: [ Strengthen the LR with recent studies in the field and ethics of ChatGpt such as: 1. George Reyes, C.E., Avello Martínez, R. and Buenestado Fernández, M., 2025. Perceptions of ChatGPT and the Complexity of Its Impact Among Higher Education Students: Evidence Across Ten Countries of Latin America and Europe.]

Response 6: We appreciate the comment. We have addressed it by strengthening the literature review with recent studies in the field and the ethics of ChatGPT, including the reference George Reyes, C. E., Avello Martínez, R., & Buenestado Fernández, M. (2025). Perceptions of ChatGPT and the complexity of its impact among higher education students: Evidence from ten countries in Latin America and Europe. This work has already been cited in the literature review and added to the reference list.

7. Point-by-Point Response to Author Comments and Suggestions

Comments 7: [Explicitly number and bold hypotheses (H1–H6) for quick scanning.]

Response 7: We appreciate the comment. We have addressed it: the hypotheses are now explicitly numbered and bolded as H1–H6 to enable quick reading throughout the manuscript.

8. Point-by-Point Response to Author Comments and Suggestions

Comments 8: [1. Simple random sampling is claimed, but recruitment via institutional channels and WhatsApp groups is more like convenience sampling with random elements. This needs clarification to avoid misrepresentation.

Response 8: We appreciate your comment. We agree with your assessment: the procedure corresponds to non-probabilistic convenience sampling (via institutional channels and WhatsApp groups), not to simple random sampling. The text has been revised in Section 2.2 (Participants), and the observation has been fully addressed. Thank you [305-310].

9. Point-by-Point Response to Author Comments and Suggestions

Comments 9: [Move beyond confirming prior findings—highlight where your results diverge from other regions or contradict earlier studies.

Response 9: Thank you for the observation. We have updated the Discussion Section to meet the requirement: the results were reviewed and corrected, and the discussion was rewritten to emphasize the discrepancies with the reference study (rather than confirming findings). We appreciate the valuable feedback [635-727].

10. Point-by-Point Response to Author Comments and Suggestions

Comments 10: [Move Explicitly address why complexity had a positive rather than negative effect (interesting theoretical implication).

Response 10: Thank you for the observation. We have updated the Discussion to explain why complexity showed a positive effect [686-695].

 

 

 

Reviewer 2 Report

Comments and Suggestions for Authors

The study is interesting, but it needs to be revised in several sections:

1. Your study includes a user survey. Please provide the ethical approval document as supplementary material.

2. Your paper only mentions ChatGPT while ignoring Gemini (likely a typo: Gemmi → Gemini), Claude, Grok, and other LLMs. It therefore lacks background literature and does not explain why ChatGPT was chosen instead of other models. Please explain the selection rationale in the Introduction and Literature Review (e.g., availability, popularity among respondents, institutional access, task fit), and add citations comparing relevant LLMs.

3. Small writing correction: Diffusion of Innovations Theory (DOI) should be written as Diffusion of Innovations (DOI) theory. After the first full mention, use DOI theory (for tables and subsequent text) — you do not need to repeat the full phrase each time.

4. The study pools data from multiple countries without testing configural, metric, and scalar invariance. If measurement properties differ across contexts, pooled PLS-SEM estimates can be misleading. For a multi-country study claiming Latin American generalisability, testing measurement invariance — or at minimum running multigroup analyses — is essential. The manuscript does not report such tests.

Author Response

We appreciate the valuable comments and have addressed all of the reviewer's suggestions, which have contributed to the improvement of the manuscript.

"We used the journal’s English editing service to enhance the quality of this manuscript."

Comments 1: [Your study includes a user survey. Please provide the ethical approval document as supplementary material]

Response 1. We appreciate your observation. Accordingly, we confirm that the study was reviewed by the Research Ethics Committee of the Universidad Nacional Jorge Basadre Grohmann and approved under Certificate No. 2024-069-CEIUNJBG, dated October 24, 2024. The certificate is attached.

Comment 2: [Your paper only mentions ChatGPT while ignoring Gemini (likely a typo: Gemmi → Gemini), Claude, Grok, and other LLMs. It therefore lacks background literature and does not explain why ChatGPT was chosen instead of other models. Please explain the selection rationale in the Introduction and Literature Review (e.g., availability, popularity among respondents, institutional access, task fit), and add citations comparing relevant LLMs]

Response 2: We appreciate the comment, so we have added a new paragraph to the Introduction Section, where recent scientific references comparing the performance of these AI models have been added. In addition, the choice of ChatGPT as the subject of study is justified. Likewise, the typographical error has been corrected and we have replaced Gemmi with Gemini [62-76]

Comment 3: [Small writing correction: Diffusion of Innovations Theory (DOI) should be written as Diffusion of Innovations (DOI) theory. After the first full mention, use DOI theory (for tables and subsequent text) — you do not need to repeat the full phrase each time].

Response 3: We appreciate your comment. We have revised the wording as suggested: on first mention we use “diffusion of innovations theory (DOI)” and, in subsequent mentions, “DOI theory.” We also confirm that we used the journal’s English translation and copyediting service; the corresponding certificate is attached.

Comments 4: [The study pools data from multiple countries without testing configural, metric, and scalar invariance. If measurement properties differ across contexts, pooled PLS-SEM estimates can be misleading. For a multi-country study claiming Latin American generalisability, testing measurement invariance — or at minimum running multigroup analyses — is essential. The manuscript does not report such tests.]

Response 4: We appreciate your comment. In the revised version of the manuscript, we have incorporated the MICOM procedure proposed by Henseler et al. (2016), which is recommended for evaluating invariance in PLS-SEM models. The results can be seen in Table 10, which shows that most constructs comply with the first two steps, confirming partial invariance [429-436,589-632, -713-727].

Given the partial invariance, a multigroup analysis (MGA) has also been added for paths using the Bootstrap-MGA approach. The results are presented in Table 11 [429-436,589-632, -713-727].

Likewise, a paragraph has been added to the discussion of the results regarding compositional invariance and multigroup analysis, reinforcing the generalization of the measurement model [429-436,589-632, -713-727].

Finally, a paragraph has also been added on statistical analysis to explain the verification of the validity of comparison between countries using the MICOM and MGA procedures [429-436,589-632, -713-727].

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript is very well written, with a clear introduction and well-developed arguments regarding the topic.

However, in Sections 4 Discussion and 5 Conclusion, the connection to the underlying theoretical frameworks is not sufficiently sustained. To strengthen the paper’s value, the findings should be more explicitly linked to the theories, so that the contribution of this emerging AI topic to the advancement of existing theory becomes evident.

Author Response

We have taken on board all the suggestions and made the changes indicated, ensuring that the changes reflect the comments made. We thank the reviewer-editor for his comments, which have helped to improve the quality and clarity of the work.

"We used the journal’s English editing service to enhance the quality of this manuscript."

 Point-by-point response to Author Comments and Suggestions

Comments 1: [However, in Sections 4 Discussion and 5 Conclusion, the connection to the underlying theoretical frameworks is not sufficiently sustained. To strengthen the paper’s value, the findings should be more explicitly linked to the theories, so that the contribution of this emerging AI topic to the advancement of existing theory becomes evident...]

Response 1. Thank you for the observation. We have revised Sections 4 (Discussion) and 5 (Conclusion) to strengthen the explicit linkage between our findings and the underlying theoretical frameworks, clarifying how our results contribute to the advancement of existing theory and highlighting their practical implications. We appreciate the valuable feedback [635,782].

 

 

 

 

Reviewer 4 Report

Comments and Suggestions for Authors

This study entitled ‘Relative Advantage and Compatibility as Drivers of ChatGPT 2 Adoption in Latin American Higher Education: A PLS-SEM 3 Study Toward Sustainable Digital Education’ is well documented and presented in a coherent manner in relation to the methodological design employed. The study is novel in relation to the research problem.

In short, the study addresses a current and necessary topic such as sustainability, as it is an area of great interest and the article touches on very relevant points.

The data are presented with detailed procedures and an explanation of the analysis and methodologies used in the study. The suggestions for future research seem valuable and insightful. Below, I offer a more detailed review of the manuscript.

To this end, I will comment on the different sections.

The abstract of the article is not adequate, as the authors should specify the objectives of the study explicitly. I suggest that the objectives be introduced.

The keywords are also adequate.

The title of the article is appropriate for the content of the manuscript. The document is well structured and contains all the sections of a scientific manuscript, maintaining consistency between the objectives, methodology and results, making it easy for readers to understand the study. It uses vocabulary that is understandable to a wide audience, without excessive technical terms. Another positive aspect is that you not only describe problems, but also propose possible solutions or reflections.

The introduction details the basis on which the study is based. The literature review is consistent with the problem under study. Thus, the works referenced are current and novel. The contribution that this study would make to research in relation to the growing importance of digital transformation in education is also highlighted, making it crucial to understand how emerging technologies can effectively support teaching and learning.

Method: It is based on the Theory of Diffusion of Innovations (DOI), which provides a clear conceptual framework for studying the adoption of ChatGPT. On the other hand, the use of a correlational-explanatory and cross-sectional design is appropriate for exploring relationships between variables in a context where evidence is still limited.

With regard to the data collection instrument, it was constructed from previous instruments and underwent pilot testing and expert validation, reinforcing its validity and reliability.

In the data analysis, I consider that the use of PLS-SEM is relevant for exploratory models with multiple constructs, which strengthens the statistical analysis.

The results are correctly explained and represented in Tables 5, 6, 7, 8, 9 and Figure 1 on structural model estimation. In general, all the tables used contribute to facilitating the reader's understanding. This evaluator considers that the results shown in relation to the study problem are presented appropriately in relation to each of the questions studied.

The discussion is carried out correctly with the various results of the literature review.

The conclusions are derived from the results and the limitations are explicitly stated.

This reviewer considers section 7 on implications to be a very positive aspect.

With regard to ethical aspects, it should be noted that measures of informed consent, anonymity, and confidentiality were applied, which gives the study ethical robustness.

However, I recommend that a special section be devoted to the conclusions, as this will facilitate the reader's understanding.

The authors explicitly state the limitations of their study.

In short, I consider this to be a good piece of work, with a clear justification of the problem and an innovative use of technology, which will undoubtedly contribute to a greater understanding of the influence of certain factors that affect the adoption of artificial intelligence among higher education students in Latin America.

 

 

Author Response

We have taken on board all the suggestions and made the changes indicated, ensuring that the changes reflect the comments made. We thank the reviewer-editor for his comments, which have helped to improve the quality and clarity of the work.

"We used the journal’s English editing service to enhance the quality of this manuscript."

 

1. Point-by-point response to Author Comments and Suggestions

Comments 1: The abstract of the article is not adequate, as the authors should specify the objectives of the study explicitly. I suggest that the objectives be introduced.

Response 1. We appreciate the observation that the abstract did not explicitly state the study’s objectives. We have addressed this comment by explicitly including the objectives in the abstract, as suggested. [15-31]

Comments 2: However, I recommend that a special section be devoted to the conclusions, as this will facilitate the reader's understanding.

Response 1. We appreciate the suggestion to have a dedicated conclusions section to aid reader comprehension. We have addressed this comment by improving the existing “Conclusions” section—clarifying the synthesis of findings, highlighting the implications, and strengthening alignment with the study objectives—as suggested. . [730-782]

 

 

 

 

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors handled all my concerns. happy to accept.

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have correctly responded my concerns.