Measuring Access and Practice: Designing a Survey Methodology for the Hygiene, Sanitation and Water Sector

Access to safe water and sanitary means of excreta disposal are essential elements of human development and poverty alleviation. It is estimated that one in four people in the developing world lacks access to water while over half the population has no access to sanitation. From the Alma-Ata declaration in 1978 to the recent Millennium Development Goals, efforts to improve this situation have been hampered by the lack of meaningful indicators to measure hygiene, sanitation and water coverage and establish progress towards the goals and targets set out by the international community. This thesis aims to determine if measuring prevalence of access to water~ sanitation and the practice of hygienic behaviour in hous~hold surveys can be.improved. With no indicators available in current international' laws and targets, various aspects of access and practice were examined to design indicators for field-testing. By using - existing data sets, the research established that there is a high geographic clustering of the measures of interest, which results in large design effects (deff) and rates of homogeneity (roh) in cluster surveys. Based on the calculated roh optimum numbers ofcluster and sample size were calculated for the field trials. This requires introducing survey costs in the sample size calculations. The high clustering of water and sanitation indicator require large sample sizes, resulting in large amounts of data which organisations in the four field trials in Kosovo, South Africa, Kenya and Laos found difficult to handle. Practical problems in the implementation of the survey method resulted in non-sampling errors and could cause reluctance in adoption the methodology. The research improved water and sanitation indicators but found that for individual behaviour such as hygiene the household is not a suitable sampling unit. It also showed that observation among interviewers have to be better standardised to reduce the inter-surveyor.variation. Representative sampling is the current bottleneck in the development of such a survey method. Current method requires a good understanding of sampling theory as well as reliable sample frames, which are rarely available to implementing organisations. Alternative sampling methods are suggested, and recommendations are made for the further development ofthe survey method designed in this research, which to date may be too complex for widespread use.