Choosing a sample for a research study is a very important step that can greatly affect the results of the research. If a sample is not chosen well it can bias the conclusions or even make the results unusable.
Random sampling is the simplest form of sampling methods where all members of a given population have an equal chance of being chosen for the sample group. One technique for insuring a random sample is to assign numbers to the population and choose the sample via unsystematic selection of numbers. However, to achieve the perfect random sample, the sample must be finite and all members of the population must be known and listed to avoid bias. In addition, the sample selected may be difficult to reach due to geographic location or other factors making it a costly and time-consuming method.
A method similar to random sampling, systematic sampling is the process of selecting a predetermined random member from a sampling list. For example, every fourth data point would be selected from a random list of population members. For this reason, systematic sampling is also called Nth name selection. Systematic sampling is slightly simpler than random sampling as the Nth method is a built in randomizer creating less bias. However, care must be taken that the sampling list is not in some kind of order, as this would lead to bias and errors.
Stratified sampling is thought to be superior to random or systematic sampling because it reduces possible errors. With this sampling method, groups and subgroups are listed within the greater population by factors they have in common. The statistician then determines their percentage of representation in the population and randomly selects a proper number from each group or stratum to sample in order to represent the population as a whole. This reduces the error that could occur with a simple random sample not including groups that have a low incidence in the overall population.
Random sampling, systematic sampling and stratified sampling are examples of probability sampling. In probability sampling, each member of the population has a known chance of being chosen for the sample. For example, if the population consists of 1,000 people and sampling is completely random, any one person has a 1 in 1,000 chance of being selected. An advantage to these sampling methods is that sampling error can be calculated since the selection is random and should represent the entire population.
Types of sampling that are not random are referred to as non-probability sampling methods. Examples include convenience sampling, where the easiest group is sampled to save time and costs, and judgment sampling, where the researcher chooses a representative sample based on their understanding of the population as a whole. Because the samples are not random, researchers cannot determine the sampling errors, or how much the sample might differ from the population. This could lead to problems with the results and incorrect conclusions from the data.
Each of these methods of sampling has pros and cons, and which one is chosen for a particular project depends on the scope, budget and subject of the research. Understanding the benefits and problems of each sampling method helps the researcher to choose the most appropriate one for the situation.