This is the documentation for SuperSTAR 9.8

SuperSTAR 9.9 is now available.
View this page in the SuperSTAR 9.9 documentation or visit the SuperSTAR 9.9 documentation home.

Skip to end of metadata
Go to start of metadata

In order to reduce costs, statistical surveys typically assess a sample of the population rather than complete enumeration. Weighting is a technique used with sample surveys to provide estimated results for the total population from which the sample is drawn. Each individual sample is assigned a factor (its weight) to make the sample's effect on the total reflect its importance. SuperSTAR supports weighting functionality.

In most sample surveys a weight variable should be included in the data set to be analysed. This contains information about the sampling design in use and in some cases also about auxiliary information used for improving precision and/or adjusting for non-response. The principal purpose of weighting is to obtain as accurate parameter estimates as possible with the chosen sampling and estimation procedures. The basic form of weight (called sampling weight or design weight) is defined as the inverse of the inclusion probability of a selected element. However, the sampling weight often needs to be adjusted after sampling and data collection for possible sources of bias, such as biases caused by errors in sampling frames, non-response and measurement errors. Moreover, an estimation procedure aiming at efficiency improvement may involve further modification of the original weights. This derivation of adjusted or modified weights is called re-weighting. The weight variable to be used in estimation procedures is produced by completing the weighting procedure (including re-weighting when necessary).

  • No labels