Sampling from Populations: Sample Surveys

Abstract

Much empirical data arises from experiments, in which the investigator interacts in some way with the units of observation and actually influences the conditions of the units leading to the measurements. Many other sets of data result from simply observing, that is, making a survey. It is to such investigations that we now turn our attention. Usually one cannot observe every individual in the population, and often this would not even be desirable, for many individuals are similar. One does not need to eat the whole bowl to learn how the soup tastes; a spoonful will suffice, provided that the soup has been adequately stirred. The “spoonful” is a sample from the bowl (population), and “stirring” corresponds to drawing a random sample.