In choosing a sample, two decisions have to be made: sample size and method of selection. In most cases, it is impossible to survey everyone in the target population. In our academic performance example, you cannot survey every student in your school. Therefore, it is necessary to select a representative portion of the target population. The goal of a sample is to select a group of subjects whose responses would be representative of the population as a whole.
The size of a sample is more important than the percentage of the total population sampled. (This assumes, of course, that the sample was correctly selected.) Accurate samples of the total United States population can consist of as few as 2000 people. A sample of this size represents only about .00001 of the total population, but when correctly selected, can give a highly accurate representation of the entire population. If the population itself is much smaller, a survey must be based on a sample of a much larger proportion of the population in order to keep the actual size large enough so that it can be analyzed properly. In those cases where the target population is less than 50, the survey should include all members of the population. The term sample size refers to the actual number of responses you receive to your questions, not the number of questionnaires you distribute
Sampling can be done either through random or non-random methods. In survey research, the word random is used in a very specific, technical sense. A random sampling procedure is one in which all subjects have an equal chance of being selected. In this case, random does not mean haphazard or arbitrary as it frequently does in ordinary conversation. In non-random sampling, subjects are not selected by chance. Non-random sampling is also called haphazard sampling. Examples include contacting shoppers at a shopping center, calling up the first 100 people on a phone list, or posting a survey on the Internet. Non-random sampling does not allow the survey designer to generalize beyond the people being interviewed. Although random sampling is preferred in every case, time and respondent availability sometimes make it very difficult.
Two of the most frequently used random sampling methods are simple random sampling and cluster sampling. If you have a complete list of everyone in a target population and have equal access to those persons, you can use a simple random sampling procedure. If you don't have a complete list of everyone in your target population, you can identify a location where your target population may be and randomly select persons from that location. This is a variation on simple random sampling called clustering.
| Simple Random Sampling Example | Cluster Sampling Example |
| Your target population is the student body of Lancaster High School. You have a complete list of all Lancaster High students. You randomly select every fourth student on your list until you have reached your desired sample size. Then you mail a survey to those students selected, at their homeroom addresses. | Your target population is the student body of Lancaster High School. You do not have a complete list of all Lancaster High students. You randomly select homerooms on each grade level. For each homeroom you randomly select a student until you reach your desired sample size. Then you contact the students in their homerooms. |
The sampling procedure may oversample or undersample certain categories of respondents. For example, a list of all the doctors in a certain region may not be completely up to date and thus will underreport younger doctors and those who have recently moved into the area. Sampling directly from the telephone book will also have some bias. It will bypass any individuals who do not have a telephone, or who have unlisted telephone numbers.
Make sure that the bias likely to be found in your sampling procedure will not seriously affect the results of your survey in any way that is critical. For example, if you feel that omitting people with unlisted telephone numbers will seriously misrepresent your target population, you may undertake any of the following steps:
| BACK | STEP 3: DECIDING ON A METHOD OF CONTACT |