Why should someone take a sample instead of studying the entire population? In many instances, a sample can provide reliable information more quickly and for less cost than a census—and the data can be of higher quality. But these advantages are realized only if the sampling is done according to statistical principles. This chapter sets out the features of a good sample and introduces the concepts of target population, sampling frame, selection bias, measurement error, and sampling error. Answers to a survey depend on the questions that are asked, and principles of questionnaire design are briefly discussed.