Metodi per il trattamento delle diverse componenti della mancata risposta totale applicati all’indagine Istat sulla Disabilità

The negative effects of non-response on the estimates of a sample survey must be properly treated. The methods, generally used for this purpose, are based on the use of auxiliary information known both for respondent and not respondent units, without taking into account the causes of the non-response. In this paper, methods, which consider this aspect, have been tested for the treatment of non-response in the Istat survey on Disabilities. The methods applied are the sequential weight adjustment method and the method based on a sample selection model with multiple selection equations; the first was adopted to correct the sample weights of the respondent units to the survey, the second to verify the assumptions which underlie of the first method and to analyze the impact of the bias effects produced by different causes of non-response on some survey estimates.