A new design and a weighting strategy of a survey on tourist activity of the Croatian population

Until 2014, the sample for the statistical survey Tourist Activity of the Population of the Republic of Croatia, shortly named as TADS, was a quota sample. The sample selection and the data collection for the survey in the Computer Assisted Telephone Interviewing (CATI) mode were conducted by a hired marketing research agency. In the summer of 2014, the Croatian Bureau of Statistics (CBS) set up its own call centre for CATI surveys. At the beginning of 2015, the CBS started developing a new probability sample design and a weighting strategy of TADS. The surveys of all the four quarters of 2015, fully conducted by the CBS, were treated as test surveys. The purpose of this paper is to describe and evaluate the finally accepted overlapping dual-frame survey design and a weighting procedure based on the sample data collected in all the quarters of 2015. The paper shows how the new design is incorporated in the existing circumstances in which surveys are conducted. The overlapping dual-frame survey design of TADS produces survey estimates related to the whole survey population based on two independent samples from two overlapping sampling frames (the first sampling frame consists of landline phone numbers from the phone book and the second one consists of mobile phone numbers from the phone book). In spite of a large undercoverage of the survey population in the phone book, using both phone types in TADS improves the total coverage of the survey population and relieves a problem of coverage and response bias. The applied weighting procedure is described systematically. The probability sampling allows calculating quality indicators such as standard errors of survey estimates and helps optimizing the survey process.