Random Selection

Random selection (sometimes called random sampling) is the way in which a particular subset of a population (i.e., sample) is chosen. For selection to be random, two criteria must be met: (1) the members of a population have an equal statistical probability of being selected for the sample, and (2) the probability of being selected is independent of whether any other member has been selected. This entry compares random selection to nonrandom selection, distinguishes random selection from random assignment, describes several types of random selection, and finally provides basic examples of random selection.

Random selection is a crucial part of most research designs. It determines the participants of the study, which in turn provides the data the researcher uses to draw conclusions. Appropriate use of random selection lays the foundation for a strong research study.

Nonrandom sampling introduces potential biases by failing to ensure the equal statistical probability of being selected for a sample. Random selection is essential for ensuring that a sample adequately represents the population to which the researcher intends to infer (i.e., generalizability). Thus, all inferential statistics virtually assume that samples have been randomly selected from the target population.

Random selection is commonly confused and used interchangeably with random assignment. However, the terms denote different foci. Random selection refers to the method by which the sample is selected from the population for inclusion in a particular study. In comparison, random assignment refers to the method by which study participants are randomly assigned to experimental conditions (e.g., treatment vs. control). Thus, random selection would typically precede random assignment for an experimental study.

[Page 1367]There are multiple types of random selection, such as simple random sampling (SRS), systematic sampling, stratified sampling, and cluster sampling. In SRS, each individual has an equal chance of being selected. In systematic sampling, every kth individual is chosen, with k being the population size (N) divided by the sample size (n). In stratified sampling, the population is divided into strata (i.e., groups) by the researcher. Within these strata, SRS is used to select the sample. In cluster sampling, the population is divided into natural groups which are preexisting. Within these natural groups, SRS is used to select the sample.

The type of random selection used is determined by the researcher, often based upon the researcher’s level of access to the population of interest. For example, if a researcher has access to all members of a population (e.g., employees at company X), then the researcher may utilize SRS. However, if the researcher finds it important that desk workers and sales representatives at company X are represented proportionately, the researcher may employ stratified sampling.

See also Cluster Sampling; Generalizability; Inferential Statistics; Random Assignment; Stratified Random Sampling; Survey Methods; Systematic Sampling; Validity; Validity Generalization