Please use this identifier to cite or link to this item: http://hdl.handle.net/10419/174460
Authors: 
Padilla Terán, Alberto Manuel
Year of Publication: 
2017
Series/Report no.: 
Working Papers 2017-07
Abstract: 
In probability sampling, variance estimation of an estimated mean or total requires developing a mathematical expression that depends on the design used to extract a sample. These formulae can be difficult to build and sometimes involve computation of joint inclusion probabilities of selection, which can be hard to obtain. For some sampling designs it is not possible to obtain an unbiased estimator of the variance. These designs include the selection of one element or one large primary sampling unit within some strata, or systematic selection of units or primary sampling units within strata. The problem of variance estimation may also arise from an analytical perspective, while estimating means or totals in unplanned domains, it is possible to arrive at only one unit or cluster within some strata. In this article, we propose a linear regression variance estimator which is very simple to compute and gives a solution to the aforementioned problems. Some examples using different designs are given.
Subjects: 
Linear regression
variance estimation
expansion factor
unplanned domains
collapsed strata
JEL: 
C80
C83
Document Type: 
Working Paper
Appears in Collections:

Files in This Item:
File
Size
726.48 kB





Items in EconStor are protected by copyright, with all rights reserved, unless otherwise indicated.