ABSTRACT
The chapters that precede this develop theories and methods of survey sampling under the suppositions that we have a target population of individuals that can be identified and, using labels for identification of the units, we choose a sample of units of a desired size and derive from them values of one or more variables of interest. However, to execute a real-life sample survey, one usually faces additional problems. Corresponding to a target population one has to demarcate a frame population, or frame for short, which is a list of sampling units to choose from, or a map in case of geographical coverage problems. The target and the frame often do not exactly coincide. For example, the map or list may be outdated, may involve duplications, may overlap, and may together under or over cover the target. Corresponding to a frame population one has the concept of a survey population, which consists of the units that one could select in case of a 100 percent sampling. These two also need not coincide because during the field enquiry one may discover that some of the frame units
may not qualify as the members of the target population and hence have to be discarded to keep close to the target. The field investigation values may be unascertainable for certain sections of the survey populations, or, even if ascertained, may have to be dropped because of inherent inconsistencies or palpable inaccuracies at the processing stage. Consequently, the sample data actually processed may logically yield conclusions concerning an inference population, which may differ from the survey population. MURTHY (1983) elegantly enlightens on these aspects.
