JMP's method - deriving progress estimates
For each country, survey and census data are plotted on a timescale from 1980 to the present. A linear trend line, based on the least-squares method, is drawn through these data points to provide estimates for all years between 1990 and 2010 (wherever possible). The total estimates are population weighted average of the urban and rural numbers.
Sanitation trend analyses at country level are made for improved sanitation facilities and open defecation. The estimates for improved sanitation facilities presented are discounted by the proportion of the population that shared an improved type of sanitation facility. The ratio (proportion of the population that shares an improved sanitation facility between two or more households, including those used a public toilet) derived from average of all available ratios from household surveys and censuses is subsequently subtracted from the trend estimates of improved sanitation facilities, and this gives the estimates for shared sanitation facilities.
Drinking-water trend analysis at the country level is carried out for the following categories: piped water into dwelling, plot or yard; improved sources of drinking-water and surface water. [read more…]
The total estimates are based on the aggregate of the population-weighted urban and rural numbers, divided by the total population. The estimates for the use of improved sanitation facilities are discounted by the proportion of the population that shared an improved type of sanitation facility. The ratio (the proportion of the population that shares a sanitation facility of an otherwise adequate type among two or more households) derived from the latest household survey/census is subtracted from the trend estimates of improved sanitation facilities. This results in the estimates for shared sanitation facilities.
Regional averages
Regional estimates are calculated when the available data cover at least 50 per cent of the population in a region. For this report, insufficient data were available to disaggregate each of the rungs of the sanitation ladder for the developed regions, the Commonwealth of Independent States and Oceania. The MDG regional groupings have been used in all regional analyses and tabulations.
Forecasting
Wherever necessary, JMP extrapolates the linear regression line up to two years before or after, the earliest or the latest data point. [read more…]
Outside of these time limits, the extrapolated regression line is flat for up to four years. If the extrapolated regression line reaches 100% coverage or beyond, or 0%, a flat line is drawn from the year prior to the year where estimates would reach 100% (or 0%).
Where insufficient data exists for linear regression, the slope of the regression is assumed to be zero, i.e. that no progress is made. This might occur in two scenarios: when only one valid datum is available, in which case the value is therefore taken forward; or when there are two or more valid data points available, but they are spaced four or fewer years apart, in which case the average of the available data points is used. When the slope of the regression is assumed to be zero, the projection can be made up to a maximum of six years forwards and backwards in time from the data point. When the use of improved facilities is at 95% or above, or at 5% or below, the projection can be made without limitations.
