Factors Affecting New Agricultural Business Entities’ Adoption of Sustainable Intensification Practices in China: Evidence from the Main Apple-Producing Areas in the Loess Plateau

Round 1

Reviewer 1 Report

The authors present results from their survey investigating the various aspects of SIACS adoption, technology, and organization on the adoption intention and decision of SIPs.  The background and literature review demonstrate a well-developed and thoughtful research plan and analysis. 

Major concerns:

• There are 16 hypotheses of interest, but no adjustment for multiple comparisons is mentioned. With 16 comparisons, there is almost a 50% chance of a significant result at an alpha=0.05 purely by chance.  Some type of multiple comparison adjustment needs to be considered for the results, particularly because the exact p-values are not provided in Table 6.
• The organization of the paper needs improvement. The results appear in the Methods section and the discussion of the results is listed under Results. 

Minor concerns:

• Line #38 – the numbers of degraded land do not appear to agree. (1.45 Mha/1964 Mha is not equal to 7.4%).
• Section 4.2 – the description of the number of counties surveyed seems to contradict. Line 52 states “5 apple-producing counties” and line 58 states “six survey counties”.
• The Fornell and Larcker paper referenced is not the original paper, but a response. Double-check if the reference should point to the original paper.

Author Response

Dear Editor and Reviewer,

Thank you for giving us the opportunity to revise and resubmit this paper, and to the Reviewer for their thoughtful and helpful review. We respond to each of the Reviewer’s comments individually below (see **s), and highlight in red font both here and in the revised manuscript where substantive revisions have been made.

Yours sincerely,

The Authors

Reviewer #1: Referee Report on " Factors Affecting New Agricultural Business Entities’ Adoption of Sustainable Intensification Practices in China: Evidence from the Main Apple Producing Areas in the Loess Plateau "

Comments and Suggestions for Authors

The authors present results from their survey investigating the various aspects of SIACS adoption, technology, and organization on the adoption intention and decision of SIPs.  The background and literature review demonstrate a well-developed and thoughtful research plan and analysis.

Major concerns:

1. There are 16 hypotheses of interest, but no adjustment for multiple comparisons is mentioned. With 16 comparisons, there is almost a 50% chance of a significant result at an alpha=0.05 purely by chance. Some type of multiple comparison adjustment needs to be considered for the results, particularly because the exact p-values are not provided in Table 6.

** We thank the reviewer for these references and have added an explanation of the reasonableness of the choice of the validation model (structural equation model) on page 13, as well as an explanation that the sample size for the current study meets the sample size requirements for the structural equation model. And we have chosen appropriate model estimation methods where the sample size of the current study is adequate but relatively small. The specific content is as follows: 

We used a structural equation model (SEM) with observed and latent variables to test the conceptual model and assess the strength of the research hypotheses, namely the effects of technical, organizational and environmental factors on SIP adoption in-tentions and decisions. As the data analysis involves observed variables, endogenous latent variables (adoption intentions and adoption decisions), and exogenous latent variables (influencing factors), the variance explained by the model is higher than when other methods (e.g., regression analysis), are used. Schumacker and Lomax (2004) found that most SEM studies had a sample size between 200 and 500. Further, studies have been conducted to evaluate the effect of sample size on the results of SEMs (Hu and Bentler, 1995; Shevlin and Miles,1998). These investigations have shown that a minimum of 100 cases should be used in the latent variable analysis. Fewer than 100 observations results in the estimates of the population parameters becoming unreliable. The sample size of this study is 206, which meets the requirement of the SEM method. Partial least squares (PLS) was used to estimate the model. Partial least squares (PLS) was used to estimate the model. PLS is more suitable for small samples (Fornell and Bookstein, 1982), as in the case of the current research; though the sample size of 206 is adequate, it is nevertheless relatively small.

** Having explained the reasonableness of the methodology and sample size of this study, we also provide a description of the parameters that demonstrate a good fit of the measurement model, as follows：

Confirmatory factor analysis (CFA) was applied to assess the reliability and va-lidity of the measurement model. For a measurement model to have sufficiently good model fit, the x2 value normalized by degrees of freedom (x²/df) should not exceed 3, and non-normed fit index (NNFI) and comparative fit index (CFI) should exceed 0.9 (Bagoz-zi and Yi, 1988). For the current CFA model, x²/df) was 1.629 (x²=188.924; df=116), NNFI was 0.908, CFI was 0.961, suggesting adequate model fit.

** Following the reference of the reviewer, we have added the exact values of the Standard Error, Critical Ratio and p-values in Table 6 below. In addition, regarding the multiple comparison adjustment for model results, we resampled the original sample (206 samples) using bootstrapping, according to Amaro and Duarte (2015) and Ahuja et al. (2016). The new data from the bootstrapping re-sampling (500 samples) was used to test the original hypothesis, and the results are shown in Table a. Based on these, we compared the p-values in Table a with those in Table 6 and found that the significance levels of the results in Table 6 were generally consistent with those in Table a, thus demonstrating the robustness of the results of the structural equation model in this study. In this study, the validity of the SEM model was tested using a two-step procedure. The test results of the first stage demonstrated the good model fit, convergent validity and discriminant validity of the measurement model. The test results of the second stage on the structural model showed that technical, organisational and environmental context account for 36.4% of the variance in adoption intention. Moreover, together with adoption intention explained 87.1% of the variance in adoption decision. These two figures illustrate that the model fits the data well. This suggests that the model explains a large proportion of the variation in the endogenous variables. In addition, the model fit indices were within acceptable thresholds: the ratio of x² to degrees of freedom was 1.868 (x² = 229.805; df = 123), CFI = 0.897, GFI = 0.915, AGFI = 0.861 and RMSEA = 0.065, demonstrating the robustness of the model results. Similar studies such as Yadegaridehkordi et al. (2016) and Liébana-Cabanillas et al. (2018), which validated theoretical models with structural equation modelling, also used a two-step procedure to test the robustness of the model results. Therefore, we have not presented the bootstrapping re-sampling process in the original text.

Table 6. Results of estimation structural model (206 samples).

Table a. Results of estimation structural model of the bootstrapping re-sampling data (500 samples).

** We followed the Reviewer’s suggestion and and added "5. Data analysis and results " and moved "4.3. Reliability and validity " and "4.4. Hypothesis Test of the Structural Equation Model " from the original article into this section, and revised them to "5.1. The reliability and validity of the measurement model " and "5.2. Hypothesis Test of the Structural Equation Model ". In addition, we have changed "5. Results" to "6. Discussion" in the original text.

Minor concerns:

3. Line #38 – the numbers of degraded land do not appear to agree. (1.45 Mha/1964 Mha is not equal to 7.4%).

**We thank the Reviewer for this suggestion. The true figure is 145 Mha and we have made the correction.

4. Section 4.2 – the description of the number of counties surveyed seems to contradict. Line 52 states “5 apple-producing counties” and line 58 states “six survey counties”.

** In fact, we did fieldwork in 6 counties, but our negligence led to such a low-level error in the text. We are very grateful to reviewer for pointing out this error. We have changed line 52 to "6 apple-producing counties".

5. The Fornell and Larcker paper referenced is not the original paper, but a response. Double-check if the reference should point to the original paper.

** We followed the Reviewer’s suggestion and double-checked this reference and made sure it pointed to the original text. We have provided an accurate reference.

References

1. Schumacker, R. E., & Lomax, R. G. (2004). A beginner’s guide to structural equation modeling. New York, NY: Psychology Press.
2. Hu, L., & Bentler, P. M. (1995). Evaluating model fit. In R. H. Hoyle (Ed.), Structural equation modeling: Concepts, issues, and applications (pp. 76-99). Thousand Oaks, CA: Sage.
3. Shevlin, M., & Miles, J. N. V. (1998). Effects of sample size, model specification and factor loadings on the GFI in confirmatory factor analysis. Personality and Individual Differences, 25(1), 85–90. doi:10.1016/s0191-8869(98)00055-5
4. Fornell, C., & Bookstein, F. L. (1982). Two Structural Equation Models: LISREL and PLS Applied to Consumer Exit-Voice Theory. Journal of Marketing Research, 19(4), 440. doi:10.2307/3151718
5. Bagozzi, R. P., & Yi, Y. (1988). On the evaluation of structural equation models. Journal of the Academy of Marketing Science, 16(1), 74–94. doi:10.1007/bf02723327
6. Ahuja, R., Jain, M., Sawhney, A., & Arif, M. (2016). Adoption of BIM by architectural firms in India: technology–organization–environment perspective. Architectural Engineering and Design Management, 12(4), 311–330. doi:10.1080/17452007.2016.1186589
7. Amaro, S., & Duarte, P. (2015). An integrative model of consumers' intentions to purchase travel online. Tourism Management, 46, 64e79.
8. Yadegaridehkordi, E., Nilashi, M., Nasir, M. H. N. B. M., & Ibrahim, O. (2018). Predicting determinants of hotel success and development using Structural Equation Modelling (SEM)-ANFIS method. Tourism Management, 66, 364–386. doi:10.1016/j.tourman.2017.11.012
9. Liébana-Cabanillas, F., Marinkovic, V., Ramos de Luna, I., & Kalinic, Z. (2018). Predicting the determinants of mobile payment acceptance: A hybrid SEM-neural network approach. Technological Forecasting and Social Change, 129, 117–130. doi:10.1016/j.techfore.2017.12.015

Author Response File: Author Response.docx

Reviewer 2 Report

Dear Authors Thank you very much for your efforts. I hope that the conveyor will be used by other researchers and interested parties, considering the reviews that are fully mentioned in the text. The volume of the content of the article, as it is high, its quality should also be increased, which I am sure will be done with your efforts. 

Author Response

Dear Editor and Reviewer,

Thank you for giving us the opportunity to revise and resubmit this paper, and to the Reviewer for their thoughtful and helpful review. We respond to each of the Reviewer’s comments individually below (see **s), and highlight in red font both here and in the revised manuscript where substantive revisions have been made.

Yours sincerely,

The Authors

Reviewer #2: Referee Report on " Factors Affecting New Agricultural Business Entities’ Adoption of Sustainable Intensification Practices in China: Evidence from the Main Apple Producing Areas in the Loess Plateau "

Comments and Suggestions for Authors

Dear Authors Thank you very much for your efforts. I hope that the conveyor will be used by other researchers and interested parties, considering the reviews that are fully mentioned in the text. The volume of the content of the article, as it is high, its quality should also be increased, which I am sure will be done with your efforts.

1. The volume of the content of the article, as it is high, its quality should also be increased, which I am sure will be done with your efforts.

** We thank the reviewer for these references. To enhance the structural integrity of the article, we have added an explanation of the reasonableness of the choice of the validation model (structural equation model) on page 13, as well as an explanation that the sample size for the current study meets the sample size requirements for the structural equation model. And we have chosen appropriate model estimation methods where the sample size of the current study is adequate but relatively small. The specific content is as follows: 

We used a structural equation model (SEM) with observed and latent variables to test the conceptual model and assess the strength of the research hypotheses, namely the effects of technical, organizational and environmental factors on SIP adoption in-tentions and decisions. As the data analysis involves observed variables, endogenous latent variables (adoption intentions and adoption decisions), and exogenous latent variables (influencing factors), the variance explained by the model is higher than when other methods (e.g., regression analysis), are used. Schumacker and Lomax (2004) found that most SEM studies had a sample size between 200 and 500. Further, studies have been conducted to evaluate the effect of sample size on the results of SEMs (Hu and Bentler, 1995; Shevlin and Miles,1998). These investigations have shown that a minimum of 100 cases should be used in the latent variable analysis. Fewer than 100 observations results in the estimates of the population parameters becoming unreliable. The sample size of this study is 206, which meets the requirement of the SEM method. Partial least squares (PLS) was used to estimate the model. Partial least squares (PLS) was used to estimate the model. PLS is more suitable for small samples (Fornell and Bookstein, 1982), as in the case of the current research; though the sample size of 206 is adequate, it is nevertheless relatively small.

** Having explained the reasonableness of the methodology and sample size of this study, we also provide a description of the parameters that demonstrate a good fit of the measurement model, as follows：

Confirmatory factor analysis (CFA) was applied to assess the reliability and va-lidity of the measurement model. For a measurement model to have sufficiently good model fit, the x2 value normalized by degrees of freedom (x²/df) should not exceed 3, and non-normed fit index (NNFI) and comparative fit index (CFI) should exceed 0.9 (Bagoz-zi and Yi, 1988). For the current CFA model, x²/df was 1.629 (x²=188.924; df=116), NNFI was 0.908, CFI was 0.961, suggesting adequate model fit.

** To enhance the quality of the article, we have added an analysis of the reasons why relative advantage and perceived barriers influence adoption intentions and adoption decisions on pages 16 and 17. The specific content is as follows: 

6.1. Technological context

The results show that relative advantage makes a significant contribution to the adoption intention (standardized coefficient of 0.285). This is similar to the findings of Zeweld et al. [21] that respondents with positive attitudes are more willing to adopt SIPs. Relative advantage shows a significant facilitation effect on the adoption decision (standardized coefficient of 0.409). This finding is consistent with the argument below that perceived benefits are closely related to the acceptance and use of technology by potential users [88]. The large scale of NABEs requires the employment of labour, and the loss of labour and the ageing population in China's rural areas has led to difficulties in finding labour and rising labour costs in their daily operations. The SIACS can replace manual labour with machinery, which can greatly reduce labour costs and help NABEs to solve this dilemma. In addition, SIACS has significantly increased yields while improving product quality compared to traditional models.  These may be important reasons why the perceived advantages significantly influence adoption intentions and adoption decisions.

Perceived barriers to SIACS adoption intention and adoption decisions show a significant inhibitory effect: when adopters perceive SIACS as complex and risky, their intention to adopt SIACS and the probability of making a decision will be significantly reduced by 0.344 and 0.382, respectively. This conclusion is supported by several studies [88,89], which suggest that the relative complexity and relative risk of technologies have a significant impact on adoption, and that potential adopters hesitate to adopt complex and risky technologies. SIACS is a completely new model and the experience of the traditional model is not at all applicable to it, which invariably makes it more difficult for producers to master this new set of technologies. In addition, SIACS has high input costs, and both the complexity and high inputs increase the risk of SIACS. There is also the fact that agricultural production is often exposed to high natural disaster risk and market risk, and China's current agricultural insurance system is not very well developed, which results in NABEs being less resilient to risk. The combination of these factors may be the reason why perceived barriers significantly inhibit adoption intentions and adoption decisions.

2. Extensive editing of English language and style required

** We followed the Reviewer’s suggestion and revised the language. We also had the revised text retouched by a professional polishing agency. 

References

1. Schumacker, R. E., & Lomax, R. G. (2004). A beginner’s guide to structural equation modeling. New York, NY: Psychology Press.
2. Hu, L., & Bentler, P. M. (1995). Evaluating model fit. In R. H. Hoyle (Ed.), Structural equation modeling: Concepts, issues, and applications (pp. 76-99). Thousand Oaks, CA: Sage.
3. Shevlin, M., & Miles, J. N. V. (1998). Effects of sample size, model specification and factor loadings on the GFI in confirmatory factor analysis. Personality and Individual Differences, 25(1), 85–90. doi:10.1016/s0191-8869(98)00055-5
4. Fornell, C., & Bookstein, F. L. (1982). Two Structural Equation Models: LISREL and PLS Applied to Consumer Exit-Voice Theory. Journal of Marketing Research, 19(4), 440. doi:10.2307/3151718
5. Bagozzi, R. P., & Yi, Y. (1988). On the evaluation of structural equation models. Journal of the Academy of Marketing Science, 16(1), 74–94. doi:10.1007/bf02723327

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

Hello, many thanks for your efforts.

regards.