Online Program

Generalized calibration to deal with non-ignorable nonresponse in the German job vacancy survey
*Hans Kiesl, Regensburg University of Applied Sciences 

Keywords: job vacancy survey, generalized calibration, nonresponse adjustment

The German job vacancy survey is a quarterly business survey (stratified by sector and size class). Calibration is used to deal with high nonresponse rates; the most important calibration variable being the number of registered job vacancies (business units may inform the Federal Employment Agency of their vacancies; these are called registered vacancies). It turns out that design weighting (ignoring nonresponse) results in biased estimates of the total number of registered vacancies. Thus, nonresponse seems to be correlated with the number of registered vacancies. Since we assume that nonresponse is correlated with the number of any vacancies (registered or not), nonresponse in our survey is non-ignorable. Up to now, the strategy for dealing with this has been to calibrate to the number of registered vacancies under the additional constraint that within each stratum sampling units with vacancies get identical calibrated weights. In this paper, we compare this strategy in terms of bias and variance with two other approaches by means of a simulation study: a two-stage GREG estimator and the generalized calibration estimator proposed by Deville (2002) and Chang and Kott (2008).