The margin of error when using a random sample of 2000 people out of a population of 260M is less than 2.19% with a 95% confidence level. You are probably maki

The margin of error when using a random sample of 2000 people out of a population of 260M is less than 2.19% with a 95% confidence level. You are probably making the common mistake of thinking that the larger the population the bigger a sample needs to be to accurately poll. That is true for lower populations but the "sample needed vs. population size" curve is asymptotic. For example, suppose we want to determine what percent of people believe X, with a 5% margin of error and a 95% confidence level. Here's a table of the random sample size needed for various populations: Sample Population 80 100 278 1 000 370 10 000 383 100 000 385 1 000 000 385 10 000 000 ... ... 385 ∞