Journal of Official Statistics, Vol.3, No.4, 1987. pp. 359–373
Some Markov-Chain Models for Nonresponse in Estimating Gross Labor Force Flows
Elizabeth A. Stasny
Abstract:The longitudinal data bases available from panel surveys may be useful for estimating period-to-period gross flows among survey classifications. One problem in gross flow estimation is how to handle the nonresponse in the data. Typically, we do not believe that such nonresponse occurs at random with respect to the variable of interest. In this paper, I consider a model-based approach to the problem of handling nonresponse in data from a panel survey. I use a Markov-chain models for the underlying gross flow process and propose some Markov-chain models for the process that generates the possibly nonrandom nonresponse. The interacting Markov-chain models are fit using maximum likelihood estimation to data from the Canadian Labour Force Survey.
Keywords:Longitudinal data; maximum likelihood estimation; nonrandom nonresponse; panel survey.
Copyright © Statistics Sweden 1996-2018. Open AccessISSN 0282-423XCreated and Maintained by OKS Group