| Verfügbarkeit |
Autoren | Rohwer, Götz |
Titel | Uses of probabilistic models of unit nonresponse. |
URL | https://www.neps-data.de/Portals/0/Working Papers/WP_V.pdf |
Erscheinungsjahr | 2011 |
Seitenzahl | 18 S. |
Verlag | Bamberg: Otto-Friedrich-Universität |
Reihe | NEPS Working Paper. Band 5 |
Dokumenttyp | Monographie; Discussion Paper / Working Paper / Konferenzbeitrag; online |
Beigaben | Literaturangaben |
Sprache | englisch |
Forschungsschwerpunkt | Bildungspanel (NEPS) |
Schlagwörter | Bildungsforschung; Statistisches Modell; Statistische Methode; Modellierung; Variable; |
Abstract | Attempts to cope with unit nonresponse in surveys are often based on probabilistic models which posit ‘probabilities of response’, and then think of these probabilities as being dependent on a set of identifiable (and known) variables. This paper discusses how to understand and use such models. The discussion is based on a distinction between two purposes for which the sampled data can be used. (a) Descriptive estimation of statistical distributions which are defined for a particular target population from which the sample is drawn. (b) Estimation of functional models (models formalizing probabilistic rules) which concern the behavior of a generic unit conditional on known values of some variables. The paper also discusses how probabilistic models of unit nonresponse can be used for the construction of nonresponse weights. It is shown that there are no general answers to the question whether such weights lead to a reduction of nonresponse bias. The paper ends with a suggestion for including information about unit nonresponse into data files for scientific use. (DIPF/Orig.) |
Förderkennzeichen | 01GJ0888 |