PropertyValue
?:abstract
  • "The aim of survey statistics is to depict the public opinion by using representative samples of the whole population. Therefore, in this field of research it is focused on collecting data with a minimum of errors, which is an important prerequisite for the subsequent steps of data analysis. However, there are many sources of errors having different influence on parameter estimation. Those impacts can be quantified with the help of the Mean Squared Error (MSE), comprising both, systematic and random error. Viewing the different error sources and their impacts leads to the concept of the Total Survey Error (TSE). It contains the different components sampling error, specification error, coverage error, nonresponse error, measurement error, processing error and includes further constraints and theories. One aim of this thesis is to determine the impacts of the different error sources by simulating a realistic data set which orientates on the structure of the Allbus 2014 and applying constructed error models to this data set. This way, direction and magnitude of the different error sources can be evaluated. It is shown that nonresponse error is a major error component of the TSE. Hence, in the second part of this thesis, it is concentrated on different approaches for handling missing data as consequence of nonresponse error. Besides common weighting and imputation methods, likelihood-based approaches, either ignoring or explicitly modeling the missing process, will be introduced, applied and discussed critically especially with respect to their assumptions. The application of the correction methods again is based on simulated data sets including missing values following the different missing mechanisms Missing Completely at Random (MCAR), Missing At Random (MAR) and Missing Not At Random (MNAR). It results that the performances of missing data methods strongly depend on these underlying processes. While in the case of MCAR data simple ad-hoc procedures should be preferred, for MAR or MNAR data more advanced methods are required". (xsd:string)
?:author
?:comment
  • (ALLBUS) (xsd:string)
?:dataSource
  • ALLBUS-Bibliography (xsd:string)
?:dateCreated
  • Aufgenommen: 32. Fassung, März 2018 (xsd:gyear)
?:dateModified
  • 2016 (xsd:gyear)
?:datePublished
  • 2016 (xsd:gyear)
?:duplicate
is ?:hasPart of
?:inLanguage
  • english (xsd:string)
is ?:mainEntity of
?:name
  • Impacts of the Total Survey Error with a Focus on the Nonresponse Error and its Handling (xsd:string)
?:publicationType
  • mastersthesis (xsd:string)
?:reference
?:sourceInfo
  • Bibsonomy (xsd:string)
?:studyGroup
  • ALLBUS (xsd:string)
?:tags
  • 2016 (xsd:string)
  • ALLBUS (xsd:string)
  • ALLBUS2014 (xsd:string)
  • ALLBUS_input2017 (xsd:string)
  • ALLBUS_pro (xsd:string)
  • ALLBUS_version32 (xsd:string)
  • FDZ_ALLBUS (xsd:string)
  • checked (xsd:string)
  • english (xsd:string)
  • mastersthesis (xsd:string)
rdf:type
?:uploadDate
  • 27.11.2017 (xsd:gyear)
?:url