Design of Estimation - Some Practical Issues (Theme)


This module discusses a set of issues to consider in the design of the estimation of a business survey with emphasis on practical problems rather than on theory. Decisions have to be made together with other design decisions. The issues discussed mostly have neither obvious nor simple solutions. There is some theoretical ground to use, though, together with practical experience. The design decisions should be reasonable from these perspectives. The estimation procedure needs to work in practice also with a pressure on timeliness. It is important to save information about the procedure, including its weak points, in order to be able to improve successively.

There are different types of statistical inference depending on the combination of data and models used. The practical problems include, for instance, skewed distributions, outliers, coverage deficiencies, non-response, organisational changes, early estimates, small domains of estimation, and handling of administrative data.

The requests on quality include, for example, accuracy, coherence, and timeliness. Most business surveys are used both for primary statistics and as input to the National Accounts. In order to achieve coherence similar methods must be used in all business surveys.

Several types of statistical units are used in business surveys. Here the term ‘enterprise’ is used in most of the cases discussed, out of convenience.


To read the entire document, please access the pdf file (link under "Related Documents" on the right-hand-side of this page).


Your feedback is appreciated. Please send your remarks, suggestions for improvement, etc. to memobust@cbs.nl.