Sampling, Probability

Four Basic Types of Probability Samples

There are multiple methods commonly used in social science research for drawing a sample, used individually or in some combination. The most straightforward probability sample is the simple random sample, in which each subset of n units in the population has an equal probability of being included in a sample of size n. A researcher could generate a simple random sample by taking a list of every unit in the population and drawing n random numbers (where n is the desired sample size) from the set of integers that range from 1 to the number of units in the population. Drawing a set of random numbers is straightforward with most statistical software or sites that can automatically [Page 1539]provide randomized numbers. Units from the population list corresponding with the random numbers would be included in the sample.

In a systematic sample, the researcher takes a list, selects a random starting point, and selects each kth person for the sample, where k is selected to result in the desired sample size. This approach is useful as long as there is no systematic pattern of the list (e.g., every kth person in a yearbook used as the list is a class officer).

In a stratified random sample, strata, or mutually exclusive groups in the population, are identified. Strata may consist of racial or ethnic groups (when drawing a sample from a population of adults), size (when sampling of U.S. cities), type of program (when sampling television programs), or any identifiable, mutually exclusive groups in the population. The researcher draws a simple random sample from each of the groups of interest with the proportion of the sample from each stratum equal to the proportion in the population. Among other advantages, stratified random samples can have greater precision in estimating population parameters than simple random samples.

...