Next Article in Journal
Landslide Risk Assessment Along Railway Lines Using Multi-Source Data: A GameTheory-Based Integrated Weighting Approach for Sustainable Infrastructure Planning
Previous Article in Journal
Disaster Chain Evolution and Risk Mitigation in Non-Coal Underground Mines with Fault Zones: A Complex Network Approach
Previous Article in Special Issue
Research on the Evolutionary Game of Quality Governance of Geographical Indication Agricultural Products in China: From the Perspective of Industry Self-Governance
 
 
Article
Peer-Review Record

Socio-Economic Determinants of Climate-Smart Agriculture Adoption: A Novel Perspective from Agritourism Farmers in Nigeria

Sustainability 2025, 17(12), 5521; https://doi.org/10.3390/su17125521
by Ifeanyi Moses Kanu * and Lucyna Przezbórska-Skobiej
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Sustainability 2025, 17(12), 5521; https://doi.org/10.3390/su17125521
Submission received: 25 April 2025 / Revised: 15 May 2025 / Accepted: 3 June 2025 / Published: 16 June 2025

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Dear Authors,

Overall, the work makes a positive impression. However, a shortcoming has been identified that is recommended to be addressed.

1) The models contain a large number of statistically insignificant coefficients, which may be due, among other factors, to multicollinearity among the independent variables. To improve the statistical significance of coefficients in the Multivariate Probit (MVP) model, it is recommended to conduct a multicollinearity analysis of the quantitative variables (AgExp, Edu, HHZ, FMZ, AgY), to standardize them for numerical stability, and to reassess the significance and role of the binary predictors (Cred, InfoC, Coop, Teno, Ext), especially in cases of imbalance between 1 and 0. According to Table 2, Socio-economic and CSA Characteristics of Agritourism Farmers in Nigeria, imbalance is observed for the following dependent variables:

  • Organic Farming (Yes = 1, No = 0): 0.80
  • Crop Rotation and Intercropping (Yes = 1, No = 0): 0.82
  • Agroforestry System (Yes = 1, No = 0): 0.78

These three variables exhibit modeled class imbalance, which may explain the weak statistical significance of some coefficients. This issue is not critical, but it should be re-evaluated after performing a multicollinearity check (e.g., using VIF) for AgExp, Edu, HHZ, FMZ, and AgY.

2) Optional (not mandatory): You may also consider conducting a Principal Component Analysis (PCA) on the independent variables AgExp, Edu, HHZ, FMZ, and AgY, and then use the resulting components as independent variables in the model equations.

 

Beat regards,

Reviewer

Author Response

 

We sincerely appreciate your constructive feedback on our manuscript, "Socio-economic Determinants of Climate-smart Agriculture Adoption: A Novel Perspective from Agritourism Farmers in Nigeria." Your insights have greatly contributed to the improvement of our study. Below, we address each of your concerns systematically.
  1. Standardization: Based on your recommendation, we standardized the quantitative independent variables (AgExp, Edu, HHZ, FMZ, AgY) by converting them to z-scores for numerical stability and re-estimated the Multivariate Probit (MVP) model. The results of this new MVP model with standardized X1, X2, X3, X4 and X6 variables shows improvement in the model, with additional 3 significant variables. So, we adopted the new MVP model. The diagnostic result are attached in MS-Excel.
  2. Multicollinearity Assessment: Following your recommendation, we conducted a Variance Inflation Factor (VIF) analysis to examine the presence of multicollinearity among the quantitative independent variables (AgExp, Edu, HHZ, FMZ, AgY). Our updated VIF results are as follows:   AgExp (X1) = 1.60 Edu (X2) = 1.05 HHZ (X3) = 1.02 FMZ (X4) = 4.14 AgY (X6) = 2.98   None of the VIF values exceed the commonly accepted threshold of 5, which suggests that multicollinearity is not a critical issue. Although X4 (Farmland Size) has a slightly higher VIF (4.14), it is still below problematic levels and captures essential economic dimensions. Given these results, we retain our previous model specification while acknowledging the need for careful interpretation of coefficients.
  3. Class Imbalance and Model Robustness: We recognize that the dependent variables Organic Farming (80%), Crop Rotation/Intercropping (82%), and Agroforestry System (78%) exhibit a degree of class imbalance. However, this imbalance reflects the actual adoption patterns observed in our study area rather than a methodological flaw. Removing or transforming these variables would not align with the novel empirical reality of agritourism farmers' climate-smart agriculture practices [just like the opening sentence of the article: "The existing body of scholarly work on the adoption of CSA in Africa and Nigeria has predominantly concentrated on the experiences and practices of smallholder farmers....."]. So, we maintain our current approach while ensuring a nuanced interpretation of our results.
  4. Normality Check (Shapiro-Wilk Test Results): We conducted a Shapiro-Wilk normality test for all continuous independent and dependent variables. The test yielded very small p-values, confirming deviations from normality. However, because our study employs a Multivariate Probit (MVP) model, normality of independent variables is not a prerequisite for valid estimation. The MVP model does not assume normal distribution for predictors, and thus, these deviations do not negatively impact model validity.
  5. Heteroskedasticity Check (Breusch-Pagan Test Results): We performed a studentized Breusch-Pagan test on an auxiliary linear regression model to assess heteroskedasticity. The test results are as follows:   BP Statistic = 1.4786 Degrees of Freedom (df) = 5 p-value = 0.9155   Since the p-value is very high (> threshold 0.05), we fail to reject the null hypothesis of homoskedasticity. This means that variance stability is not a concern and confirms that the MVP model does not suffer from heteroskedastic distortions.  
  6. Correlation Analysis: A correlation matrix was computed to examine potential relationships between independent and dependent variables. X4 (Farmland size) and X6 (AgY - Agritourism annual income) have a correlation of 0.80. This strong positive correlation is the primary reason for the VIF of X4 being 4.14 and X6 being 2.98. It makes sense that larger agritourism farms might generate more income. More specifically, the two variables are theoretically important for understanding the socio-economic determinants of CSA adoption in the context of agritourism. Removing them would lead to model misspecification and loss of valuable insights, as both farm scale and economic returns are crucial aspects of farmers' decision-making.  
All in all, we sincerely appreciate the reviewers' insightful recommendations.

Reviewer 2 Report

Comments and Suggestions for Authors

Dear authors,

Your research is very interesting and the topic is extremely important in our days. Generally the manuscript is good and well structured. It helps the reader to understand the topic of the research. Although, at some points it can be shortened so it easier to follow. I have made some suggestions below that I think your manuscript can benefit from.

In the abstract section, please state what MVP stands for.

In the Study area section of the Materials and Methods, I would like to see the authors concetrate more to the Agricultural part of the zones, and the socio-economic parameters. It is nice to see the cultural festivals etc, but it does not provide anything to the manuscript. On the other hand I could say that it distracts the reader from the point of the research.

In lines 317-318, you mention that you selected randomly 2 agritourism farmers from each LGA. Does this provide representativeness beyond geography? Please justify this in this section.

I would like to see the questionnaire as supplementary material.

In lines 487-488 you mention “Acces to credit and other financial resources”. Can you explain what this means? Does this refer to subsidies? It should be clear.

Tables are good, but the manuscript would benefit from 1 or 2 graphs of your results. It would be easier for the reader.

In the Discussion, I would like to suggest to avoid referring to your statistical model and numbers (for example lines 549-551, 584-585, 624-627 etc). Go straight to your point and compare your findings with previous research (as you have done correctly).

In Conclusions section, as I mentioned in the previous one, avoid referring to numbers (lines 711, 713, 714 etc).

In Conclusions or Policy recommendations, I would like to see a suggestion by the authors on how their results can benefit not only Nigeria region, but also other regions with similar status. It would be very good for your research to have this type of replicability.

Finally, I would suggest to somehow summarize the limitation of the study and incorporate the to the beginning of the discussion.

The use of english language in this manuscript is appropriate and does not need nothing more than a revision by the authors.

For example

Long sentences, such as in lines 33-36, should be avoided.

Some sentences, like in lines 54-56, can be rephrased more fluently, so it would be easier for the reader to understand.

Author Response

Dear Reviewer ,

We extend our sincerest gratitude for your insightful comments and constructive feedback on our manuscript. Your thorough review and thoughtful suggestions have been invaluable in helping us to refine and enhance the quality of our work. We truly appreciate the time and effort you dedicated to this process.

 

We are pleased to inform you that we have carefully considered all your recommendations and have effected the necessary corrections and revisions throughout the manuscript. Please find below a detailed response to the points raised and the changes implemented:

 

  • Paragraph structure and readability: We acknowledge the concern regarding lengthy paragraphs. We have diligently worked through the manuscript to shorten overly long paragraphs, thereby improving the overall flow and readability.

 

  • Retention and refinement of cultural information in study area section: We have decided to retain the information about cultural festival/new yam festival as it is part of the unique agritourism activities in the study area. Most readers don't really know about agritourism in Nigeria, it will be nice if they get a glimpse of agritourism activities in the study area on their journey to the results and discussion section. We improved on the section by adding a seamless connecting sentence.

 

  • Justification of Random Selection of Agritourism Farmers: We have incorporated a clear justification for the random selection of two agritourism farmers from each LGA. We have also attached the full research questionnaire used for data collection as supplementary material.

 

  • Clarification of Access to Credit: Access to credit in the context of the article refers to the availability of formal credit facilities (i.e. structured financial lending programs) intended for agritourism-related activities. It does not include subsidies, grants, or general financial aid programs. We have removed the phrase: "...and other financial resources”.

 

  • Inclusion of graphs: Based on your suggestion, we've included additional 2 graphs to the article: {Fig 4. Heatmap of bivariate correlations between adoption of CSA practices and exogenous factors among the randomly sampled agritourism farmers in Nigeria} and {Fig 5. Boxplot showing the influence of land ownership on agritourism income in Nigeria}.

 

  • Diagnostics test section: As recommended by Reviewer 2, we performed series of diagnostics statistical test, which are required before running the MVP model. We created new section in the article to report these test statistics - Section 2.5: Diagnostics Test (with sub-section 2.5.1. Check for Collinearity, 2.5.2. Test for Heteroskedasticity, and 2.5.3. Standardization). Following variable standardization, we re-performed the MVP analysis based on standardized std_X1, std_X2, std_X3, std_X4 and std_X6 variables. The revised results are presented in Table 3.

 

  • Policy Implications for Other Regions: We expanded the policy recommendation section to illustrate how our findings can benefit other regions that share similar characteristics.

 

  • Opinion on research limitation:We acknowledge Reviewer 1 suggestion to integrate the study limitations within the discussion section. However, we have retained the limitations in its original section to prevent excessive critique of the results. Integrating limitations within the discussion might inadvertently diminish the impact of the findings, making them seem less meaningful. If limitations were embedded in the discussion section, it could weaken the perceived significance of the findings, leading readers to fixate on the research flaws. A separate limitation section ensures that the methodological constraints are acknowledged without diminishing the importance of the core research findings, as practiced in numerous journals. Moreso, the limitation of the research are not overly unique/peculiar, but are the general limitations inherent in any cross sectional research involving the use of primary data.

 

  • Lastly, we revised and rephrased the manuscript as you recommended to improve the overall fluency and readability. And I have attached the questionnaire as supplementary material.

Author Response File: Author Response.pdf

Back to TopTop