An Approximate Design Effect for Unequal Weighting When Measurements May Correlate with Selection Probabilities

Bruce D Spencer

An Approximate Design Effect for Unequal Weighting When Measurements May Correlate with Selection Probabilities

Statistics and Data Science

Research output: Contribution to journal › Article › peer-review

Abstract

It is common practice to estimate the design effect due to weighting by 1 plus the relative variance of the weights in the sample. This formula has been justified when the selection probabilities are uncorrelated with the variable of interest. An approximation to the design effect is provided to accommodate the situation in which correlation is present.

Original language	English
Pages (from-to)	137-138
Journal	Survey Methodology
Volume	26
State	Published - 2000

Cite this

@article{ae50b07850fa45c5be236025a9d1e01e,

title = "An Approximate Design Effect for Unequal Weighting When Measurements May Correlate with Selection Probabilities",

abstract = "It is common practice to estimate the design effect due to weighting by 1 plus the relative variance of the weights in the sample. This formula has been justified when the selection probabilities are uncorrelated with the variable of interest. An approximation to the design effect is provided to accommodate the situation in which correlation is present.",

author = "Spencer, {Bruce D}",

year = "2000",

language = "English",

volume = "26",

pages = "137--138",

journal = "Survey Methodology",

issn = "0714-0045",

publisher = "Statistics Canada",

}

TY - JOUR

T1 - An Approximate Design Effect for Unequal Weighting When Measurements May Correlate with Selection Probabilities

AU - Spencer, Bruce D

PY - 2000

Y1 - 2000

N2 - It is common practice to estimate the design effect due to weighting by 1 plus the relative variance of the weights in the sample. This formula has been justified when the selection probabilities are uncorrelated with the variable of interest. An approximation to the design effect is provided to accommodate the situation in which correlation is present.

AB - It is common practice to estimate the design effect due to weighting by 1 plus the relative variance of the weights in the sample. This formula has been justified when the selection probabilities are uncorrelated with the variable of interest. An approximation to the design effect is provided to accommodate the situation in which correlation is present.

M3 - Article

SN - 0714-0045

VL - 26

SP - 137

EP - 138

JO - Survey Methodology

JF - Survey Methodology

ER -

An Approximate Design Effect for Unequal Weighting When Measurements May Correlate with Selection Probabilities

Abstract

Fingerprint

Cite this