Skip to Main Content


In the process of defining a research question, the researcher must also decide who will be studied. The goal, of course, will be to make generalizations beyond the individuals studied to others with similar conditions or characteristics. Generalization is basic to all types of research, where scientists continually draw conclusions about human behavior and the environment based on limited experiences and measurements. The purpose of this chapter is to describe how the responses of a small representative group can be used with confidence to make predictions about the larger world.


The larger group to which research results are generalized is called the population. A population is a defined aggregate of persons, objects or events that meet a specified set of criteria. For instance, if we were interested in studying the effects of various treatments for osteoarthritis, the population of interest would be all people in the world who have osteoarthritis; however, it is not reasonable to test every person who has osteoarthritis. Working with smaller groups is generally more economical, more time efficient, and potentially more accurate than working with large groups because it affords better control of measurement. Therefore, through a process of sampling, a researcher chooses a subgroup of the population, called a sample. This sample serves as the reference group for estimating characteristics of or drawing conclusions about the population.

Populations are not necessarily restricted to human subjects. Researchers may be interested in studying characteristics of institutions or geographical areas, and these may be the units that define the population. In test-retest reliability studies, the population will consist of an infinite series of measurements. The sample would be the actual measurements taken. An epidemiological study may focus on blood samples. Industrial quality control studies use samples of items from the entire inventory of a particular manufacturing lot. Surveys often sample households from a population of housing units. A population can include people, places, organizations, objects, animals, days or any other unit of interest.

Sampling Bias

To make generalizations, the researcher must be able to assume that the responses of sample members will be representative of how the population members would respond in similar circumstances. Human populations are, by nature, heterogeneous, and the variations that exist in behavioral, psychological or physical attributes should also be present in a sample. Theoretically, a good sample reflects the relevant characteristics and variations of the population in the same proportions as they exist in the population.

Although there is no way to guarantee that a sample will be representative of a population, sampling procedures can minimize the degree of bias or error in choosing a sample. It is not so much the size of a sample that is of concern. A small representative sample of 50 may be preferable to an unrepresentative sample of 1,000. For example, in 1968 the Gallup ...

Pop-up div Successfully Displayed

This div only appears when the trigger link is hovered over. Otherwise it is hidden from view.