Characteristics and Implications of Sample Attrition in the Korea Health Panel Survey

Abstract

Panel surveys are prone to undergoing respondent dropout as the waves progress over time. Panel attrition in effect reduces the representativeness of the sample and can lead to inferential bias and erroneous conclusions, thereby rendering the credibility of the surveys compromised. Therefore, it is important to identify the causes of attrition and hold them in check in advance on an ongoing basis. In this article, I examined data from the 2nd Korea Health Panel Survey and analyzed the characteristics and response types of participating household members. I then statistically explored the effect of each different response-type group from various angles. Based on the findings, I suggest four areas for focused panel management to minimize sample attrition.