Target Population

The terms statistical population, population of interest, universe, and simply population are often used interchangeably when referring to a target population. What they have in common is that they define a group of people or other units that are the focus of a study and to whom the results are intended to generalize. The term population derives from the origins of statistics being used to describe human populations. However, a population may be people (σuch as the population of smokers), objects (σuch as hospital records), events (σuch as deaths or births), or measurements on or of the people, objects, or events (σuch as ages of smokers or occurrence of births).

A group may be defined as a population due either to an inherent characteristic of the group itself (σuch as residence in a particular city) or to a particular characteristic of interest to the researcher (σuch as having a particular health condition). A population may be very large (σuch as the population of the United States, estimated at more than 300 million in 2007) or very small (σuch as the population of [Page 1030]patients with progeria, of whom only 42 were known in the world as of 2006).

If a target population consists of people, objects, or events, then it is the set of sampling units about which investigators would like to draw conclusions or the set of all the members of the group under consideration. This population is the entire set of units to which findings of the survey are to be generalized. If the population consists of measurements taken on people, objects, or events, then, ideally, it is the set of all measurements that a researcher would like to have in answering a scientific question. These data are all possible or all hypothetically possible observations of those measurements.

In a statistical study, the researcher must define the population being studied. Typically, defining the target population is easy: It is all subjects possessing the common characteristics that are being studied. However, often for practical reasons the researcher must also define a study population, meaning the actual population of people from whom the sample will be drawn. For instance, if a researcher working in Ohio is interested in the effects of smoking on systolic blood pressure in adults aged 18 to 65 years, this would be the target population (σometimes called the ‘theoretical population’), while for practical reasons, the study population from whom the sample would be drawn might be all people between the ages of 18 and 65 years on January 1, 2007, residing within Greene County, Ohio. Additionally, each set of measurements that might be drawn on these individuals may be considered a population, so we could speak of the population of systolic blood pressure measurements or the population of smoking status indicators (whether each individual smokes) for adults aged 18 to 65 years in Greene County, Ohio.

It is important to note that whether a data set is considered a population or a sample depends on the context in which the data are to be viewed. In the previous example, if the researcher is interested in generalizing the study to all adults within southwestern Ohio, then the set of data for all adults in Greene County would be a sample. However, if the researcher is interested only in studying the relationship of smoking and blood pressure within Greene County, then that same set of data would constitute the population. Additionally, the population ‘adults aged 18 to 65 years in Greene County’ could be used to draw a sample for study: If the results are intended to generalize all adults in southwestern Ohio (or the entire United States), this would be a study population; if a sample was drawn with the intent of generalizing only the adults of Greene County, it would be the target population.

...