Please use this identifier to cite or link to this item: http://hdl.handle.net/10419/20328
Full metadata record
DC FieldValueLanguage
dc.contributor.authorBollinger, Christopher R.en_US
dc.contributor.authorChandra, Amitabhen_US
dc.date.accessioned2009-01-28T16:13:18Z-
dc.date.available2009-01-28T16:13:18Z-
dc.date.issued2004en_US
dc.identifier.urihttp://hdl.handle.net/10419/20328-
dc.description.abstractIn empirical research it is common practice to use sensible rules of thumb for cleaning data.Measurement error is often the justification for removing (trimming) or recoding (winsorizing)observations whose values lie outside a specified range. We consider a generalmeasurement error process that nests many plausible models. Analytic results demonstratethat winsorizing and trimming are only solutions for a narrow class of measurement errorprocesses. Indeed, for the measurement error processes found in most social-science data,such procedures can induce or exacerbate bias, and even inflate the variance estimates. Weterm this source of bias ?Iatrogenic? (or econometrician induced) error. Monte Carlosimulations and empirical results from the Census PUMS data and 2001 CPS datademonstrate the fragility of trimming and winsorizing as solutions to measurement error in thedependent variable. Even on asymptotic variance and RMSE criteria, we are unable to findgeneralizable justifications for commonly used cleaning procedures.en_US
dc.language.isoengen_US
dc.publisher|aInstitute for the Study of Labor (IZA) |cBonnen_US
dc.relation.ispartofseries|aIZA Discussion paper series |x1093en_US
dc.subject.jelJ1en_US
dc.subject.jelC1en_US
dc.subject.ddc330en_US
dc.subject.keywordmeasurement error modelsen_US
dc.subject.keywordtrimmingen_US
dc.subject.keywordwinsorizingen_US
dc.subject.stwModell-Spezifikationen_US
dc.subject.stwStatistischer Fehleren_US
dc.subject.stwBiasen_US
dc.subject.stwTheorieen_US
dc.titleIatrogenic Specification Error : A Cautionary Tale of Cleaning Dataen_US
dc.typeWorking Paperen_US
dc.identifier.ppn384725759en_US
dc.rightshttp://www.econstor.eu/dspace/Nutzungsbedingungen-

Files in This Item:
File
Size
361.78 kB





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