Lyberga and Cassela, 2004

Lyberga, L., and C. M. Cassela. 2004. Sample surveys: The field. International Encyclopedia of the Social & Behavioral Sciences:13475-13480.

Survey sampling is the art of selecting a sample of units from a population, measuring these sampled units and then estimating population parameters based on these measurements. The origin of survey sampling dates about 100 years ago. Early contributors to ‘the representative method’ were Kiaer, Bowley, and Tsuchprow. They were later joined by Neyman, Hansen, Hurwitz, Madow, and Cochran just to mention a few. Today the survey industry employs more than 130,000 people just in the USA. Surveys provide useful information to governments. Typically, parameters such as unemployment rates, consumer price indexes, amounts of toxic waste, gross national product are regularly estimated thereby allowing governments to make informed decisions. Modern survey sampling employs different kinds of inferential procedures. We discuss design-based, model-based and model-assisted inference and advocate the latter as a compromise between inference schools. There are a number of challenges associated with the development of the field. Among them are the need for designs that better take nonsampling errors into account, the estimation of mode effects, and better measures of international comparisons.