New Approaches to Monitor Inequalities in Access to Water and Sanitation: The SDGs in Latin America and the Caribbean

Round 1

Reviewer 1 Report

This manuscript addresses the very important problem with reporting aggregate data regarding access to water and sanitation services in the context of meeting the UN SDGs in Latin America and the Caribbean. The authors rightly point out that aggregate data conceal uneven WaSH development. The authors also make the important point that SDG-related reporting on inequalities does not examine the intersection of different kinds of inequality, which has the potential to compound disadvantaged communities. This leads to underestimation of the problem.

 

To start to move the conversation in the right direction, the authors use the IPUMS-International Project data and develop a novel approach to examine intersectional differences with regard to WaSH insecurity, which is then compared to an inequality index at the country level for cross-evaluation.

 

This is an important paper tackling a very critical subject. I recommend it for publication in the journal. At the same time, I have a few questions that should be addressed in the manuscript to improve clarity:

 

1. The key term, AGPi, is calculated through logistic regression (page 4). What are the values? Are they odds ratios? B values (intercepts/coefficient for the constant)? If the latter, then how does the AGPi term incorporate the B’s standard error?

 

2. The response variables (page 4) for water (access to a piped system), do not really mean the households have water, just that they have access to a water distribution system. In many places in Latin America, households have access to a piped water system, but no water or very poor water quality (so no potable or drinkable water). This is a significant problem using this variable. The authors mention this at the end of the discussion section on page 16, but this very important limitation should be noted earlier in the methods section. In fact, there are several limitations to the data used, and so the authors should consider adding a sub-section on data limitations in section 2.

 

3. The explanatory variables were derived from “bibliographic research” (pages 4-5). Can you please be more specific?

 

4. The ‘race or color’ variable “white/non-white” (page 5) seems very problematic. Overall, the use of “white” and “black” is confusing in this article, such as on page 12. Race is a cultural construction and so what “white” means in one cultural context is not the same across all contexts. The authors will need to justify and defend the use of this variable, which weakens the utility of the analysis.

 

5. There seems to be an overlap between the ‘literate/non-literate’ variable and the ‘educational attainment’ variable, since one cannot attain education if they are illiterate (page 5). For the model, this will result in double-counting the weight of this variable. The authors should remove one of these or justify and defend the use of both, noting the problems with the overlap.

 

6. How does the Inequality Factor equal a percentage, as displayed in Figures 1 and 2. A percentage of what? Or do the authors mean for this to be displayed as a proportion or a scaled value (0-1)?

 

7. On page 14, the authors note: “There was a correlation of approximately 60% between the two methods,” citing an R-squared of 0.6019. The correct interpretation of this term is that the relationship (linear model) explains 60% of the variation in the dependent variable (“intersectionality”—which is misspelled on the Y axis in Figures 11 and 12).

Author Response

Dear Reviewer,

 

We appreciate the time and effort that you have dedicated to providing your valuable feedback on our manuscript. We revised the text and believe that the result is a clearer and improved paper. Here is a point-by-point response to your comments and concerns:

 

 

1. The key term, AGPi, is calculated through logistic regression (page 4). What are the values? Are they odds ratios? B values (intercepts/coefficient for the constant)? If the latter, then how does the AGPi term incorporate the B’s standard error?

 

Thank you for pointing this out. However, in the case of our study, presenting the beta values seems slightly out of scope because the focus was on the aggregated analysis provided by the Inequality Factor (D) and not in the explanatory variables separately. We clarified these issues in the lines 130-140.  

 

 

1. The response variables (page 4) for water (access to a piped system), do not really mean the households have water, just that they have access to a water distribution system. In many places in Latin America, households have access to a piped water system, but no water or very poor water quality (so no potable or drinkable water). This is a significant problem using this variable. The authors mention this at the end of the discussion section on page 16, but this very important limitation should be noted earlier in the methods section. In fact, there are several limitations to the data used, and so the authors should consider adding a sub-section on data limitations in section 2.

 

You have raised an important point here. Although we did not create a new sub-section, we have revised the text to clarify data limitations. See Table 1 and lines 156-161.

  

1. The explanatory variables were derived from “bibliographic research” (pages 4-5). Can you please be more specific?

 

We agree with the suggestion and have incorporated the following references (line 136):

3 - REZENDE, S.; WAJNMAN, S.; CARVALHO, J. A. M.; HELLER, L. Integrando oferta e demanda de serviços de saneamento: análise hierárquica do panorama urbano brasileiro no ano 2000. Eng. Sani. Ambient., v. 12, n. 1, 2007; 90-101.

17 - WHO/UNICEF/JMP. Progress on household drinking water, sanitation and hygiene 2000-2017. Special focus on inequalities. New York: United Nations Children’s Fund (UNICEF) and World Health Organization (WHO), 2019.

23 - WHO/UNICEF JMP. Inequalities in sanitation and drinking water in Latin America and the Caribbean. A regional perspective based on data from the WHO/UNICEF Joint Monitoring Programme (JMP) for using recent national household surveys and censuses, 2016; 11.

 

1. The ‘race or color’ variable “white/non-white” (page 5) seems very problematic. Overall, the use of “white” and “black” is confusing in this article, such as on page 12. Race is a cultural construction and so what “white” means in one cultural context is not the same across all contexts. The authors will need to justify and defend the use of this variable, which weakens the utility of the analysis.

Table 1 was changed to better explain the “Race or Color” variable and its importance.  

1. There seems to be an overlap between the ‘literate/non-literate’ variable and the ‘educational attainment’ variable, since one cannot attain education if they are illiterate (page 5). For the model, this will result in double-counting the weight of this variable. The authors should remove one of these or justify and defend the use of both, noting the problems with the overlap.

 

Table 1 was modified to explain the variables “Literacy” and “Educational Attainment” in more detail and to stress the differences between them. We would like to take this opportunity to clarify that the latter refers to the educational attainment of the household head. We apologize for the mistake.  

 

1. How does the Inequality Factor equal a percentage, as displayed in Figures 1 and 2. A percentage of what? Or do the authors mean for this to be displayed as a proportion or a scaled value (0-1)?

 

Thank you for pointing this out. We have modified the text to clarify this question (lines 190-191 and 195-196).

 

1. On page 14, the authors note: “There was a correlation of approximately 60% between the two methods,” citing an R-squared of 0.6019. The correct interpretation of this term is that the relationship (linear model) explains 60% of the variation in the dependent variable (“intersectionality”—which is misspelled on the Y axis in Figures 11 and 12).

 

We agree and have rectified the interpretation (lines 317-318 and 332-333).

Author Response File: Author Response.pdf

Reviewer 2 Report

This is a very interesting paper and I congratulate the authors for it. It is a good contribution and fill a gap in the literature. 

In my view, the paper will be ready for acceptance after the following recommendations: 

a) The english language needs to be polished and several typos need to be removed

b) The authors should be more prospective in conclusions and highlighting some policy implications of their research, in particualr how the inequality can be mitigated

Author Response

Dear Reviewer,

 

We appreciate the time and effort that you have dedicated to providing your valuable feedback on our manuscript. We revised the text and believe that the result is a clearer and improved paper. Here is a point-by-point response to your comments and concerns:

 

 

 

1. The english language needs to be polished and several typos need to be removed

 

Thank you for pointing this out, we apologize for this. We have revised the text to correct spelling and grammatical errors.

 

1. The authors should be more prospective in conclusions and highlighting some policy implications of their research, in particualr how the inequality can be mitigated

 

Although our paper does not provides indications on how to mitigate inequalities, we have modified the conclusions to suggest how the results can contribute to improve monitoring and, thus, support the formulation of more effective public policies (lines 388-393).

Author Response File: Author Response.pdf

Reviewer 3 Report

The manuscript advances two methodologies to address inequalities in analysis of access to WASH services, the first one adjusting to the indicators to of access to these services, and the second one offers an assessment of intersecting forms of inequality. I strongly believe that this paper will be very useful to WASH policy makers, practitioners and researchers, as the methodologies proposed are novel and well justified. I also think that the greater merit of this paper is the fact that both of these methodologies can be used to complement one another to provide a more nuanced analysis of aggregate data of inequality. The paper is also very well written and presented. I recommend that this paper be published in its current form. My only recommendations are the following:

It would be very interesting to run an analysis to see which explanatory variables (e.g. urban-rural status, geographical location, race, indigenous group, literacy etc) and their combinations are most significant. I don't know if this is possible, but it if I highly recommend to include it in the document.

 

Author Response

Dear Reviewer,

 

We appreciate the time and effort that you have dedicated to providing your valuable feedback on our manuscript. We revised the text and believe that the result is a clearer and improved paper. Here is a point-by-point response to your comments and concerns:

 

 

It would be very interesting to run an analysis to see which explanatory variables (e.g. urban-rural status, geographical location, race, indigenous group, literacy etc) and their combinations are most significant. I don't know if this is possible, but it if I highly recommend to include it in the document.

 

Thank you for pointing this out, it is indeed a valuable suggestion. However, it seems slightly out of scope in the case of our study because the focus was on the aggregated analysis provided by the Inequality Factor (D) and not in the explanatory variables separately. Nevertheless, we have changed the text to incorporate this suggestion as a recommendation for future research. (lines 139-140).

Author Response File: Author Response.pdf