Historical Self-Comparison of Water Consumption as a Water Demand Management Tool
Round 1
Reviewer 1 Report
General comments:
After carefully reading the manuscript, I believe that this paper could be of interest to the broader international audience of Water only after major revision and clarification of some aspects. Therefore, I have some major suggestions the authors might want to consider:
1. Sampling strategy: The authors state that the households were “randomly sampled from a roster of survey registrants with a research company”. I think this statement is a bit vague and more information on the sampling strategy is missing: How was the sampling implemented? Does it ensure that the socioeconomic characteristics and consumption profiles are similar in the control and treatment groups? Some descriptive statistics regarding those characteristics in the three groups and means for the whole Census should be provided in order to ensure that the sampling is both random and representative of the population. Otherwise the study could not considered to be valid.
2. Empirical strategy:
The authors divide the households in the sample in four clusters, studying the effectiveness of the policy intervention from this perspective. Although the analysis is interesting, in my opinion it deviates the attention from the actual analysis of the effectiveness of the experiment/policy developed in the paper. I think it would be more illustrative to include a figure depicting the evolution of consumption in the control, and two treatment groups during the studied period, instead that the current Figure 3.
In addition, if the cluster analysis is included, some information on the method employed for the clustering should be given.
Other minor comments are the following:
1. Material and methods: I have a doubt concerning the experiment design: Had all the households to report their consumption or only the households in the intervention groups? The mere reporting could be able to generate both awareness and consciousness about water conservation, so it would be very difficult to isolate this effect from the effect of the use of the blue/colour drops if the control group does not have to report consumption as well.
2. Section “Water consumption data and evaluation of change”: The authors explained that water consumption per capita of each household was converted into a value assuming to be for a one-person household. Some more information on how they did this conversion as well as for the threshold to be designated as “high consumer” or “low consumer” should be provided.
3. Results section:
a. It would be interesting to provide also descriptive statistics of the finally employed sample (479 households) to know whether characteristics have changed a lot from the initial sample. That could potentially have implications for the study that should be offset.
b. The authors report the Chi-square tests for statistical differences among groups, but they do not report whether they are testing differences in means, medians or any other statistic. It would be important to include a reference.
Author Response
Thank you very much for reviewing our article. We revised our manuscript as follow:
1. Sampling strategy: The authors state that the households were “randomly sampled from a roster of survey registrants with a research company”. I think this statement is a bit vague and more information on the sampling strategy is missing: How was the sampling implemented? Does it ensure that the socioeconomic characteristics and consumption profiles are similar in the control and treatment groups? Some descriptive statistics regarding those characteristics in the three groups and means for the whole Census should be provided in order to ensure that the sampling is both random and representative of the population. Otherwise the study could not considered to be valid.
-> We inserted 3.1 Participants and described the socioeconomic characteristics of participants of the three groups (lines 143-154).
2. Empirical strategy:
The authors divide the households in the sample in four clusters, studying the effectiveness of the policy intervention from this perspective. Although the analysis is interesting, in my opinion it deviates the attention from the actual analysis of the effectiveness of the experiment/policy developed in the paper. I think it would be more illustrative to include a figure depicting the evolution of consumption in the control, and two treatment groups during the studied period, instead that the current Figure 3.
-> As the variance in residential water consumption is large, we thought it would not be appropriate to use a representative value such as an average value. Therefore, we analyzed the water consumption trend of each household.
In addition, if the cluster analysis is included, some information on the method employed for the clustering should be given.
->As described in 2.3, we used the K-Means clustering method. We added the method in Lines 118-122.
Other minor comments are the following:
1. Material and methods: I have a doubt concerning the experiment design: Had all the households to report their consumption or only the households in the intervention groups? The mere reporting could be able to generate both awareness and consciousness about water conservation, so it would be very difficult to isolate this effect from the effect of the use of the blue/colour drops if the control group does not have to report consumption as well.
-> Both the control group and the intervention groups reported their consumption. We also think that the reporting could generate awareness and consciousness. Therefore, the reporting was requested of all participants. We inserted ‘all’ (line 72).
2. Section “Water consumption data and evaluation of change”: The authors explained that water consumption per capita of each household was converted into a value assuming to be for a one-person household. Some more information on how they did this conversion as well as for the threshold to be designated as “high consumer” or “low consumer” should be provided.
-> We inserted the explanation in Lines 125-132.
3. Results section:
a. It would be interesting to provide also descriptive statistics of the finally employed sample (479 households) to know whether characteristics have changed a lot from the initial sample. That could potentially have implications for the study that should be offset.
-> We inserted 3.1 Participants and described the socioeconomic characteristics of the participants of the three groups (line 143-154).
b. The authors report the Chi-square tests for statistical differences among groups, but they do not report whether they are testing differences in means, medians or any other statistic. It would be important to include a reference.
-> A Chi-square test was used for analyzing the difference in the number of households classified into each cluster. As we mentioned before, the variance in residential water consumption is large. Therefore, we thought it inappropriate to use a representative value.
Author Response File: 
Author Response.docx
Reviewer 2 Report
Manuscript concerns important issue of efficient use of water based on historical self-comparisons of residential water consumption. Before taking decision about acceptance the following remarks should be explained. Add reference to the formula, in the case You are not the Author.
How You define “nudge” in Key words?
Number all equations.
Line 70-71. Why did You choose three groups for survey?
Line 111. LRPn, give some example of calculation in the next section.
Lines 120-122. Please give an example of the situation presented in the following question: we converted the water consumption per capita of each household in September into a value that was assumed to be for a one-person household.
Line 122-123. Give some obtained values of classification as ‘high consumer’ or ‘low consumer’, and the difference between number of residents and water usage. In section of results.
Why You choose the log-transformed relative proportion in calculation of LRPn?
Line 222. What do You mean by unlike usual nudge attempts?
On what base You decided to exclude the z values from the further research?
For the purpose of the assessment by the Reviewer, present questionnaire through Author response page.
How the obtained results will help practitioners? Add some comments to section of conclusion.
In conclusions add some information about future perspectives of work.
Author Response
Thank you very much for reviewing our article. We revised our manuscript as follow:
How You define “nudge” in Key words?
-> We think that ‘Nudge is a word that is already known to a certain extent, in that 31,900 search results come out when examined with Google Scholar.
Number all equations.
-> We inserted numbers.
Line 70-71. Why did You choose three groups for survey?
-> For statistical analysis, a certain number of cooperative families was needed. Therefore, three groups is the best for the analysis.
Line 111. LRPn, give some example of calculation in the next section.
> We inserted the example in lines 114-116.
Lines 120-122. Please give an example of the situation presented in the following question: we converted the water consumption per capita of each household in September into a value that was assumed to be for a one-person household.
-> We inserted the explanation in Lines 125-132.
Line 122-123. Give some obtained values of classification as ‘high consumer’ or ‘low consumer’, and the difference between number of residents and water usage. In section of results.
-> We inserted the explanation in Lines 125-132.
Why You choose the log-transformed relative proportion in calculation of LRPn?
-> As the variance in residential water consumption is large, we thought it better to analyze in proportion to the initial value of each household. The reason for using logs is to detect increases or decreases in value.
Line 222. What do You mean by unlike usual nudge attempts?
->As we described in line 225, usual nudge attempts did not exhibit such continuity [30,31].
On what base You decided to exclude the z values from the further research?
-> Water consumption will be 0 in the absence of the resident, and will increase if there are many visitors. As the water consumption changes greatly when an act different from normal life is performed, the z value was used to eliminate such unusual states as much as possible.
How the obtained results will help practitioners? Add some comments to section of conclusion.
-> We added the future perspectives in the conclusion (lines 245-246)
In conclusions add some information about future perspectives of work.
-> We added the future perspectives in the conclusion (lines 256-258)
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
All the comments have been properly addressed.
There is only a tipographic error, the test is “Kruskal-Wallis” instead of “Kruska” as it already reads.
Author Response
Thank you very much for reviewing our article again.
I revised a typo (Line 151).
