Population of Interest

Most scientific research has some specifie groups of interest and attempts to make generalizations about the characteristics of those groups. This is what is termed the population of interest. For example, a public health study assesses medical needs among senior citizens; an educational study examines the relationship between high school students' academic performance and their parents' academic attainment; and a marine biology project attempts to investigate the life cycle of humpback whales. The population of interest in the first study is the senior citizens; the second high school students; and the third humpback whales. The same applies to applied social science studies that employ surveys.

While closely related to one another, the population of interest is more loosely defined than the population of inference and the target population. In fact, the definition of the population of interest is too loose to be directly implemented in survey data collection. In the case of the senior citizen study, who constitutes the senior citizens? Is there any age criterion? What is the reference time period for the survey? The age criterion applied to the scope of senior citizens yields different sets of people depending on the selected ages and time period chosen. Are all U.S. senior citizens eligible for the survey? What about U.S. senior citizens living abroad at the time of the survey? What about those living in nursing homes? These kinds of questions must be answered in order to make the survey operationaliz-able. The answers to these questions narrow the population of interest to the population of inference and then narrow it further to the target population.

Strictly speaking, survey results cannot be generalized to the population of interest unless it is perfectly aligned with the target population. This, however, is not necessarily practiced in reality. For example, the target population of the General Social Survey (GSS) in the United States is defined as “all non-institutionalized English-speaking persons 18 years of age or older living in the U.S.” Most studies based on the GSS data use the results to explain behaviors and attitudes of “American adults,” which is different from the target population. This type of misalignment is common, reflecting the gap between the ideal and practical study settings.

Surveys are conducted using fixed resources, so often the budget will not allow for a survey design that reaches every person in the population of interest. It is costly to develop questionnaires in languages other than English and hire and train interviewers for non-English interviews, and it may not be the most effective way to allocate resources given that the proportion of non-English speakers in the general population is not large. Also, even in the unlikely case where there were essentially “unlimited” resources, one may not have access to all of the population elements. For example, it may be impractical, if not impossible, to try to reach prison and jail inmates for general population surveys, or U.S. residents abroad. Errors coming from these factors, however, often do not hamper the generalizability of survey findings when the noncovered portion of the population is relatively small.

...