Sampling

Sampling refers to the process of selecting the people or the objects to be used in an analysis. Although sampling is most commonly associated with surveys, sampling also can be used to select messages for analysis. For example, a practitioner may be interested in examining a subset of the news stories that appeared about a product recall, a lawsuit, or a plant opening. Although this discussion will focus on sampling people, similar principles can be applied to sampling messages.

Researchers use sampling when they cannot survey everyone from the population of interest. A population refers to the group of people of interest. A census is when researchers are able to survey everyone from the population. However, typically this is not feasible. Researchers determine a sampling frame—a complete list of the membership of the population from which they will select their sample. The sampling frame is composed of the set of people that have a chance to be sampled. Researchers use a sampling method to obtain a sample, a subset of the population that is used to represent the population. A primary goal is to ensure that the sample closely matches the population so that one can generalize from the sample to the population. In other words, you want the sample to represent the population. This discussion briefly overviews basic concepts associated with sampling. Readers should consult additional references for more detailed information.

There are two categories of sampling methods: (a) probability (scientific) sampling and (b) nonprobability (convenience or nonscientific) sampling. Probability sampling requires the use of a random selection of people to be included in the sample. Every member of the population has a known probability of being included in the sample. Probability sampling is preferred because it can provide more accurate, unbiased data and permits the calculation of the sampling error. Sampling error often is reported as the “margin of error” and reflects how confident the researcher can be in the accuracy of the results. Sampling error is calculated using a formula that assumes the random selection of cases to be included.

Sample size is a concern for researchers because it affects the accuracy of the sample. Statistics books provide tables indicating the sample size desired for a particular level of confidence in the accuracy of the results.

Commonly used strategies for probability sampling include simple random sampling, systematic sampling, and stratified sampling. Simple random sampling resembles drawing the sample from names put in a hat. Names of the members of the sampling frame are put into the hat and the desired number are drawn. Computer programs can be used to approximate pulling names from a hat.

A second sampling method is systematic random sampling. This method requires an unordered list of members of the sampling frame (e.g., members are not ordered by geographic region, age, or any other characteristic). A starting point is selected randomly (e.g., the 12th, 37th, or 101st [Page 816]name on the list) and every nth name (e.g., every 11th, 20th, or 35th name) is selected for inclusion in the sample.

...