COMPARING MULTIPLE IMPUTATION AND PROPENSITY-SCORE WEIGHTING IN UNIT-NONRESPONSE ADJUSTMENTS.

The usual approach to unit-nonresponse bias detection and adjustment in social surveys has been post-stratification weights, or more recently, propensity-score weighting (PSW) based on auxiliary information. There exists a third approach, which is far less popular: using multiple imputed values for...

Descripción completa

Detalles Bibliográficos
Publicado en:Public Opinion Quarterly Vol. 79; no. 3; pp. 635 - 662
Autores principales: ALANYA, AHU, WOLF, CHRISTOF, SOTTO, CRISTINA
Formato: Artículo
Publicado: Oxford University Press / USA Fall2015
Materias:
Acceso en línea:Ver este registro en EBSCOhost
fields @attributes:
  recordID: 1
pdfLink:
plink: https://search.ebscohost.com/login.aspx?direct=true&db=ssf&AN=109369568&site=ehost-live
header:
  @attributes:
    shortDbName: ssf
    uiTerm: 109369568
    longDbName: Social Sciences Full Text (H.W. Wilson)
    uiTag: AN
  controlInfo:
    bkinfo:
    jinfo:
      jid:
        0033362X
        POQ
      jtl: Public Opinion Quarterly
      issn: 0033362X
      maglogo: N
    pubinfo:
      dt: Fall2015
      vid: 79
      iid: 3
      pid: 622
      pub: Oxford University Press / USA
    artinfo:
      ui:
        109369568
        10.1093/poq/nfv029
      ppf: 635
      ppct: 27
      formats:
      tig:
        atl: COMPARING MULTIPLE IMPUTATION AND PROPENSITY-SCORE WEIGHTING IN UNIT-NONRESPONSE ADJUSTMENTS.
      aug:
        au:
          ALANYA, AHU
          WOLF, CHRISTOF
          SOTTO, CRISTINA
        affil:
          Centre for Sociological Research, University of Leuven, Leuven, Belgium
          Scientific director, GESIS Leibniz-Institute for the Social Sciences, University of Mannheim, Mannheim, Germany
          Professor of sociology, University of Mannheim, Mannheim, Germany
          Professor, Interuniversity Institute for Biostatistics and Statistical Bioinformatics, Center for Statistics, University of Hasselt, Hasselt, Belgium
      su:
        Social science simulation methods
        Multiple imputation (Statistics)
        Social surveys -- Methodology
        Mathematical variables
        Nonresponse (Statistics)
        Research bias
        Charts, diagrams, etc.
      sug:
        subj:
          Social science simulation methods
          Multiple imputation (Statistics)
          Social surveys -- Methodology
          Mathematical variables
          Nonresponse (Statistics)
          Research bias
          Charts, diagrams, etc.
      ab: The usual approach to unit-nonresponse bias detection and adjustment in social surveys has been post-stratification weights, or more recently, propensity-score weighting (PSW) based on auxiliary information. There exists a third approach, which is far less popular: using multiple imputed values for each missing unit of the survey outcome(s). We suggest multiple imputation (MI) as an alternative to PSW since the latter is known to increase variance substantially without reducing bias when auxiliary variables are not associated with the survey outcome of interest. Given that most social surveys have multiple target variables, creating imputed data sets may address bias in survey outcomes with less variance inflation. We examine the performance of PSW and MI on mean estimates under various conditions using fully simulated data. To evaluate the performance of the methods, we report average bias, root mean squared error, and percent coverage of 95 percent confidence intervals. MI performs better under some of our scenarios, but PSW performs better under others. Even within certain scenarios, PSW performs better on coverage or root mean squared error while MI performs better on the other criteria. Therefore, robust methods that simultaneously model both the outcomes and the (non)response may be a promising alternative in the future.
      pubtype: Academic Journal
      doctype: Article
      src: R
    language: English
    refInfo:
    copyright:
      @attributes:
        flag: N
    holdings:
      @attributes:
        islocal: N