Just as enrollment rates and exam results attract attention and controversy in the education sector, so estimates of access to clean and safe water and functionality rates (what percentage of waterpoints are functioning at any given time) are fought over at length.
Without getting too technical, the problem stems from how you collect data. When trying to measure access, do you collect up data on how much rural water supply infrastructure - pumps, pipes, protected springs, etc. - and make assumptions on how many people access water from each of those sources? Or do you collect information from a sample of households - asking them what source they actually use?
If you use infrastructure data, your final figure depends on assumptions about how many people use each waterpoint, which are easy to get wrong. This approach also leaves you vulnerable to under-reporting of breakdowns - if a broken down water supply scheme is still included, then you will get an over-estimate of access. Finally, this approach also suffers from a conflict of interest since the very people who are usually responsible for collecting information on rural waterpoints are also responsible for new investment and keeping older schemes functioning - they are effectively reporting on their own performance.
It's much easier to construct new infrastructure than to ensure people use it. Does the infrastructure breakdown - is it maintained and repaired properly? Does the water source dry up? Can people afford to pay for clean water and do they know the importance of using water from an improved source? This is very similar to the distinction between enrollment and quality in the education sector: it's relatively easy to increase enrollment by building schools and scrapping fees but much harder to ensure children are actually learning.
For all these reasons, there is consensus internationally that using household survey data (though imperfect) is a more reliable method. The UNICEF-WHO Joint Monitoring Programme (JMP), for example, which is responsible for monitoring progress against the Millennium Development Goals for water and sanitation, only uses survey data to report on access.
In Tanzania, official data (using infrastructure-based routine monitoring) reports access to be as high as 58% (and increasing), compared to 40% (and declining) reported by households themselves through the latest household surveys.
Tanzania's Ministry of Water knows this; they know all the weaknesses with using routine monitoring of infrastructure to estimate access. And yet the figures the Ministry consistently focus on are based on routine monitoring and the Ministry takes little notice of survey data.
Why? Because they don't want to admit that the data they've been collecting for years is misleading. Because they don't want to admit that access to clean and safe water in rural areas is actually much lower than they've been reporting. Because explaining the lower - but more accurate - figures to the Minister and to Parliament is too sensitive. In short, because of politics.
In the latest example, two submissions from the water sector into the ongoing process to develop MKUKUTA II* both included a specific recommendation that access to clean and safe water should be reported using household survey data rather than routine monitoring. I had a chance this week to look at the current draft of MKUKUTA II, only to find that the relevant indicators are entirely based on routine monitoring of infrastructure.
If Tanzania really wants to get to grips with rural water supply, we surely have to first be honest about the current state of access to clean and safe water. If we keep telling ourselves and our political leaders that the situation is not that bad, nothing's going to change.
* UNICEF's Meeting the Water, Sanitation and Hygiene Rights of Tanzanian Women and Children (final version not yet online, earlier draft here) and an official review of performance monitoring systems for the water sector. This blog's author was involved in preparing both reports.