Sampling Theory

Communication research deals with questions such as “How are political issues framed?” or “What effect does violent television content have on its viewers?” These and many other questions apply to a particular population to which researchers want to generalize their results (e.g., all articles selected for a newspaper issue; all television viewers of a nation): the target population. In the first step of every empirical research design, this target population needs to be defined. Sometimes, the population will be small enough to be included entirely in the study. Yet, in many cases, the population is too large to be covered completely, due to time or financial constraints—or simply because it is virtually impossible to do so (e.g., when the target population is an entire nation). Then, a sample has to be drawn in such a way that the parameters of that sample accurately represent the entire target population. This “miniature” is under investigation in research, and from this, researchers extrapolate to the target population. Hence, an accurate sampling enables researchers to make statements about a population and to draw conclusions without having to investigate the whole population.

A sample consists of several elements (also called units). These elements could be objects (e.g., newspaper articles), events (e.g., press conferences), or individuals and groups of individuals (e.g., persons, households, organizations). Sampling is defined as the process of selecting those elements from a list (or quasi-list) of the members of a population, the so-called sampling frame. Although researchers aim at selecting a sample that matches the target population perfectly, there is no such thing as the perfect sample. The sampling procedure is always prone to errors, which affect its quality. Against a common misconception, it is first the quality of the sampling procedure and the resulting sample and not its size alone that affects its generalizability, the extent to which the findings can be applied to the target population.

This entry discusses the most common sampling procedures and respective potential biases and errors. The basic logic of the procedures described in this entry applies to all kinds of samples. Yet, if the elements are individuals (and not objects or events) the fact that not all units will respond becomes a specific issue. This enhances potential biases of the sample.