Infoskripsi arrow Reference arrow Defining The Population
Defining The Population





Sampling means selecting a given number of people from a defined population, as representative of that population. One type of population distinguished by educational researchers is called the target population. By target population, also called universe, we mean all the members of a real or hypothetical set of people, events, or objects to which we wish to generalize the results of our research. The advantage of drawing a small sample from a large target population is that it saves the researcher the time and expense of studying the entire population. If the sampling is done properly, the researcher can reach conclusions about an entire target population that are likely to be correct within a small margin of error by studying a relatively small sample. The first step in sampling is to define the target population. Typical populations from which educational research samples might be drawn include school superintendents in the state of Utah; practice-teaching supervisors in state-supported teachers' colleges, bilingual children in the primary grades of the San Antonio city school district, pupils failing algebra in New York City schools, and seniors graduating from American public high schools in June of 1978. These examples illustrate that the target population may represent a large group scattered over a wide geographical area or a small group concentrated in a single area. It is seldom possible for a researcher to draw a representative sample from a target population such as all first-grade pupils in public schools in the United States. In order to obtain a representative sample of this broadly defined population, a complex method of selecting cases from different areas, different-sized communities, and different types of schools would have to be developed. Obviously the selection of such a sample and collection of data from it would involve a tremendous amount of work and expense. Instead, the researcher must usually draw his sample from an experimentally accessible population such as all first-grade pupils in the San Francisco school district.3 Even though the sample is selected from the accessible population, the researcher may want to know the degree to which the results can be generalized to the target population. This type of generalization requires two inferential leaps. First, the researcher must generalize the results from the sample she actually studied to the accessible population from which she selected the sample. Second, she must generalize from the accessible population to the target population. The leap from sample to accessible population presents no problem if a random sample of the accessible population was obtained, that is, a sample in which all members of the population had an equal chance of being selected. If the sample was not formed random, the researcher must gather data about the sample and the population on characteristics critical to the study. Often such data are available in school records, but some testing may also be necessary. It will rarely be possible to obtain all data that would be useful, but the researcher should obtain comparative information on as many critical variables as possible with the resources she has at her disposal. These data will demonstrate that the sample is either biased or unbiased. If unbiased the researcher can safely generalize the results to the accessible population. If, however, the sample is biased, she must report the nature of the bias and discuss how this bias is likely to affect the results. In order to make the second leap from the accessible population to the target population, the researcher must gather data to determine the degree of similarity between these two populations. It is possible to gather comparative data on a very large number of variables. However, if the investigator can demonstrate that the accessible population is closely comparable to the target population on a few variables that appear most relevant to the study, she has done much to establish population validity. That is, she has established that the accessible population is reasonably representative of the target population. For example, suppose a researcher wants to compare the achievement of first-grade pupils who are taught with two different reading programs. If she selects a random sample from the accessible population of first graders in the San Francisco schools but can demonstrate by comparing local test data with national test norms that San Francisco first graders are not significantly different from first graders nationwide on such important variables as reading readiness, verbal IQ, chronological age, and socioeconomic status, then she has established population validity. This means that she can generalize her results from the accessible to the target population with reasonable confidence. Often financial limitations or the nature of the research problem limit us to sampling the student population of a single school district. Studies based on this narrow accessible population are of course, less generalizable than those based on broader populations, but may still have important implications for other educators if it can be demonstrated that this population is reasonably similar to the target population on a few critical variables. It is beyond, the scope of most research projects to identify all the members of a defined population. For example, the identification by name of all fifth-grade teachers in even a single state would be major undertaking. Thus, researchers usually rely on published lists of various populations that arc of interest to educators. Most researchers will be able to draw samples from accessible populations only at the state or district level. In these cases the state or district education office should be contacted to find out if they have a list defining the population in which the researcher is interested. State and district education departments usually maintain complete list of information about schools, such as school addresses, grades, enrollment, and names of principals and teachers for administrative purposes. In certain instances the researcher may wish to consult a national association or national directory, even though the defined population is at the state or district level. For example, suppose one wished to survey a sample of educational researchers in a given state. One might obtain a national directory, such as the current membership directory of the American Educational Research Association, and mark all the educational researchers residing in that state. From this defined population, a sample of educational researchers can be selected. The Guide to American Educational Directories4 is a helpful source for locating directory lists that can be used to define populations. In using any published list to define a population, the student should check to determine whether it is complete and up to date. School enrollment and memberships of organizations are constantly changing, so frequent updating of population lists is necessary. Also it should be realized that membership in most organizations is voluntary. Thus a student who uses an organization directory to define a population faces the risk of selecting a biased sample, since joiners of organizations may differ in important respects from nonjoiners. Should this be the case, the student should probably define the accessible population as all members of a given organization rather than as all members of the profession or group which the organization serves.