Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The paper presents high quality of research and overall meritoric level. It was a great pleasure to have the opportunity to review.

The methodology is well described and all the methods used in the paper are suitable. The sample of analyzed population characterized by regional dimension and regards peach farmers in Qingdao. As he the partial scope of digital learning channels is included. The local scope of the research makes it difficult to generalize the results to other regions or sectors, but the methods used can be used in further studies in this field.

I have only a few minor general remarks:

-the literature review could be more concise to avoid redundancy between the theoretical framework and the hypothesis development parts

-there is no doi, or any other links included to the references.

Author Response

Comments 1：The literature review could be more concise to avoid redundancy between the theoretical framework and the hypothesis development parts.

Response 1：Thank you very much for this constructive suggestion. We agree that the literature review section in the original manuscript partly duplicated content from the theoretical framework and hypothesis development sections. In the revised version, we have streamlined the literature review and removed redundant descriptions of key concepts and mechanisms that were repeated later in the theoretical framework and hypothesis development. Specifically, we (1) condensed overlapping discussions of social learning, risk attitudes, and farmers’ e-commerce participation, and (2) kept the more detailed conceptual argumentation in the theoretical framework and hypothesis development, while making the literature review more focused and concise. These revisions can be found in the revised manuscript on Section 1 “Introduction”.

Comments 2：There is no doi, or any other links included to the references.

Response 2：Thank you very much for pointing this out. We agree that in the original version the References list did not include DOIs or web links, which is not fully consistent with the journal’s current reference requirements. In the revised manuscript, we have carefully checked all references and added DOIs or URLs wherever they are available.

Reviewer 2 Report

Comments and Suggestions for Authors

This article examines social learning and how it influences smallholder farmers' decisions to adopt e-commerce in agricultural products. For this purpose, the Heckman two-stage model, IV-Probit, and mediation effect models are used. The topic of the study is particularly interesting! However, the study has some weaknesses, which should be addressed in order for the article to be re-evaluated for possible publication. Below are some comments, per section:

Abstract

The abstract is long. Specifically, it exceeds 200 words, which is not in accordance with the journal's guidelines. The keywords used are almost the same words as the title. It is suggested that these be differentiated as much as possible.

Introduction

The introduction lacks a variety of bibliographical references. It should be further enriched. In the penultimate paragraph of the section, you obviously want to refer to the gap that exists in the existing bibliography and the contribution of your research. Make this clearer.

In the last paragraph, you refer to research questions. However, these questions have not been clearly mentioned above. This point should be improved.

Materials and Methods

2.1. Literature Review

Although the proposed structure of the journal is followed. The bibliographic review does not particularly fit into the Materials and Methods section. You should see how you will deal with this. Also, enrichment with more bibliographic references is suggested. It seems poorly documented.

2.2.2. The Mediating Effect of Risk Attitudes

Figure 1 should be better explained, through text.

2.3.3.Sample Selection and Data Sources

In the Materials and Methods section, it is good to clearly define the aim and objects of the study.

You can add exactly when the primary data collection was conducted.

Discussion

In the discussion section, one would expect you to compare your results with those of other research. This needs to be improved. Also, a separate section with the conclusions should be created. Future research directions can be integrated into the conclusions section.

Author Response

Comments1：The abstract is long. Specifically, it exceeds 200 words, which is not in accordance with the journal's guidelines. The keywords used are almost the same words as the title. It is suggested that these be differentiated as much as possible.

Response 1：Thank you very much for this helpful comment. We agree that the original abstract was too long and exceeded the 200-word limit required by the journal. We also acknowledge that the original keywords overlapped too much with the title. In the revised manuscript, we have shortened the abstract to comply with the journal’s guidelines (now approximately 180–190 words) by removing redundant background information and overly detailed methodological descriptions, while keeping the core research question, data, methods, main findings, and contributions. In addition, we have revised the keywords to avoid simple repetition of the title and to better reflect the key concepts and methodological focus of the study. These changes can be found in the revised manuscript on page 1,“Abstract”section (lines 10–26) and “Keywords” (lines 29-30).

Comments 2：The introduction lacks a variety of bibliographical references. It should be further enriched. In the penultimate paragraph of the section, you obviously want to refer to the gap that exists in the existing bibliography and the contribution of your research. Make this clearer.  In the last paragraph, you refer to research questions. However, these questions have not been clearly mentioned above. This point should be improved.

Response 2：

Thank you very much for this insightful comment. We agree that the original version of the Introduction did not include a sufficiently diverse set of bibliographical references, and that the research gap, contribution, and research questions were not articulated clearly enough.

In the revised manuscript, we have made the following changes in Section 1 “Introduction”: 
We have expanded the Introduction by adding more references from both China and other developing regions. In particular, we now discuss agricultural e-commerce adoption and digital market participation in Africa (e.g., mobile phones and farmers’ marketing decisions in Ethiopia) and Latin America (e.g., structural barriers and transaction costs faced by smallholders), so that our study is more clearly situated within the global literature rather than relying almost exclusively on China-focused work. These additions appear in the middle part of the Introduction (paragraphs 3–5 of Section 1).
We have rewritten the penultimate paragraph of the Introduction to explicitly highlight the gap in the existing literature and the contribution of our research. We now clearly state that, although prior studies have examined social learning and risk attitudes separately, there is still limited research that embeds both into a unified empirical framework to examine how external social learning reshapes internal risk attitudes and thereby affects e-commerce participation. We then explain that our study addresses this gap by constructing an integrated “social learning–risk attitude–e-commerce participation” framework and using micro-survey data from peach farmers in Qingdao to provide new empirical evidence and policy implications.
In the last part of the Introduction, we now explicitly list three core research questions and introduce them as a direct response to the identified research gap:
(i) whether social learning significantly promotes farmers’decision and extent of participation in agricultural e-commerce;
(ii) whether risk attitude plays a mediating role in this relationship; and
(iii) whether this influence mechanism exhibits heterogeneity after controlling for individual and contextual factors.
These questions are now clearly stated and explicitly linked to the preceding discussion of the literature gap and the proposed analytical framework. We then briefly summarize the structure of the remainder of the paper.

We hope that these revisions make the Introduction richer in bibliographical references, clearer in presenting the research gap and contribution, and more explicit and coherent in stating the research questions.

Comments 3：Although the proposed structure of the journal is followed. The bibliographic review does not particularly fit into the Materials and Methods section. You should see how you will deal with this. Also, enrichment with more bibliographic references is suggested. It seems poorly documented.

Response 3：Thank you very much for this helpful comment. We agree that, in the original version, placing the bibliographic review inside the Materials and Methods section was not ideal.

In the revised manuscript, we have reorganized the structure so that the literature review is now presented as an independent subsection, separate from the methods. Specifically, we have split the original“Materials and Methods”section into two parts:

Section 2“Literature Review and Hypotheses”, which now concentrates on the theoretical and bibliographical background and hypotheses.

Section 3 “Methodology”, which focuses solely on the data, variable definitions, and empirical model.

Comments 4：Figure 1 should be better explained, through text.

Response 4：Thank you for this helpful comment. We agree that in the original manuscript the mediating role of risk attitudes and the structure of Figure 1 were not sufficiently explained in the text. In the revised version, we have expanded the textual explanation of Figure 1 in Section 2.2.2 “The Mediating Effect of Risk Attitudes”, so that the reader can clearly understand each component of the conceptual model and how it corresponds to the hypotheses. Specifically, we now explain (1) the direct effect of social learning on farmers’e-commerce participation, (2) the indirect effect operating through risk attitudes (mediation path), and (3) the role of control variables in the model and how they relate to the interpretation of partial mediation.

These changes are included in Section 2.2.2, immediately before Figure 1 in the revised manuscript. 

Comments 5：In the Materials and Methods section, it is good to clearly define the aim and objects of the study. You can add exactly when the primary data collection was conducted.

Response 5：Thank you very much for this useful suggestion. We agree that the aim and objects of the study were not stated clearly enough in the Materials and Methods section, and that the timing of the primary data collection should be specified more precisely. In the revised manuscript, we have made the following changes in Section 1“Introduction”:In the last part of the Introduction, we now explicitly list three core research questions and introduce them as a direct response to the identified research gap. In Section 3 “Methodology”, we have supplemented the description of the survey by clearly stating when the primary data were collected.

Comments 6：In the discussion section, one would expect you to compare your results with those of other research. This needs to be improved. Also, a separate section with the conclusions should be created. Future research directions can be integrated into the conclusions section.

Response 6：

Thank you for this valuable comment. We agree that the original version of the manuscript did not sufficiently compare our results with existing literature, and that the discussion would benefit from a more critical and comparative approach. We also appreciate the suggestion to create a separate conclusions section and to integrate future research directions there.
In the revised manuscript, we have made the following changes:

In Section 5 “Discussion”, we have added several paragraphs that explicitly compare our findings with previous studies on social learning and technology adoption (e.g., Conley and Udry on Ghanaian pineapple farmers; Munshi on acquaintance-based networks), and with work on risk attitudes in farmers’ decision-making. We highlight both the consistencies and the ways in which our results extend this literature.

We have created a new Section 6 “Conclusions”, separate from the discussion. This section (i) summarizes the main findings and policy implications in a concise way, and (ii) integrates the directions for future research, including cross-regional comparisons and more detailed measures of digital learning channels.

Reviewer 3 Report

Comments and Suggestions for Authors

The article investigates the effects of social learning and risk attitudes on rural households’ participation in agricultural e-commerce activities. The study presents an analytical framework grounded in behavioral economics theory and utilizes survey data collected from 327 peach farmers in Qingdao (Shandong Province). The application of the Heckman two-stage model, IV-Probit, and mediation models provides a solid scientific foundation for the analysis.

However, several limitations are identified in the study.
Although the literature review and overall scope are comprehensive and up to date, approximately 95% of the reviewed works are China-focused, while the international context (for instance, agricultural e-commerce adoption in Africa or Latin America) has not been sufficiently discussed. The theoretical linkage between social learning and risk attitudes could also be examined in greater depth.

The study does not specify the numerical distribution of the 327 questionnaires conducted through online and face-to-face surveys. It remains unclear how many were completed online and by what method, as well as how the farmers were selected. More detailed clarification on these points is needed. The sampling method should be explicitly defined (e.g., “random,” “purposive,” etc.), and the sampling frame should be stated. The exact numbers and proportions of online and offline questionnaires must be reported, and the representativeness of the sample for the study region should also be addressed.

The social learning indicator was measured using five questionnaire items, yet no evidence of confirmatory factor analysis (CFA) or reliability testing (Cronbach’s α) was provided.
It is also unclear whether the risk attitude variable was measured solely by a single “monetary choice” question.

The choice of the entropy weighting method for scaling variables lacks sufficient justification; the authors should explain why this method was preferred over more conventional approaches such as principal component analysis (PCA) or factor analysis.
The interpretation of coefficients would be improved by discussing them in terms of marginal effects.

In several results tables, units and interpretations of coefficients are missing (for instance, the expression “13.27% increase” is ambiguous, as the measurement units of the variables are not clearly defined).
The discussion section should adopt a more critical and comparative approach to existing literature.

The heterogeneity analysis focuses solely on “cooperative membership”; it could be expanded to include other factors such as regional development level or internet infrastructure.
The proposal of a “Rural E-commerce Risk Education Program” is particularly valuable.

However, the study’s international and theoretical implications remain limited.

Thank you.

Author Response

Comments 1：Although the literature review and overall scope are comprehensive and up to date, approximately 95% of the reviewed works are China-focused, while the international context (for instance, agricultural e-commerce adoption in Africa or Latin America) has not been sufficiently discussed. The theoretical linkage between social learning and risk attitudes could also be examined in greater depth.

Response 1：Thank you very much for this helpful comment. We agree that the bibliographic review in the original manuscript was not sufficiently rich. In the revised version, we have substantially enriched the literature review section, especially by adding more international references related to rural e-commerce adoption, social learning, and farmers’ risk attitudes (including studies from regions such as Africa and Latin America). These additions help to better situate our study within the global research context and address the concern that the paper was“poorly documented.”

In addition, in Section 2.2.2 “The Mediating Effect of Risk Attitudes”, we have expanded the theoretical discussion on how social learning shapes farmers’perceptions of uncertainty and expected returns, thereby altering their risk attitudes. We more clearly explain the behavioral-economics logic behind the mediating mechanism tested in our model, thus strengthening the theoretical linkage between social learning and risk attitudes.

Comments 2：The study does not specify the numerical distribution of the 327 questionnaires conducted through online and face-to-face surveys. It remains unclear how many were completed online and by what method, as well as how the farmers were selected. More detailed clarification on these points is needed. The sampling method should be explicitly defined (e.g.,“random,”“purposive,”etc.), and the sampling frame should be stated. The exact numbers and proportions of online and offline questionnaires must be reported, and the representativeness of the sample for the study region should also be addressed.

Response 2：Thank you very much for this important and detailed comment. We agree that in the original version the description of the sampling process and the distribution of online vs. face-to-face questionnaires was not sufficiently clear.In the revised manuscript, we have substantially clarified the sampling frame, sampling method, data-collection modes, and sample representativeness in Section 3.3“Sample Selection and Data Sources”.

Comments 3：The social learning indicator was measured using five questionnaire items, yet no evidence of confirmatory factor analysis (CFA) or reliability testing (Cronbach’s α) was provided. It is also unclear whether the risk attitude variable was measured solely by a single“monetary choice”question.

Response 3：Thank you very much for this careful and important comment. We agree that the original version did not provide sufficient information on the reliability evaluation of the social learning construct, nor did it clearly describe how the risk attitude variable was measured.

In the revised manuscript, we have clarified and supplemented these points as follows: As the reviewer notes, the social learning indicator is constructed from five questionnaire items covering observational learning and reinforcement learning. In the revised version, we now explicitly report a reliability test for this scale. Specifically, in Section 4.1“Social Learning’s Impact on E-Commerce Adoption”, we add Table 3 “Reliability Analysis Results” and state that the Cronbach’s alpha for the five-item social learning dimension is 0.851, which exceeds the commonly accepted 0.70 threshold in social science research, indicating good internal consistency and satisfactory reliability of the construct.

Regarding risk attitude, we now clearly state in Section 3.2“Variable Definitions” that this variable is measured using a single monetary-choice question, following the approach developed by Chen et al. and widely used in experimental and survey-based studies. Specifically, respondents are asked which of five hypothetical gain–lottery scenarios they would be willing to choose, ranging from“certainly receiving 1,000 CNY”to“a 50% chance of receiving 0 CNY and a 50% chance of receiving 5,000 CNY,”which are coded from 1 (extremely risk-averse) to 5 (strongly risk-preferring).

Comments 4：The choice of the entropy weighting method for scaling variables lacks sufficient justification; the authors should explain why this method was preferred over more conventional approaches such as principal component analysis (PCA) or factor analysis.

Response 4：Thank you very much for this insightful comment. We agree that in the original version the rationale for choosing the entropy weighting method to construct the social learning index was not explained clearly enough, especially in comparison with more conventional approaches such as principal component analysis (PCA) or factor analysis.

We have already added a brief explanation for why we apply the entropy method to weight the five observed indicators of social learning instead of conducting a confirmatory factor analysis. As described in Section 3.2, the five indicators are heterogeneous in their information sources and variances (interpersonal vs. media-based channels), and the entropy method allows us to assign data-driven weights to indicators that carry more information. Moreover, given that our data consist of 327 observations from a single region, the entropy method imposes fewer assumptions on latent factor structure than CFA or PCA and is therefore more appropriate for index construction in this context. 

Comments 5：The interpretation of coefficients would be improved by discussing them in terms of marginal effects.

Response 5：

Thank you very much for this useful suggestion. We agree that interpreting the coefficients in terms of marginal effects can make the results more intuitive.
In the revised manuscript, we have taken two steps to address this point:

First, in Section 4.1 “Social Learning’s Impact on E-Commerce Adoption”, we now report and discuss the average marginal effects of social learning from the Probit and IV-Probit models. For example, we explicitly state that a one-standard-deviation increase in the social learning index is associated with an increase of about 0.10–0.12in the probability of participating in agricultural e-commerce, evaluated at the sample means.

Second, in Table 4, we have added a footnote clarifying that the reported effects for the Probit-type models are marginal effects, and we indicate the units of the dependent variables (“probability” and “participation ratio”) to avoid ambiguity.
These revisions make the interpretation of the coefficients more transparent and consistent with the reviewer’s recommendation to emphasize marginal effects.

Comments 6：In several results tables, units and interpretations of coefficients are missing (for instance, the expression“13.27%”increase is ambiguous, as the measurement units of the variables are not clearly defined).

Response 6：Thank you very much for pointing out this important issue. We agree that in the original manuscript some tables did not clearly specify the units and measurement definitions of key variables, which made expressions such as“13.27% increase”insufficiently precise.

Comments 7：The heterogeneity analysis focuses solely on“cooperative membership”; it could be expanded to include other factors such as regional development level or internet infrastructure.

Response 7：Thank you very much for this helpful comment. We agree that, in the original version, the heterogeneity analysis was too narrowly focused on cooperative membership and did not sufficiently consider differences in regional development and infrastructure conditions. In the revised manuscript, we have expanded the heterogeneity analysis in Section 4.3 “Heterogeneity Test”. In addition to the grouping by cooperative/association participation, we now conduct a geographical-location-based heterogeneity analysis using three main producing areas in Qingdao (Laixi, Pingdu, Chengyang). The new regressions (reported in Table 8) show that social learning significantly promotes farmers’ participation in agricultural e-commerce in all three regions, but the estimated coefficients differ in magnitude (Laixi > Pingdu > Chengyang), revealing clear spatial heterogeneity. 

Reviewer 4 Report

Comments and Suggestions for Authors

The paper looks into an interesting topic. I have the following comments for the authors to consider:

1. The introduction is a bit long without focus. Please try to better explain the story behind as well as the paper's contribution to the existing literature. Moreover, the description of the structure of the paper is missing.
2. When exactly did you conduct your survey and how did you find the respondees?
3. On what basis did you identify the variables used in your model?
4. To what extent do you find your results generalisable?
5. I would urge the authors to use a bit more non-technical language when evaluating the results.
6. Discussion section needs more elaboration in terms of usable policy recommendations.

Author Response

Comments 1：The introduction is a bit long without focus. Please try to better explain the story behind as well as the paper’s contribution to the existing literature. Moreover, the description of the structure of the paper is missing.

Response 1：Thank you for pointing this out. We agree with this comment. Therefore, we have revised the introduction section to more clearly articulate the research background and narrative, explicitly emphasize the paper's contribution to the existing literature, and add a description of the paper's structure. These changes can be found in the revised manuscript in Section 1.

Comments 2：When exactly did you conduct your survey and how did you find the respondees?

Response 2：Thank you for raising this question. We agree that specifying the exact timing and methods of data collection is crucial for ensuring the transparency and replicability of our study. Therefore, we have clarified the survey period and the detailed procedures for identifying and reaching respondents in the manuscript.

To improve data quality, a small-scale online pilot survey was first conducted; feedback from this pilot was used to adjust item wording, option design, and the order of questions. The target respondents were peach farmers, and to expand coverage we adopted a mixed offline + online data-collection strategy.
For the offline component, a sampling frame was first constructed from farmer lists provided by township agricultural departments and village committees. Enumerators then conducted face-to-face interviews in orchards and at farmers' homes. A total of 210 paper questionnaires were distributed and 192 valid responses were obtained, accounting for 58.7% of all valid cases.
For the online component, questionnaires were disseminated through WeChat mini-programs, Wenjuanxing, and WeChat official accounts, and further circulated in peach-farmer WeChat groups, fresh-peach trading groups, and fan groups of leading bloggers in the sector. This yielded 135 valid online questionnaires, representing 41.3% of the total valid sample. These changes can be found in the revised manuscript in Section3.3.

Comments 3：On what basis did you identify the variables used in your model?

Response 3：Thank you for raising this question. We agree that clarifying the theoretical and empirical basis for variable selection is crucial for ensuring the rigor of the research model. Therefore, we have added explanations regarding the rationale for selecting the core variables (dependent variable, core explanatory variable, mediating variable, and control variables) in the manuscript. These changes can be found in the revised manuscript in Section3.2“Variable Definitions” ,with corresponding support in the relevant hypothesis development and literature review sections

Comments 4：To what extent do you find your results generalisable?

Response 4：Thank you for raising this important question. We fully agree that clarifying the scope and boundaries of the research findings is crucial for interpreting their value. Therefore, we have included a dedicated and candid discussion on this very point in the “Limitations and Future Directions” section (Section 6.2) of the paper.

In summary, we argue that: The direct generalizability of the empirical results is conditional, primarily applicable to contexts similar to that of Qingdao peach farmers (i.e., producing high-value, perishable products with relatively good infrastructure). The theoretical mechanism has greater potential for generalizability. The core contribution lies in revealing the mediating pathway of “social learning influencing participation through risk attitude.” This behavioral mechanism itself holds significant theoretical value and provides a new analytical framework for understanding farmers' technology or channel adoption behaviors in other contexts. The methodology offers reference value. The social learning measurement we constructed and the econometric models we employed provide a replicable scheme for addressing similar problems.

Comments 5：I would urge the authors to use a bit more non-technical language when evaluating the results.

Response 5：Thank you for this valuable suggestion. We completely agree that using more accessible language to present the research findings enhances the paper's readability and allows a broader audience to grasp the value of the study. Therefore, we have revised the results evaluation sections throughout the paper (primarily in Section 5 “Discussion” and Section 6 “Conclusion”) to reduce the use of technical jargon and employ more intuitive language to explain the findings.

Comments 6：Discussion section needs more elaboration in terms of usable policy recommendations.

Response 6：Thank you for pointing this out. We agree with this comment. Therefore, we have substantially elaborated on the concrete and actionable policy recommendations derived from our research findings in both the “Discussion”(Section 5) and the “Conclusion” (Section 6) sections of the paper.

 

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have completely revised their manuscript, in accordance with my comments. I am very grateful for the responses they have given to the comments. In my opinion, the article is fine in its current form.

Reviewer 4 Report

Comments and Suggestions for Authors

The authors have responded to all my questions so I am happy to advise this paper to be published.