Sampling relies on the random selection of individuals or objects. Researchers standardize how they order the units in the population. For example, an inspector might look at every third batch of peanuts. Systematic samples are very simple, fast and convenient for those who already have a list of units in the population. Statisticians benefit from using systematic sampling when studying large populations because systematic sampling covers the sampling area uniformly. For example, if a state department studies how residents use sunscreen, the department should ideally sample from the entire state instead of sampling from a few counties.
Events occurring at regular intervals have periodicity. For example, a television show that airs every Tuesday at 8 p.m. has periodicity. In a study, the sampled population might have periodicity characteristics. For example, salmon might swim up the river at the same time every year. Researchers might also notice that a pattern has periodicity. For example, there might be more bears in a region whenever the salmon swim up the river. But the sample periodicity might not match the pattern periodicity.
In another example, a statistician might randomly select health club members for a study. However, the selected participants might not represent the actual proportions of health club members in the population. The selected sample might happen to all have diabetes, while not everyone who goes to health clubs has diabetes. But situations in which the sample participants have characteristics that are wildly divergent from the norms of the sampled population are unlikely and repeating the study at a later date will reveal the abnormalities in the study.
According to the Stony Brook University website, finding multiple samples and repeating the study can increase the chances that the overall results of the study will be accurate. For example, a researcher could study the prevalence of a particular disease among potatoes by examining potatoes at four different potato farms. One farm might have an unusually high number of pathogens because of poor farming practices. When the researchers select four different farms for a second study and average the results of the first and the second farm, the abnormal farm will make up only 12.5 percent of the averaged results, instead of 25 percent.
Systematic sampling is a kind of probability sampling, meaning that the researcher must ensure that the sample equally represents all the members of the population. If they do not, the statistician will have skewed results, which are results that diverge from the actual characteristics of the population. For example, a college study might report that 70 percent of residents in Missouri oppose immigration reform. However, the college performs the survey by asking students attending the college. The results will be skewed, since the research will not represent all of Missouri but only the students.
Statisticians can avoid bias if they select units for the sample in a systematic way. For example, instead of relying only on the college students, researchers could call every 100th resident listed in the phone book to ask them questions about immigration reform.