Skip to main navigationSkip to main content
The University of Southampton
Southampton Statistical Sciences Research Institute

Survey estimation

The target of survey estimation is characteristics (often referred to as parameters) of a finite population on the basis of a sample of units taken form this population. Common examples of finite-population parameters include totals, means and quantiles of one or several measurements of interest that are associated with each unit of the population.

The core concern in survey estimation is the so-called sampling error. Relevant topics are:

  • Sampling design, which may be a randomisation scheme involving stratification, unequal inclusive probabilities, multiple stages of selection and/or multiple sampling frames, or a purposive selection of units.
  • Sample coordination of multiple surveys, cross-sectional and/or over time. 
  • Use of auxiliary values, which may be known throughout the population or only observed in the sample.
  • Robust methods in the presence of outliers.
  • Assessment of uncertainty, which may be exclusively based on the probability sampling design. Or the sampling design may be ignored and the evaluation is entirely model-based in order for the inference to be appropriately conditional.

In addition, survey estimation must also deal with appropriate adjustments of both the estimator and its associated uncertainty due to non-sampling errors such as survey nonresponse and measurement errors.

Share this research area Share this on Facebook Share this on Twitter Share this on Weibo
Privacy Settings