[BLANK_AUDIO] Our guest today is Mark Jutland, Mark is an assistant professor at the Sanford School of Duke University, in Durham, North Carolina. He's an environmental engineer and water resources systems analyst, he teaches in the area environment development, and does research in these areas, including the area global public health. He's done some really interesting research recently on household water demand in India, in Cambodia and he's going to talk to us about that today. So Mark I'm just delighted you're here, thanks so much for coming over. >> Thank you inviting me. >> So let's start in let me ask you about your research questions, I mean what were your research question in Cambodia and India, what were you trying to do? >> Sure. And I think, as I answer this question, it's probably best to give a little bit of history about these, these two sites. The first one, which is in India, in Andhra Pradesh, this is Southeastern India was actually a study that I got involved with at a later stage and it was a follow-on on a community-level trial of these water treatment plants, and they're advanced water treatment plants. The idea of these plants was that if you installed highly advanced treatment at the community level. And made that available to households for a fee, that they would perhaps purchase better quality water and consume better quality water. What they found in that study was that use rates were quite low from these treatment plans. And so, we asked the question of whether maybe people weren't buying this water, because they didn't realize how contaminated their drinking water was. >> This was the RTI study that Chrissy Polis was involved in. >> That's correct. >> Right? Right. >> Yeah. >> So as I remembered the water was pretty expensive. >> It was not that expensive actually, if you, I mean I guess, relative to a free option that you have in the, in these communities, which is you know, hand pump, or, some. >> Mm-hm, mm-hm. But they had to pay when they went to the. >> They, they would pay about 2.5 to $0.05 depending on the community per 20 liter jerry can. So that's, you know, really not that expensive, and if we think about point of use treatment alternatives they would be comparable. >> Right, right. >> so. The water wasn't that expensive but there were other questions about convenience of these supplies, households had to walk long distances to get to these kiosks. And also you know just general inertia in households wanting to do what they've what they've routinely done. So the idea, then, was to do some simple water quality testing and information provision to households to see if a signal of contamination would motivate them to go and purchase. >> So in the previous day, they didn't know, really, whether the water was contaminated or not, so you're going to give them the information. >> That's right. The second study's a little bit different. Actually, at the inception phase of that study it helped to think about the design with a colleague, Joe Brown, who's now at the Georgia Institute of Technology. We were interested in the question of whether providing chlorinated treatment of irregular and kind of low dependability supplies of drinking water would help provide some health impact to communities. And this, these were in, in communities where water supplies were actually fairly convenient for households, so they had alternatives like rain water harvesting but also water networks. But the piped water networks didn't work all the time. They were intermittent and they were not typically treated in dependable way by system operators, so these were private systems, by and large. At the design phase, we thought we would try cluster randomized trial approach, in communities where some of them would get some kind of, inline or, unobtrusive chlorination technology, that could be put at the network level or even screwed on to taps to auto-chlorinate their water. But our funder for the study kind of backed out after our, our initial baseline study, didn't want to fund the full cluster randomized trials, so we decided we still had this rich set of baseline data on these communities, and we knew something about demand for water treatment following our baseline study. We decided to do a similar information experiment, but combine it with sale of point of use technologies and in this case aqua tabs, which is a tablet of chlorine, a disinfectant product. That households could use to, to treat their water. So, it was a similar idea in the end but a different history in terms of. >> Mm-hm. >> How the study was. >> Mm-hm. >> Developed. >> Mm-hm. So, so, what were your research questions then in the two place, as you have, you have both the interventions of the chlorine and you also have the information. >> Right. >> Treatment. What were you thinking your hypothesis would be? >> So in the first one it was a really simple question. It was this idea that we wanted to know of households would respond to a contaminated water signal. And purchase more water from these treatment plants that they weren't using. so, it was in part a, a question about whether people really know how contaminated their water is. And we didn't do extensive work trying to understand what their perceptions of, of drinking water quality were, but. We use this randomized information experiment to see if there would be response in among those who got information, and those who did not. In the other study the question was more about stimulation of demand for point of use treatment and there are lots of questions about point of use treatment and whether it's, it's a good idea; whether it's something that households see the value of. So we wanted to test that idea more specifically, and we did some more careful measurements of perceptions of drinking water quality, so we wanted to isolate the mechanisms through which demand might be effected. And see if it was simply salespeople armed with water quality tests that gives people some nervousness about their water supply or whether they actually understood the contamination signal that they got from the information. And so we tested that. >> So what did you find? let's, maybe we'll start with India. >> Right so in India we found that use of the packaged drinking water from these treatment plants increased by 50%, so that looks like a big effect, right, but it was actually a very modest effect. Because it was working off of a low baseline, a few. So, about 5 to 10% of households were purchasing households. So 50% effect is another 5% on top of that. And so, we see that some people do respond to the information. But that the vast majority don't change. Anything, really. And, there are multiple reasons why that might be. >> What do you think? Like what? >> It's kind of discouraging, right? I mean. >> Among them could be the possibility that this water is expensive, and they just decide that, you know, they're cash constrained. >> Mm-hm. >> They don't have the possibility. Another possibility is, is that they really know that this treated water is maybe not all it's cracked up to be. In fact they have to bring it home and store it. >> Mm-hm. >> And during the storage process that water can be contaminated, so this is the typical argument in favor of point of use treatment is that if you do community level interventions people don't end up better off because the water gets contaminated later. >> Mm-hm. >> Anyway. >> Mm-hm. >> We did also try to track other behaviors because the information was accompanied by a script that said, you can go and buy this water from drinking plants, the, the treated drinking water plants. Or you can do other things. You can treat your water in house. You can clean your containers more often. You can avoid storing water for long periods of time in your house. You can use safe methods of taking water out of your containers. And we did see some modest responses along each of those margins as well. But again, there's, there's, there was no kind of, signal of really large effect on any of those behaviors. They were all pretty modest effects. And this points, I think, to the complexity of water and sanitation interventions, and the fact that there's a whole set of things that house holds can do. And so you know, just targeting a single thing is often not going to get you the whole way towards safe water. >> I've got some more questions about India, but why don't we, why don't we just get the overview of Cambodia, what you found in Cambo, in, in Cambodia. >> Okay so I mentioned in, in Cambodia that we did some careful baseline surveys, ahead of doing this information experiment. And we learn some things in those baseline surveys. I highlight three things in particular, so the first thing that we learned, was that water quality is really bad, at the household level. We tested for ecoli, found high contamination rates. And it didn't really matter what the source was. So, we tested water from piped water supplies, rainwater storage, and as well as public taps. And we found contamination in all of them, by the time the household was consuming this water. And in fact, quality got worse in the house. So, we did the degradation in the water quality. Second thing we saw, because we did some contingent evaluation surveys at the time and questions. Was that, perceptions are not unrelated to demand for water quality. That's not surprising obviously it's very hard to say something casual about that link because the things that drive demand are are also maybe driving these perceptions. So that in large part motivated our experiment which was randomized. Information provision which would kind of shock perceptions. >> Mm-hm, mm-hm. >> And give us a clear signal through perceptions and, and demand. But we found by in large that people who thought their water was safer, actually had higher demand for water treatment, further water treatment. Which is consistent with the story that these are risk-averse people who are really worried about their water, and they just want to do everything they can to make, make it better. And then there's a whole other set of people who, sort of, don't really think their water's very safe but they also have very low demand. There are also maybe poor people and they are perhaps more constrained in their choices. So, those two things were some of the key findings from our baseline work. From the information experiment, we then learned that people do respond to information in the way we would expect. So a contaminated signal does change perceptions in the way that we would expect. I'm happy to talk about how we measured perceptions if that. >> Yeah. >> If that's of interest later. but, and, and, that that also translates into higher demand. So, a negative movement in perceptions does increase demand, and we see this, pretty consistently with the overall treatment effect. But the second main finding here is that there's heterogeneity in that impact, and I told you about these two types of people, well, it turned out that the two communities we worked in were basically different along the socioeconomic and risk profile lines. And that the community with initially sort of lower perceptions of drinking water quality, was actually the one that responded to information. So these were the people who didn't really care initially and had lower demand to start with. They were the ones who responded to information. The people who had higher demand initially. Didn't their perceptions didn't really change, so they were already thinking about what we're saying. >> The emphasize, the, the information here was, was actually information about the their quality, their individual water. >> That's correct. >> And how, how did you measure that at the household level? >> Right, so that the information, was provided through a household water test, and this was randomized to half of the households, stratified by these two communities. And it was a simple, presence absence test, which is not very specific for ecoli or specific contaminants in water. But indicates the presence of coliforms, so it's, it's called an H2S Test. The reason we chose it was because it provides saline information. It's easy to understand. Your test tube turns black if it's contaminated, it stays clear. >> Do they see it? Is it in their house? >> If they saw it. >> Yeah. [COUGH] >> And left the sample with them overnight. >> Mm-hm. >> So they observe the, the change. and, and so it's suited our purposes for that reason. You didn't have to take it back to a lab and incubate it and then come back and tell them your water is contaminated, they might not believe that,so this was simple in that sense. >> So you said in the case of Entraperdesh that the magnitude of the effect was small, what about the magnitude of the information treatment effect? In Cambodia. >> So similarly it was somewhat of a modest effect. But it's it's hard to compare the size of the effects because we were selling a very different product. >> Mm-hm. >> And in the case of Cambodia every household whether they got the information or not. Got a sales pitch to buy these Aqua Tabs products. And we found that about 50% of the sample purchased from Aqua Tabs, the quantity varied quite a bit and then, with the information, in one community. The effect was about ten percentage points. So. >> More. >> A 20%. >> Yeah. >> Effect. >> Right, right. >> But in the other community, no impact. So, again, the effect, sort of, in, in percentage points is roughly the same. It's about five percentage points. In the base line is, is different. >> So, in both cases, you've got a lot of people receiving information, and not responding. >> Right. >> To it. >> But in Cambodia interestingly enough half of the sample did buy the product even without the information. >> Mm-hm. >> So that tells us something about demand for for point of used treatment, at least short term demand. >> What what do you know about the the duration of the treatment effect, in these places. I mean, how long do you think this, this bump up in use is going to last. >> Right. So, this is a really good question. And we don't know a lot. Because we did these follow up surveys, relatively quick succession after the intervention. Under Pradesh, it was, the follow-up survey was one-month after. Information was provided. And in Cambodia was about six weeks. What we didn't do in India was then a subsequent sales of, something. Because they were responding to what their community had available or the behaviors that were available. We weren't combining it with a marketing approach, but in Cambodia we did try to sell products again at the second at the six week follow up to see if demand persisted. But it's hard to interpret those signals because a lot of households still had some Aquatabs. So they may not have seen the need to buy more. We didn't see a lasting effect. >> Why did they still have Aquatabs, why wouldn't they use them? >> Yeah, that's a really good question. So, one of the interesting, subsequent findings, if you dig into the data from the Cambodia study,. And we had asked people to keep their packaging. Now, that's kind of complicated because lots of households don't really care and they would throw away their packaging. And, but we found that those who have responded to the information. So that, those who got the test result and most of them were contaminated test results, they responded in this one community by purchasing. More Aquatabs, but that that impact was concentrated on people who actually did not use much of the Aquatabs, so that tells us that maybe. This information really doesn't accomplish that much, because households, in fact those that you get with the information, they're also the least likely to use the products over time. >> Do you have any information on health outcomes, actually? I mean, I mean that would tell you a little bit about whether they're using the Aquatabs regularly, if it was reliable. >> So this in term of broader implications of these studies. And in both of them we don't see a real health signal. We rely on self-reported, diarrheal disease, which is notoriously hard to measure. And the studies were not powered to see impacts on diarrheal disease. But even so, we don't see a health signal. So it's hard to say what the net effect of, of this, all this really is. and, and in fact if, what, what we see in, in other studies that I've been involved with is people sometimes compensate. If they do one thing they do less of another protective behavior. >> Actually I wanted to ask you about that in terms of the positive test, a positive test meaning you know? >> Yeah. >> Your water is not contaminated so. Do people stop doing good things when they find out from you that their, their water isn't contaminated after all? >> Right, so statistically, this is a hard question to answer with our sample, because we had 900 households in Cambodia. And about 90% of those tested for contaminated water. So we have a very small group of households, about 100, who didn't have contamination but we do see for them that their perceptions of water quality actually improve. So they thought their water was safer, the signal you would expect and that that seems to translate into lower demand for aqua tabs. Okay, so on one level maybe we shouldn't be worried. Their water was safe so they shouldn't be wasting their money on Aquatabs. But on another level, if they started doing other things that are compensating behaviors, taking other risks. >> Well it's just one sample too, right? I mean, they have one data point on the quality of their water so it's hard to judge you know? >> That's right. >> So, what would you say about the sort of broader implications of your work. What does it mean outside of Cambodia, in Andhra Pradeshterms of, what you're thinking about, sort of, information treatments as a policy intervention, and also point of use treatment as a policy intervention? >> I think I can pretty safely say that, Based on the literature and not just my own studies, but other findings in the literature, that information provision can be an effective strategy for promoting these health interventions and adoption of environmental health improvements. >> Mm-hm, mm-hm. >> But the effect is modest. So I mean, obviously it has to be combined with other strategies, if we really believe that a particular strategy is effecting. In terms of generalizability of these specific findings I would say that would we have to be pretty cautious about the results on these sites. And, and there are a couple of reasons for that. One is, in Cambodia self we see heterogeneity. Within communities in terms of who responds and who doesn't and that's related to all kinds of preferences and some of these socioeconomic factors that drive demand. But also across the sites we see different responses. And we see that across the two communities in Cambodia that we see completely different. One place doesn't respond to information, the other does. In India we see some responses, but again, the shift in behaviors isn't always concentrated on buying treated drinking water. Sometimes it's doing other behaviors. so, it's hard to say exactly what the effect of information will be in any given community. And, you know, we've thought about this in the context of how we think about cost and benefits of water and sanitation interventions, using simulation approaches. >> Mm, mm-hm. >> The parameters that we measure in these. Specific studies measuring these impacts or a, a single parameter that determines these cost and benefits and there is a whole other set of things. These households have other priorities. They also have other priorities with respect to drinking water. So, convenience maybe more important than quality. Or sometimes the aesthetics, the taste of the drinking water may be important. Another thing that we tested in Cambodia was whether households were sensitive to the taste of chlorine, and we did this double blinded taste tests. So neither the enumerator nor the household knew which of their samples had chlorine in it and we asked them their to identify their preferred samples. >> Where what'd you find. >> And we do see that people don't like the chlorinated samples once they reach a level that's the recommended treatment level for chlorine. So if the concentration is lower than that recommended level they don't really detect it, but if we start getting to the recommended treatment concentration. People don't like the water. So, we, we could assume that they would then probably respond in a point of view setting by not using the product as much. >> Do you think that would happen in a pipe system corner? >> So, that's a different question? And if they had alternatives, they might switch away and this was actually one of the original hypothesis that we wanted to test with these inline. Or non-intrusive chlorination devices was whether we'd see people switching to rainwater. Because in Cambodia they have lots of rainwater harvesting. But in systems where there's no choice and you get all of your water through the piped water supply, and that's the most convenient thing, then. Maybe the chlorine distaste is not such a big issue, and that's sort of the case in the U.S., and in developed countries, where we just drink the tap water because that's, that's what's there. >> Right. So, so how would you respond to somebody that says that they read your papers and they, and they said. They, they learn that the effect is modest from the information treatment, and people have this big gap between the reality of the water quality and what their perceptions of the water quality are. And, and somebody says we should just get on with this and get these people piped, water is chlorinated and good for their health. And this point of use treatment, information treatments, this is all kind of a dead end. >> I think it's an important question to ask. But the truth is that piped water systems are a long way off in a lot of these communities. At least dependable piped water systems that give people high quality water. So, I think in the meantime there are still some things that, that need to be worked on. And, and one of those things is kind of bridging this disconnect between perceptions of water quality and, and actual water quality. Water quality's really hard to observe. We find that households in Cambodia, their perceptions are actually completely uncorrelated with the tests. And I mentioned these Ecoli tests at baseline. We actually find a negative correlation between safety perceptions and the actual Ecoli contamination levels. So, people don't know. They don't observe this contamination and, and so on some level that's a problem because households should know, here in the US we know that our water's safe, we get these drinking water report cards and we take it for granted that, that this water is basically undrinkable. People in developing countries, they just don't have that information. They don't know. So they can't make these informed choices. So at some level, information is useful on, for at least informing people. But you can't do it in the one-shot sort of case. Like we do in these tests. I think you have to provide information over time. and, and test water and that may get expensive. >> Do you have any sense of the cost of information treatment relative to other interventions or sort of the cost effectiveness of information strategy like you've used in these research projects. >> So I think if you were to use these H2S tests, which are the tests we use in these papers, which are salient and give a presence perhaps and signal. It's pretty cheap. It's about a dollar a test. So, and it can be manufactured cheaper. We actually worked in Cambodia with a local lab to manufacture these products. And at scale, they could manufacture it I'm sure for fractions of a dollar a test. That is not very expensive and it may pay for itself. If you are the packaged treatment water provider. >> Mm-hm. >> You are selling this water at about $0.05 per jerry can. So you, if you got people buying for a month. You could provide them a monthly test. >> Interesting. >> That would pay for itself. >> Yeah. >> But whether it's a good business strategy, you'd have they'd have to work out whether that demand bump was worth it. >> If you were thinking about using an information treatment not for research but for a public health strategy and intervention to help people. You know, forget about your research papers. What would you recommend? When you, how would, how would you, how would you do this for. >> Now, I think that the public health case is a bit more complicated, than perhaps, the business case trying to sell one of these products or, or sell package treatment, treated drinking water. And the reason is because that H2S test is not a reliable or an indicator of health impact. So, if you wanted to really tell people their water was unsafe, you'd need to do more sophisticated microbial tests, and those start to become expensive. so, it, people have been trying to work on some cheaper ways of testing for Ecoli other pathogens of concern. In a variety of different labs and universities around the world. But those tests are still considerably more expensive than the simple H2S test. And at that point, you have to ask whether its cost effective, particularly given the fact that we don't see a clear impact on health. From providing this information. >> Do you think it would ever make sense for the businesses themselves that are selling the chlorine tablets to provide the test to households, to try to bump up demand? >> Yeah, I think, this was one of the things that we thought would maybe appeal to our partner who was thinking about these chlorinated products. And they are aware of the fact that households don't tend to know much about contamination. And they were pretty interested by the results, seeing that there would be this impact done to man. I think that they certainly would consider doing something like this. >> Well thank you so much Mark it's been a great, is there anything else you'd like to say about your findings, or anything that the students would want to know. >> No I think this has been great. >> Yeah. >> Thanks for having me. >> Oh, it's a pleasure, we're going to post the references for Mark's papers on both the Cambodia and the India work. At the end of the interview and so you'll be able to get access to the papers. >> Thank you. >> Thanks again. [BLANK_AUDIO]
