Interpersonal Conflict over Water Is Associated with Household Demographics, Domains of Water Insecurity, and Regional Conflict: Evidence from Nine Sites across Eight Sub-Saharan African Countries

Round 1

Reviewer 1 Report

This is an interesting contribution to our understanding of water-related conflict and how it associates which demographic and contextual factors.

The paper is based on empirical evidence, and the empirical basis of the paper is clearly described.

The empirical basis for the paper seems to entail the following weaknesses:

Based on my own field work experience, I would hesitate to rely on self-reported data obtained on income levels, based on recall. Adding to this, the sd of income levels reported in tables 2 and 3 also seem quite large, and also the somewhat self-contradictory result which you report (line 330-331) could also indicate that the data on income level is not reliable. You may therefore consider not using this data, due to doubtful reliability.

I note that the question which you use as a proxy for within household water-related conflict is phrased in a manner which may actually cover a number of other aspects apart from 'within household conflict', namely as "...problems with water that caused difficulties within your household?” (line 158-159). Such a problem could also be unsafe water which caused diarrhea, i.e. something quite different from 'water-related conflict'. In my view, that limits your ability to meaningfully analyse intra-household water-related conflict on the basis of your material.

A higher degree of engagement with and awareness of context in the interpretation of results would also be welcome. For instance with respect to within household conflicts, more detailed consideration of household composition (age of kids, number of adults etc.) would be in place, as well as a more general consideration of whether what you identify as intra-household water-related conflicts are in fact conflicts related specifically to water or e.g. conflicts related to the distribution of work load within the household that also occur with respect to other aspects of household duties. 

I would encourage you to consider splitting your sample into an urban and a rural sample, to test whether your conclusions hold true in each of the sub-samples. The reason for that being that water infrastructure but also a number of the demographic and contextual factors which you examine are quite different in rural and urban settings, respectively.

Finally, I would encourage you to pay even more attention to not suggest causal relationships based on your conclusions, when what you find is simply statistically significant correlations which merit further research.

Author Response

To The Editor,

Thank you for the opportunity to further revise this manuscript: ‘Interpersonal conflict over water is associated with household demographics, domains of water insecurity and regional conflict: Evidence from 9 sites across 8 sub-Saharan African countries’, in accordance with minor comments made from the reviewers. We have substantially modified the manuscript throughout. Importantly, we provide thorough edits to the Introduction, Methods, and Discussion to respond to reviewers’ comments.

Edits to the manuscript are provided in red font.

Our individual responses to the reviewer comments are outlined below:

Reviewer 1

Based on my own field work experience, I would hesitate to rely on self-reported data obtained on income levels, based on recall. Adding to this, the sd of income levels reported in tables 2 and 3 also seem quite large, and also the somewhat self-contradictory result which you report (line 330-331) could also indicate that the data on income level is not reliable. You may therefore consider not using this data, due to doubtful reliability.

RESPONSE: Indeed, measuring income is challenging, particularly across various settings in sub-Saharan Africa. The income variable was a relative measure within each site (rather than absolute income) to hopefully overcome some of these concerns. Encouragingly, if we omit income z-scores from our analyses (Models A and B), the associations between conflict and relative socioeconomic position did not change (remained negative and statistically significant at the p<0.10 level). For these reasons, we opted to keep this variable in our models. We now provide some citations on the known limitations of self-reported income in the Discussion section. We included a discussion of issues associated with self-reported income into the paper. This discussion appears in lines 371-382 which now read as follows:

“Here, it is important to note some differences in results in regards to the specification of income. We found higher levels of within-household conflict were associated with higher income and lower relative SEP. These conflicting results may be explained by issues associated with self-reported income in survey work [33-36]. Studies note for example that high income respondents report lower than actual earnings while low income respondents report higher than actual earnings [34, 35, 37]. However, in the absence of publicly available income tax and/or pension data, survey data are the only means of collecting income information, albeit imperfect [36]. To attenuate reporting problems associated with income, we included other measures of socio-economic status (e.g. relative socio-economic position).”

I note that the question which you use as a proxy for within household water-related conflict is phrased in a manner which may actually cover a number of other aspects apart from 'within household conflict', namely as "...problems with water that caused difficulties within your household?” (line 158-159). Such a problem could also be unsafe water which caused diarrhea, i.e. something quite different from 'water-related conflict'. In my view, that limits your ability to meaningfully analyse intra-household water-related conflict on the basis of your material.

RESPONSE: This is an important distinction that we should have made clearer in the manuscript. Enumerator training for these two questions focused on our intention to capture conflicts related to problems with water. We modified the Methods section and cited the training manual to clarify this.

A higher degree of engagement with and awareness of context in the interpretation of results would also be welcome. For instance with respect to within household conflicts, more detailed consideration of household composition (age of kids, number of adults etc.) would be in place, as well as a more general consideration of whether what you identify as intra-household water-related conflicts are in fact conflicts related specifically to water or e.g. conflicts related to the distribution of work load within the household that also occur with respect to other aspects of household duties. 

RESPONSE: We now include a more detailed discussion of household composition, divisions of labor and conflict within the home, in relation to our findings. This addition appears in lines 364-371 of the manuscript which now reads as follows:

“These results are consistent with prior work which finds below poverty level income [66] and the presence of children in the household [67-69] increase the risk of interpersonal violence. They are also consistent with studies of water collection in Sub-Saharan Africa which find that women and children spend a disproportionate amount of time fetching water [70, 71] and that the amount of time spent fetching water is associated with higher risk of spousal abuse for women within households [54].”

I would encourage you to consider splitting your sample into an urban and a rural sample, to test whether your conclusions hold true in each of the sub-samples. The reason for that being that water infrastructure but also a number of the demographic and contextual factors which you examine are quite different in rural and urban settings, respectively.

RESPONSE: The reviewer correctly mentions that many contextual factors differ in rural vs. urban areas and encourages splitting the sample rural vs. urban. In our main analyses, our models include contextual variables more directly that relate to contextual differences between rural and urban (e.g., water infrastructure/source, etc). We also include population density, as a measure of urbanicity. However, as a sensitivity analysis, we ran our most complex models (E and F) stratified by rural/urban status (new Tables 7 and 8). We also added descriptive statistics for conflict levels in rural areas in Table 2. We now discuss these findings and their implications in the Discussion.

Finally, I would encourage you to pay even more attention to not suggest causal relationships based on your conclusions, when what you find is simply statistically significant correlations which merit further research.

RESPONSE: We have modified language in the Discussion to ensure that causation is not inferred from our cross-sectional analyses.

 Reviewer 2

• the last author is not a person and should not be listed as a coauthor.. editors should have picked on this before sending this paper out for review. 

RESPONSE: There was an error with the last author’s email address, so perhaps this was the reason for the reviewer thinking the last author was ‘not a person’. We have corrected this in the portal.

• what is striking is that while the literature review includes discussion on the relation between water scarcity and conflict, it does not include the latest literature on the topic, omitting in particular the UK based research on the topic (unsurprusingly given the affiliation of most coauthors) being USA based. See and include therefore the research on water scarcity and transboundary water governance of UK scholarship, including: Tony Allan (SOAS/KCL), Naho Mirumachi (KCL), Leyla Mehta (Sussex), Mark Zeitoun (UEA), Hussam Hussein (Oxford), Dustin Garrick (Oxford). 

RESPONSE: We revised the introduction to the paper to incorporate this vital aspect of work on water and both conflict and cooperation. The new text appears in lines 51-64 of the manuscript which now reads as follows:

“These changes and disparities are exacerbated by water intensive lifestyles related to increasing rates of urbanization and economic development, which promote unsustainable withdrawals of fresh water [4]. Macroeconomic changes and politial upheaval further undermine equitable water availability, accessibility, affordability, quality, and reliability [5, 6].

The increasing interconnectedness and transboundary nature of water flows and transactions render water governance a global issue (Mehta and Movik, 2014; Cooley et al., 2013; Pahl-Wostl et al., 2013). Consequently, conflict occurrence over water resources is multiscalar (Carius et al., 2004; Burke, Hsiang, Miguel, 2015) and highly contextual (Garrick et al., 2019) that ranges from visible conflicts in the form of water wars to less visible forms of conflict in the form of contested power relations (Zeitoun et al., 2020). At the nation-state level transbounary water agreements are a common strategy for grappling with the governance challenges associated with the uneven distribution of water and transboundary nature of waterways (Zeitoun et al., 2020).”

New references incorporated into the paper were as follows:

• The literature should also include the issue of power in transboundary water governance, and therefore mention of the critical hydropolitics literature (see the work of Jeroen Warner, of Mark Zeitoun, and of Ana Elisa Cascao, among others). 

This should help in engaging with the literature in the first round of revision. 

RESPONSE: We revised the introduction to the paper to incorporate this aspect of work on water and both conflict and cooperation. The new text appears in lines 64-71 of the manuscript which now reads as follows:

“A critical hydropolitcal perspective on these agreements suggests they are shortsighed stop-gaps to conflict that ignore or fail to consider the socio-ecological dynamics driving conflict over water resources [12-14]. From this perspective, transboundary agreements are a good example of hydro-hegemony [15, 16] because they suffer from asymmetric power-relations that better position the powerful party to leverage a variety of tactics and strategies to move towards consolidated control over water resources [17].”

 OVERALL RESPONSE: Thank you for your helpful reviews which have improved this manuscript.

Reviewer 2 Report

I have read with interest this paper on water conflict at different scales. My comments to improve the paper are as follow:

• the last author is not a person and should not be listed as a coauthor.. editors should have picked on this before sending this paper out for review. 
• what is striking is that while the literature review includes discussion on the relation between water scarcity and conflict, it does not include the latest literature on the topic, omitting in particular the UK based research on the topic (unsurprusingly given the affiliation of most coauthors) being USA based. See and include therefore the research on water scarcity and transboundary water governance of UK scholarship, including: Tony Allan (SOAS/KCL), Naho Mirumachi (KCL), Leyla Mehta (Sussex), Mark Zeitoun (UEA), Hussam Hussein (Oxford), Dustin Garrick (Oxford). 
• The literature should also include the issue of power in transboundary water governance, and therefore mention of the critical hydropolitics literature (see the work of Jeroen Warner, of Mark Zeitoun, and of Ana Elisa Cascao, among others). 

This should help in engaging with the literature in the first round of revision. 

Author Response

To The Editor,

Thank you for the opportunity to further revise this manuscript: ‘Interpersonal conflict over water is associated with household demographics, domains of water insecurity and regional conflict: Evidence from 9 sites across 8 sub-Saharan African countries’, in accordance with minor comments made from the reviewers. We have substantially modified the manuscript throughout. Importantly, we provide thorough edits to the Introduction, Methods, and Discussion to respond to reviewers’ comments.

Edits to the manuscript are provided in red font.

Our individual responses to the reviewer comments are outlined below:

Reviewer 1

Based on my own field work experience, I would hesitate to rely on self-reported data obtained on income levels, based on recall. Adding to this, the sd of income levels reported in tables 2 and 3 also seem quite large, and also the somewhat self-contradictory result which you report (line 330-331) could also indicate that the data on income level is not reliable. You may therefore consider not using this data, due to doubtful reliability.

RESPONSE: Indeed, measuring income is challenging, particularly across various settings in sub-Saharan Africa. The income variable was a relative measure within each site (rather than absolute income) to hopefully overcome some of these concerns. Encouragingly, if we omit income z-scores from our analyses (Models A and B), the associations between conflict and relative socioeconomic position did not change (remained negative and statistically significant at the p<0.10 level). For these reasons, we opted to keep this variable in our models. We now provide some citations on the known limitations of self-reported income in the Discussion section. We included a discussion of issues associated with self-reported income into the paper. This discussion appears in lines 371-382 which now read as follows:

“Here, it is important to note some differences in results in regards to the specification of income. We found higher levels of within-household conflict were associated with higher income and lower relative SEP. These conflicting results may be explained by issues associated with self-reported income in survey work [33-36]. Studies note for example that high income respondents report lower than actual earnings while low income respondents report higher than actual earnings [34, 35, 37]. However, in the absence of publicly available income tax and/or pension data, survey data are the only means of collecting income information, albeit imperfect [36]. To attenuate reporting problems associated with income, we included other measures of socio-economic status (e.g. relative socio-economic position).”

I note that the question which you use as a proxy for within household water-related conflict is phrased in a manner which may actually cover a number of other aspects apart from 'within household conflict', namely as "...problems with water that caused difficulties within your household?” (line 158-159). Such a problem could also be unsafe water which caused diarrhea, i.e. something quite different from 'water-related conflict'. In my view, that limits your ability to meaningfully analyse intra-household water-related conflict on the basis of your material.

RESPONSE: This is an important distinction that we should have made clearer in the manuscript. Enumerator training for these two questions focused on our intention to capture conflicts related to problems with water. We modified the Methods section and cited the training manual to clarify this.

A higher degree of engagement with and awareness of context in the interpretation of results would also be welcome. For instance with respect to within household conflicts, more detailed consideration of household composition (age of kids, number of adults etc.) would be in place, as well as a more general consideration of whether what you identify as intra-household water-related conflicts are in fact conflicts related specifically to water or e.g. conflicts related to the distribution of work load within the household that also occur with respect to other aspects of household duties. 

RESPONSE: We now include a more detailed discussion of household composition, divisions of labor and conflict within the home, in relation to our findings. This addition appears in lines 364-371 of the manuscript which now reads as follows:

“These results are consistent with prior work which finds below poverty level income [66] and the presence of children in the household [67-69] increase the risk of interpersonal violence. They are also consistent with studies of water collection in Sub-Saharan Africa which find that women and children spend a disproportionate amount of time fetching water [70, 71] and that the amount of time spent fetching water is associated with higher risk of spousal abuse for women within households [54].”

I would encourage you to consider splitting your sample into an urban and a rural sample, to test whether your conclusions hold true in each of the sub-samples. The reason for that being that water infrastructure but also a number of the demographic and contextual factors which you examine are quite different in rural and urban settings, respectively.

RESPONSE: The reviewer correctly mentions that many contextual factors differ in rural vs. urban areas and encourages splitting the sample rural vs. urban. In our main analyses, our models include contextual variables more directly that relate to contextual differences between rural and urban (e.g., water infrastructure/source, etc). We also include population density, as a measure of urbanicity. However, as a sensitivity analysis, we ran our most complex models (E and F) stratified by rural/urban status (new Tables 7 and 8). We also added descriptive statistics for conflict levels in rural areas in Table 2. We now discuss these findings and their implications in the Discussion.

Finally, I would encourage you to pay even more attention to not suggest causal relationships based on your conclusions, when what you find is simply statistically significant correlations which merit further research.

RESPONSE: We have modified language in the Discussion to ensure that causation is not inferred from our cross-sectional analyses.

 Reviewer 2

• the last author is not a person and should not be listed as a coauthor.. editors should have picked on this before sending this paper out for review. 

RESPONSE: There was an error with the last author’s email address, so perhaps this was the reason for the reviewer thinking the last author was ‘not a person’. We have corrected this in the portal.

• what is striking is that while the literature review includes discussion on the relation between water scarcity and conflict, it does not include the latest literature on the topic, omitting in particular the UK based research on the topic (unsurprusingly given the affiliation of most coauthors) being USA based. See and include therefore the research on water scarcity and transboundary water governance of UK scholarship, including: Tony Allan (SOAS/KCL), Naho Mirumachi (KCL), Leyla Mehta (Sussex), Mark Zeitoun (UEA), Hussam Hussein (Oxford), Dustin Garrick (Oxford). 

RESPONSE: We revised the introduction to the paper to incorporate this vital aspect of work on water and both conflict and cooperation. The new text appears in lines 51-64 of the manuscript which now reads as follows:

“These changes and disparities are exacerbated by water intensive lifestyles related to increasing rates of urbanization and economic development, which promote unsustainable withdrawals of fresh water [4]. Macroeconomic changes and politial upheaval further undermine equitable water availability, accessibility, affordability, quality, and reliability [5, 6].

The increasing interconnectedness and transboundary nature of water flows and transactions render water governance a global issue (Mehta and Movik, 2014; Cooley et al., 2013; Pahl-Wostl et al., 2013). Consequently, conflict occurrence over water resources is multiscalar (Carius et al., 2004; Burke, Hsiang, Miguel, 2015) and highly contextual (Garrick et al., 2019) that ranges from visible conflicts in the form of water wars to less visible forms of conflict in the form of contested power relations (Zeitoun et al., 2020). At the nation-state level transbounary water agreements are a common strategy for grappling with the governance challenges associated with the uneven distribution of water and transboundary nature of waterways (Zeitoun et al., 2020).”

New references incorporated into the paper were as follows:

• The literature should also include the issue of power in transboundary water governance, and therefore mention of the critical hydropolitics literature (see the work of Jeroen Warner, of Mark Zeitoun, and of Ana Elisa Cascao, among others). 

This should help in engaging with the literature in the first round of revision. 

RESPONSE: We revised the introduction to the paper to incorporate this aspect of work on water and both conflict and cooperation. The new text appears in lines 64-71 of the manuscript which now reads as follows:

“A critical hydropolitcal perspective on these agreements suggests they are shortsighed stop-gaps to conflict that ignore or fail to consider the socio-ecological dynamics driving conflict over water resources [12-14]. From this perspective, transboundary agreements are a good example of hydro-hegemony [15, 16] because they suffer from asymmetric power-relations that better position the powerful party to leverage a variety of tactics and strategies to move towards consolidated control over water resources [17].”

 OVERALL RESPONSE: Thank you for your helpful reviews which have improved this manuscript.

Round 2

Reviewer 1 Report

I appreciate the efforts made, particularly with respect to citing a wider body of research (however, without this apparently having implications for the presentation or the discussion of research results), but also with respect to applying detailed analysis for rural and urban samples. By doing so, it seems that all household specific characteristic-associated variables are no longer significantly associated with inter- or intrahousehold water-related conflict in rural areas. Thus water-related conflict is associated with aspects which relate to the quality of water-supply infrastructure. In urban areas, you still find household characteristics associated with water-related conflict. My best guess is that that is because your category urban areas cover poor as well as non-poor neighbourhoods. THerefore, I would still challenge your conclusion that "Overall, our results indicate that households that experienced internal conflict and those that experienced conflict with neighbors differed from households that did not experience these types of conflict." (lines 343-345) in the sense that rather than individual features of the household, the underlying cause is that low-income etc. household, the to be living in neighbourhoods with largely insufficient water supply infrastructure. Your current presentation of data still masks this potential explanation. 

Author Response

We agree and have added text at the end of this paragraph acknowledging the potential role of the water infrastructure (and potential correlation with income levels) in lines 355-360. This addition now reads as follows:

“But we note that income may be correlated with availability and accessibility of different forms of water infrastructure. Residential areas with a high proportion of low income households are more likely to have poorer water infrastrcuture than residential areas with high-income households so it is important to acknowledge that the relationship between income and water insecurity may be associated with factors beyond the household context.”

Reviewer 2 Report

Looks much better. One little comment: on line 57, when you discuss multiscalar, I would suggest you to include also the work:

• da Silva, L. P. B. (2019). Production of scale in regional hydropolitics: an analysis of La Plata River Basin and the Guarani Aquifer System in South America. Geoforum, 99, 42-53.

apart from this, well done. 

Author Response

We have inserted this reference into the text on line 57.