Title: Estimation techniques for discrete response variables in dual frame surveys
Authors: Antonio Arcos - Universidad de Granada (Spain)
Maria Giovanna Ranalli - University of Perugia (Italy)
David Molina - University of Granada (Spain) [presenting]
Abstract: In a dual frame survey, two different samples are drawn; one from each of the sampling frames that are available, and then the information collected is adequately combined to get estimates. Although each of these frames may be incomplete considered separately, it is assumed that, jointly, they cover the entire target population. Since the emergence of dual frame surveys a noticeable number of estimators have been formulated to estimate the population total or mean of a continuous variable and, currently, dual frame surveys are widely used by statistical agencies and private companies due to their proven benefits. Surveys in general, and dual frame surveys in particular, usually include questions in which the respondents have to select one in a series of options. Although customary dual frame estimators can be used, the estimates they provide are inconsistent since they do not add up to 1 through all the possible categories of the response. New techniques for appropriately addressing variables with discrete outcomes in dual frame surveys are presented. To check the efficiency of the proposed procedures some Monte Carlo experiments were carried out.