The Problem of Sampling
In most research, we save time by examining only a sample of an entire Gulliver whatever we are studying, whether it is tomato plants, laboratory animals, college freshmen, or working wives. If the sample is properly . selected, it will give an accurate picture of the entire universe under study. But to do this, the sample must be representative; that is, all kinds of people (or tomato plants, or whatever) must appear in the sample ill tile same proportions as they appear in the universe being studied, Thus, a .representative sample of the student body must contain the same proportion of freshmen, males, blacks, commuters, business majors, and married students as found in the entire student body. The most common, way of doing his is to select a random sample. ' The' term,' "random," suggests a selection without any system or design, such as choosing anyone who is handy-people passing a particular street corner, or climbing the library steps. But this would be an calculi(roped sample, for there are no controls to insure that it will be representative. A random sample is selected so that each person in the universe being studied has an equal chance of, being in tk sample.
We might take "every-tenth, or fiftieth, or hundredth name in the student directory. Or we might feed all the student numbers into the computer and program it to make a random selection. Every tenth' address on the community's mail-delivery routes, every twentieth hospital admission, or every hundredth driver's license would give random samples of local residents, hospital patients, or automobile drivers While a random sample is quite represent administrative random sample is still more perfectly representative. In such a sample, we first determine what percentage of each category of the universe under study would
be in the sample and then program the computer to select a random sample of each category. For example, suppose that our university student body is 32 percent freshmen, 49 percent male, 12 percent black, and 45 percent commuters (plus other categories). In a representative sample, each 100 members of the sample should include 32 freshmen, 4q males, 12 blacks, and 45 commuters. The computer is then programmed to make a random selection of 32 freshmen from all the freshmen, 49 males from all the males, and so on. A sample is composed of volunteers, such as persons who write letters to the editor or to their senator or who mail in magazine questionnaires. It is unknown how these volunteers compare with those who did not volunteer. Is it mainly the "far-outs" or the" squares" who mail in the questionnaires? Thus,
The Hit Report (Hite, 1976), a sex book pretending to be a research study of women's sex lives, Was based upon a 3 percent return of mailed questionnaires. Ms. Hite's sequel, The Hite Report on Ma/~ Sexuality" (Hite, 1981) did a little better.' getting a 6 percent return. With such any returns, these books should be viewed as popular entertainment, not research