To obtain information about large populations, researchers use four probability sampling methods: simple random, systematic, stratified and cluster. Everyone in a given population has a known and equal chance of being selected in probability sampling, and, most importantly, people are chosen randomly.
Probability Sample's Usefulness
Imagine how difficult and costly it would be for a company to survey everyone in the United States every time it wants to know something about Americans. If a sample is created randomly and everyone had a chance to participate, then the sample's results would be close to the results of a census, which surveys everyone. Probability sampling is a crucial, time-saving and far less expensive way to obtain information from society than a census because its results can reflect a large population even though it surveys a small number of people. If a sample was not created randomly, which is non-probability sampling, then it’s unlikely the results reflect the entire population.
Simple Random and Systematic Sampling
In simple random sampling, people are randomly selected from a complete population list. Typically, each person or household in the population is given a number and a computer generates random numbers indicating who is chosen for the sample. Lotteries are a purely random sample. All ticket holders are in a lottery, but only a few are randomly chosen.
Systematic sampling is similar to simple random sampling with one difference: a pattern to the selection of participants. For example, a researcher may start at a random point and take every 100th name he finds in the Atlanta, Georgia, telephone book. This sampling method is used widely for consumer mail and telephone interviews.
Stratified and Cluster Sampling
Stratified sampling is useful when comparing different parts of a population. Researchers divide or segment the population in a way relevant to their needs and take a simple random sample in each segment. The segments are called subpopulations or strata. If you want to compare how 1,000 women and men feel about health care, then you could segment or stratify the population by gender and randomly chose 500 men and 500 women. You may segment or stratify a population in many ways, including age, education, income and location.
Cluster sampling includes two random processes. The first step is to divide the population into specific groups and then randomly select groups, not specific people. Then researchers run a simple random sample only in each chosen group. Researchers often use postal codes or large city areas to create a group.
A researcher may want to know how all Americans feel about health care by surveying 520 people. If he has a list of every American and randomly selects 520 people from all over the country, then that is simple random sampling. If instead he starts at a random point on the list of every American and selects every 700,000th person, then that’s systematic sampling.
If he divides the list of every American into 50 states and randomly draws 10 people from each state, then he uses stratified sampling. If he randomly chooses 26 states from the 50 states and then randomly draws 20 people from each of the 26 states, then he uses cluster sampling.