Sampling, generally speaking, is one of the most useful methods for researchers to draw conclusions about a broad group of people. By asking a portion of this group—the sample—specific, carefully selected questions, researchers can gain valuable insights about the attitudes, beliefs, and behaviors of the broader group as a whole.
As you might expect, there are many different types of sampling methods and data collection strategies. These methods vary in cost, in the ways in which the sample is selected, and in the conclusions they can yield. While a simple random sample will, ideally, generate a truly random subset of the population (think, drawing from a hat), other sampling methods might choose to use a different approach.
One of the most popular methods for generating samples is systematic sampling. Essentially, a systematic sample will take a large list of people and generate the sample by selecting individuals that are a fixed, periodic interval away from one another. This can help generate a seemingly random sample, without the time and cost constraints that come with truly simple random sampling.
Systematic sampling can be used to better understand large groups. Currently, systematic sampling techniques are utilized by a variety of different businesses and other organizations. In this article, we will answer some of the most common questions that people have about systematic sampling and explain how systematic sampling actually works. By taking the time to take a closer look at systematic sampling, you can decide if this particular strategy is right for your enterprise.
Suppose you have a population of 50,000 people and want to learn something about this particular group’s attitudes, behaviors, or beliefs (such as their views on a particular product or who they plan to vote for). In a perfect world, you’d be able to ask every member of this group their opinion and could then simply tally up the results. But we don’t live in a perfect world and doing this would be impractical, time-consuming, and very expensive. Instead, you might select a statistically significant portion of this population—say, 500 people (one percent of the population)—and use the responses of the selected group to draw conclusions about the population as a whole.
The question that remains, however, is how do you decide which 500 people to survey?
With systematic sampling, a researcher will take a list of every possible respondent, start at a random point, and then select the sample group using a fixed, periodic interval. In this particular scenario, the researcher will choose a random starting point and then select every 100th person on the list. The reason the researcher is selecting every 100th person is that the total population (50,000) divided by the desired sample size (500) equals 100.
Eventually, this will create a seemingly random (or unrelated and uncorrelated) group of 500 people that, statistically speaking, is large enough to adequately represent the broader group as a whole. If the “first” person selected on the list was the 86th person overall, the sampled population would then consist of the 186th person, the 286th person, the 386th person, and so on, eventually reaching the 49,986th person on the list and amassing a total sample size of 500.
Systematic sampling is appealing to researchers because it is both simple and capable of producing what—in most cases—is a truly random group. However, in some cases, systematic sampling can have its drawbacks and, without a list of the entire population, generating the final sample population can be difficult. This is why systematic sampling is just one of many possible sampling methods that researchers might consider using.
One question that many researchers might ask is, “when does it make sense to use systematic sampling?”
First, the researcher must consider whether they have access to a complete list of members of the sample. Without a complete list, it becomes impossible to select every 100th member (or nth member) of the sample because the researcher cannot know who these members actually are. If there is an incomplete list, the data will necessarily have been filtered—in what was likely a non-random way—meaning any conclusions drawn from this sample might be manipulated.
A systematic sample might make sense when calling every 100th name in the phonebook, but it doesn’t necessarily make sense when trying to interview every 100th customer at a grocery store—in the latter example, an alternative form of sampling known as cluster sampling will likely be used.
Second, systematic sampling is usually much more useful when the population being sampled is relatively large. With SurveyMonkey’s Audience panel, the audience being sampled exceeds 50 million, meaning that systematic sampling can often be very effective. But if you are trying to draw conclusions about a group of 100, surveying every 10th person might not produce statistically significant outcomes, even if these individuals have been assigned a random number are seemingly uncorrelated.
Systematic sampling is also ideal for situations in which there are no patterns present between the intervals. Asking every 100th person on a list organized alphabetically by last name can be useful because there is no evidence that these people have anything to do with one another. However, imagine a town that is organized with large properties on every street corner, with five small properties in between them. In this specific situation, it would not make sense to survey every sixth person in the (numerically organized) address book because the sample will be skewed to overrepresent the views of the larger property owners.
Systematic sampling is ideal for researchers that have budget constraints because it is usually the most affordable way to generate an observably random sample. So, ultimately, systematic sampling is ideal for large and complete data sets, data sets void of systematic patterns, and research projects with limited resources. If this describes the research question you want to have answered, you may want to consider generating a systematic sample.
Using the resources provided by SurveyMonkey, it is easier than ever to create a systematic sample. However, before you begin the surveying process, you will need to do some basic preparations. Generally speaking, creating a sample using systematic sampling can be broken down into four main parts.
Before you can sample a population—using any sampling method—you will need to adequately identify and define the entire population. If you are the owner of a business, the population you want to sample might include previous customers at your store. If you are considering marketing a new product, you might want to sample the population at large or sample a specific group of people (broken down by demographics, geography, etc.). Regardless, recognizing the population you want to learn more about will be a crucial first step.
Next, you will need to identify the size of the population and the ideal sample size. Be sure that the sample size is statistically significant (this will depend on the size of the entire population). Having both these figures in hand will be crucial for creating a systematic sample.
Every member of the broader population will need to have an assigned number. If the population is already organized in a seemingly random way, such as alphabetically, assigning a number should be easy. If not, you will need to randomize the population prior to numbering. This will help you know who is 86th on the list, 186th, 49,986th, and so on.
The interval, in the case, will be the nth person that is included in the survey. To calculate the interval, simply divide the total population by the number of people you want to be included in the survey. If you have a population of 50,000 and want to include 500 people in the survey, your interval will be 100.
Once you have a randomly (or essentially randomly) organized list of the entire population and the interval, you will then be ready to determine who will be included in your sample. In the example above, a researcher will then choose a random number between 1 and 100 (using a random number generator or equivalent), which will help them determine their starting point. From there, they will be able to complete the sample and will finally be ready to begin the surveying process.
Generally speaking, there are three different ways to generate a systematic sample:
Systematic random sampling is the type of systematic sampling described above. The researcher, working with a fixed sampling interval, will choose a starting point between 1 and n (the number of people being sampled), they will continue down their list at fixed intervals until their entire population set is completed.
Linear systematic sampling is very similar to systematic random sampling. To create a linear systematic sample, researchers will arrange the population in a classified sequence, determine the sample size, and then calculate the sampling interval. They will then select a random number between 1 and the sampling interval, and then linearly keep adding new members of the population, until the desired sample size is reached. It is okay to round to the nearest integer in the event that the sampling interval is not an even integer.
Circular systematic sampling is generally useful for researchers working with smaller populations. To create a circular sample, you will need to begin by calculating the sampling interval and then select a number between 1 and the sampling interval. Next, you will continue skipping the sampling interval until the desired population has been reached—this may include “circling” back and passing the initial starting point yet again. The key difference with circular sampling is that the sample sequence begins again at the same point after ending, rather than skipping forward.
As you will quickly discover with any type of sampling, systematic sampling is something that has both advantages and disadvantages associated with it. Before committing to conducting a sample using systematic sampling—as opposed to other types of sampling—consider these following benefits:
Systematic sampling is one of the easiest ways to create a truly randomized group of people. If there are no patterns present between the intervals, then the final sample population will indeed be completely random. Randomness is often crucial for achieving statistically significant results. In many cases, this makes systematic sampling vastly preferable to “less random” sampling methods, such as cluster sampling or sampling filtered by a self-selection bias.
Systematic sampling is also very practical. While there are countless specialized approaches to sampling, many of these approaches are only feasible in a very small set of circumstances. Systematic sampling, on the other hand, can generate a random sample population in seemingly any situation where the prerequisites are met (having a complete and randomized data set). Furthermore, when compared to other sampling methods, systematic sampling is generally among the most affordable options available.
In statistics, simplicity is often what produces that most profound—and significant—results. When researchers try too hard to manipulate sample populations or otherwise modify their data, their initially good intentions could end up skewing results. All things considered, systematic sampling is systematically simple, which should generally be considered to be a desirable quality.
Of course, as you might expect, systematic sampling also has some drawbacks, which is why other sampling methods are often utilized by researchers.
Systematic sampling can be very useful when researchers have access to a complete and randomized data set, but unfortunately, this is not always the case. In many instances, researchers will only be able to survey a specific, non-randomized subset of the population, making it difficult to draw any firm conclusions about the population as a whole. If you are considering conducting a systematically sampled survey, be sure that you have the information needed to do so.
If the organized data set is not organized in a random way, this can lead to clustering issues and systematic biases. For example, if addresses ending in the number 5 correspond with a specific type of residency (as was illustrated earlier), an interval that overrepresents this subpopulation can lead to incorrect conclusions about the population at large. Furthermore, if the dataset was created using specific groups, such as units of a company, this can create a clustering effect. Once again, it becomes clear that the reliability of the conclusions being drawn from the data will only be as good as the underlying approach to sampling.
If the sampling interval is 100, the research team can theoretically have 100 different subpopulations available to choose from. Suppose the research team is gauging support for a ballot initiative, which we will refer to as Proposition A. If, in reality, the proposition truly has 50-50 support, this means that most randomly selected subpopulations will exhibit a level of support between 40 and 60 percent. If a supporter of Prop A conducts a survey that reveals 40 percent support, they might select another data set that is closer to their desired support of 60 percent.Internally, an organization could—quite simply—choose not to manipulate the data. However, because these data sets could be manipulated by ill-intentioned researchers, this particular sampling method can generate scrutiny, particularly in industries that are prone to statistical manipulations, such as politics.
There are several reasons why SurveyMonkey Audience is ideal for systematic sampling. With an audience consisting of more than 50 million nearly-randomized people, it is easy to conduct a variety of different types of surveys and generate statistically significant results.
Furthermore, SurveyMonkey makes it easy to find a random population, assign the population specific numbers, and then systematically generate a subpopulation. With limited time and resources, organizations of all kinds can gain greater insights and adjust their own behaviors in response.
Systematic sampling is a sampling methodology that samples a larger population via fixed, repeated intervals. When all else is equal, it is a sampling methodology that is simple, straightforward, and easy to conduct. While there are indeed some drawbacks, such as strict pre-requisites and clustering issues, this type of sampling is still incredibly popular among researchers around the world.
Collect market research data by sending your survey to a representative sample
Get help with your market research project by working with our expert research team
Test creative or product concepts using an automated approach to analysis and reporting