Bitte verwenden Sie diesen Link, um diese Publikation zu zitieren, oder auf sie als Internetquelle zu verweisen: https://hdl.handle.net/10419/29618 
Erscheinungsjahr: 
2002
Schriftenreihe/Nr.: 
Diskussionspapier No. 42b/2002
Verlag: 
Friedrich-Alexander-Universität Erlangen-Nürnburg, Lehrstuhl für Statistik und Ökonometrie, Nürnberg
Zusammenfassung: 
On the basis of real data sets it is shown that splitting a questionnaire survey according to technical rather than qualitative criteria can reduce costs and respondent burden remarkably. Household interview surveys about media and consuming behavior are analyzed and splitted into components. Following the matrix sampling approach, respondents are asked only the varying subsets of the components inducing missing data by design. These missing data are imputed afterwards to create a complete data set. In an iterative algorithm every variable with missing values is regressed on all other variables which either are originally complete or contain actual imputations. The imputation procedure itself is based on the socalled predictive mean matching. In this contribution the validity of split and imputation is discussed based on the preservation of empirical distributions, bivariate associations, conditional associations and on regression inference. Finally, we find that many empirical distributions of the complete data are reproduced well in the imputed data sets. Concerning these long media and consumer questionnaires we like to conclude that nearly the same inference can be achieved by means of such a split design with reduced costs and minor respondent burden
Dokumentart: 
Working Paper

Datei(en):
Datei
Größe
91.87 kB





Publikationen in EconStor sind urheberrechtlich geschützt.