ABSTRACT
Kiaer of the Central Bureau of Statistics Norway initiated the introduction of sampling in a scientific way as the population in a miniature form to be carefully chosen and studied as an adequate 'representative' of the population. Though Kiaer recommended only 'purposive sampling' leaving no scope for probabilistic approach of inference-making Bowley supported the 'Sampling approach as a way to inference-making postulating normality and encouraging large sizes of samples. In the early stages of the 1920s opinion polls based on random samples were popular. The parameters of interest in survey sampling are well-behaved functions of population totals of real-variables. Each such total has to be estimated suitably by appropriate statistics based on samples to be chosen according to a 'sampling scheme' or equivalently 'sample design'. George Gallup introduced non-random 'quota sampling'. There was some exciting initial support for it.
