UM  > 社會科學學院  > 社會學系
Investigation of Ways to Handle Sampling Weights for Multilevel Model Analyses
Cai,Tianji
2016-10-28
Source PublicationSociological Methodology
ISSN14679531 00811750
Volume43Issue:1Pages:178-219
Abstract

When analysts estimate a multilevel model using survey data, they often use weighted procedures using multilevel sampling weights to correct the effect of unequal probabilities of selection. This study addresses the impacts of including sampling weights and the consequences of ignoring them by assessing the performance of four approaches: the multilevel pseudo–maximum likelihood (MPML), the probability-weighted iterative generalized least squares (PWIGLS), the naive (ignoring sampling weights), and the sample distribution methods for a linear random-intercept model under a two-stage clustering sampling design. When inclusion probabilities are correlated with the values of outcome variable conditioning on the model covariates, the sampling design becomes informative. The results show that whether a sampling design is informative and at which stage of the sampling design it is informative have substantial impacts on the estimation. The results also show that the level of variation of sampling weights is correlated with the bias of estimates. A higher level of variation of sampling weights is associated with a higher level of bias when a sampling design is informative; however, under a noninformative design, the level of variation of sampling weights may not necessarily associate with biased results. Ignoring an informative sampling design at the first stage will result in biased estimates on the intercept and variance of random effect, whereas ignoring an informative sampling design at the second stage will lead to slightly underestimated fixed effects and residual variance, in addition to the biased estimates on the intercept and variance of random effect. Including the sampling weights as the hybrid methods (MPML and PWIGLS) may still produce biased estimates on the intercept and variance of random effect and slightly underestimated fixed effects and residual variance. The sample distribution method may give unbiased estimates, but it depends on the correct specification of the sampling process. © 2013, American Sociological Association. All rights reserved.

KeywordMultilevel Model Sample Distribution Method Sampling Weights
DOI10.1177/0081175012460221
URLView the original
Indexed BySSCI
Language英语
WOS Research AreaMathematical Methods In Social Sciences ; Sociology
WOS SubjectSocial Sciences, Mathematical Methods ; Sociology
WOS IDWOS:000339954900015
Fulltext Access
Citation statistics
Cited Times [WOS]:12   [WOS Record]     [Related Records in WOS]
Document TypeJournal article
CollectionDEPARTMENT OF SOCIOLOGY
Corresponding AuthorCai,Tianji
AffiliationUniversity of MacauDepartment of Sociology,United Kingdom
First Author AffilicationUniversity of Macau
Corresponding Author AffilicationUniversity of Macau
Recommended Citation
GB/T 7714
Cai,Tianji. Investigation of Ways to Handle Sampling Weights for Multilevel Model Analyses[J]. Sociological Methodology,2016,43(1):178-219.
APA Cai,Tianji.(2016).Investigation of Ways to Handle Sampling Weights for Multilevel Model Analyses.Sociological Methodology,43(1),178-219.
MLA Cai,Tianji."Investigation of Ways to Handle Sampling Weights for Multilevel Model Analyses".Sociological Methodology 43.1(2016):178-219.
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[Cai,Tianji]'s Articles
Baidu academic
Similar articles in Baidu academic
[Cai,Tianji]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[Cai,Tianji]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

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