As the Internet and various communication devices have spread to the general public after the mid-1990s, the ratio of non-probability sampling surveys using online and wireless communication has increased rapidly. Not depending on the randomly sampled individual samples in a reasonable and complete sample sampling framework, most of these surveys are performed by individuals who have been qualified by participating in a panel or clicking on an online advertisement. At first, it was mainly used in market research, but as the cost of surveys based on the probability sampling sample theory increases and the response speed decreases, non-probability sampling surveys are becoming more common. Although mainly used in the production of national statistics, probability sampling sample surveys are facing a crisis due to reductions in the scope of inclusion, decrease in response rates, and increases in data collection costs. In addition, the rapidly increasing statistical demand and the emergence of big data are accelerating the need for changing the sample survey. It seems difficult to overcome the problems of the sample survey simply by investing more time and cost to make the sampling framework more elaborate, and to increase the response rate. In this study, we are to examine domestic and overseas cases of sample survey design based on the theory of non-probability sampling that can be reflected in the survey design of the NSO, and to identify the latest trends discussed at the Statistical Methodology Society and conferences. Furthermore, we are to explore a test and survey methodology that can be reflected or utilized in the actual survey design.


Key words: Sample survey, probability sampling, non-probability sampling, error, selectivity bias, overseas cases