Nonparametric estimation of individual contact and response probabilities in cati surveys

Both producers and users of statistics are aware that nonresponse can impair the quality of the estimates. In this paper we face nonresponse adjusting the survey estimators by means of individual contact probabilities (ICP) and individual response (collaboration) probabilities (IRP). Obviously, both ICP and IRP are not known and must be estimated from available data. Referring to a real survey, to estimate these probabilities, we propose a nonparametric method based on data available partly from the CATI automated call scheduling, partly from the selection list. Some results are presented and compared with two alternative methods of estimation.