Keywords: survey representativeness, randomness, polling organizations

Despite the increasing prevalence of public opinion surveys, there is little evidence that public understanding of survey representativeness has improved significantly. Using data from the Korean General Social Survey (KGSS), this study examines how people perceive the representativeness of survey samples and identifies key factors that shape these perceptions. First, we investigate whether the phrasing of the question—using "a random sample of 1,000" versus "a sample of 1,000"—affects respondents' answers to the question: "Do you think that a sample of 1,000 people surveyed nationwide can accurately reflect the public opinion of the entire population, or do you think it cannot?" The results indicate that approximately 33% of respondents believed a sample of 1,000 could accurately reflect national public opinion, regardless of whether the term random was included.
Next, using logistic regression analysis, we find that trust in media-reported statistics, reliance on polling results when voting, confidence in the neutrality of polling organizations, interest in participating in election polls, and conservative political orientation are positively associated with the belief that a sample of 1,000 can accurately represent national public opinion. In contrast, age, gender, education level, income, homeownership, health status, occupation, and urban or rural residence show no significant relationship with perceptions of sampling representativeness.
Our findings suggest that the general public does not place much importance on the concept of randomness when evaluating the representativeness of a sample. Moreover, public confidence in polling is not merely a methodological concern but is also shaped by institutional trust and political beliefs.