Journal of Official Statistics, Vol.4, No.4, 1988. pp. 363–374
Comparison of Measurement Errors for Telephone Interviewing and Home Visits by Misclassification Models
Gad Nathan and Tami Eliav
Abstract:The assessment of measurement errors due to the mode of data collection (home visit or telephone interview) can be carried out on the basis of repeated observations, for instance via panel surveys, for characteristics which do not change over time. A misclassification model is proposed for this case with probabilities of correct classification and of misclassification depending on the mode of collection. The estimation and comparison of these parameters is used to assess the effects of mode of collection.
The models proposed are applied to empirical data from the Israel Labour Force Survey, in which home visits and telephone interviews are used in different rounds of the survey for the same units. The effects of mode of collection on model parameters are assessed for educational characteristics which are not subject to change over time (for those who no longer study). This is done separately for the subpopulation of households who have telephones and are willing to respond by telephone and for the subpopulation of households who respond only by home visit. The results indicate that responses obtained by telephone interview have higher consistency than those obtained by home visits.
Keywords:Misclassification; response errors; mode of collection.
Copyright © Statistics Sweden 1996-2018. Open AccessISSN 0282-423XCreated and Maintained by OKS Group