PropertyValue
?:abstract
  • When the European Values Study was carried out in Iceland in 2017, it was decided to conduct a Computer Assisted Web Interviewing (CAWI) for the first time along with the traditional Computer Assisted Personal Interview (CAPI). In order to shorten the time it takes to answer the survey a matrix design was used. The questionnaire was split into 5 parts and the sample was randomly split so that each participant only answered 3 of these parts. A large amount of blank cells in the data is inevitably a consequence of this method, and when making statistical tests, these missing values must be dealt with in some way. Listwise-deletion, where all rows containing missing values are deleted, is the most common method, but since the data contains a high proportion of missing values due to design, more information is being sacrificed than needs be. In this study seven methods to deal with missing by design values will be compared. These methods are listwisedeletion, mean imputation, multiple linear regression imputation, multiple linear regression imputation with noise, as well as three methods using multiple imputation with and without auxiliary variables, namely Predictive Mean Matching (PMM), Classification and Regression Trees (CART) and Random Forest. The results show that multiple imputation with PMM and CART yielded a regression model that most resembled regression model with full data. (xsd:string)
?:author
?:comment
  • (EVS) (xsd:string)
?:dataSource
  • EVS-Bibliography (xsd:string)
?:dateModified
  • 2019 (xsd:gyear)
?:datePublished
  • 2019 (xsd:gyear)
?:duplicate
?:fromPage
  • 53 (xsd:string)
is ?:hasPart of
?:inLanguage
  • other (xsd:string)
is ?:mainEntity of
?:name
  • Tilreikningur og brottfallsgildi : Samanburður á aðferðum til að bregðast við hönnuðum eyðum í gagnasafni (Isländisch) (xsd:string)
?:publicationType
  • mastersthesis (xsd:string)
?:reference
?:sourceInfo
  • 53, 2019 (xsd:string)
  • Bibsonomy (xsd:string)
?:studyGroup
  • European Values Study (EVS) (xsd:string)
?:tags
  • 2019 (xsd:string)
  • EVS (xsd:string)
  • EVS2017 (xsd:string)
  • EVS_input2019 (xsd:string)
  • EVS_pro (xsd:string)
  • FDZ_IUP (xsd:string)
  • checked (xsd:string)
  • kbe (xsd:string)
  • mastersthesis (xsd:string)
  • other (xsd:string)
  • wa (xsd:string)
?:toPage
  • 53 (xsd:string)
rdf:type
?:url