A short history of the regression estimator in survey sampling - ARCHIVED

Regression estimation has only been used extensively by statistical organizations in recent years. The sample mean and the classical ratio estimator are particular cases of this estimator. Both have a long history that can be traced back to the ancient Greeks or earlier. For instance, the ratio estimator was used by the Egyptians and Babylonians to compute the circumference of a circle given its radius, given a constant approximating the famous pi value. Multiple examples that use the ratio estimator as a means to indirectly compute a measure of interest can be found in physics, and engineering. The ratio estimator was also used to estimate population census estimates in the past when exact counts were beyond the capabilities of the existing administrations (for example John Graunt 1662, and Laplace late eighteenth century).

In this talk, we will trace the evolution of the regression estimation in survey sampling from the 1930's to the present time. We will outline its advantages and disadvantages in using it in survey sampling. Corresponding software development will also be presented