Types of sampling for market research

Market research is crucial for any business that wants to understand the people it is selling its goods and services to. With preliminary research, businesses of all kinds can gain useful insights, identify new selling opportunities, and find ways to allocate their resources efficiently and equitably.

One of the most effective ways to conduct market research is sampling. Sampling utilizes data from a small group, such as a simple random sample, and allows marketers to draw conclusions about a much larger target population.

By ensuring that your representative group is actually representative of the population, and that the questions you are asking are effectively worded, you can pave the way for impactful and productive research. Without sampling, you will inevitably be forced to guess how to reach your audience. Not only will this be inefficient and cause you to miss out on valuable opportunities, but lack of sampling can also cause significant damage to your brand.

Fortunately, you can gain critical insights into your target audience by using the right types of sampling and strategically employing various sampling techniques. In this article, we will answer some of the most common questions that market researchers and business owners have about sampling. By taking the time to understand what sampling actually is, along with the different types of sampling that are currently in place, you can decide if committing to a broader sampling campaign makes sense for your particular organization. 

Sampling is a term used to describe the process of obtaining data from a small group (or subgroups). Once this data has been gathered, it can then be applied to a larger audience, such as a company’s target market.

Suppose a restaurant is targeting people between the age of 25 and 35 living in an urban area. The restaurant wants to decide what color it should make its logo. Rather than asking everyone in that age group what color makes them most likely to visit the restaurant, the company might take a sample of 100 people from that group and gather their opinions. If more than half of the people say blue is the most appealing color, the company can draw conclusions about 25 to 35-year-olds in general, and adapt their marketing approach in response.

Of course, the conclusions that can be drawn from sampling will only be as good as the sampling frame itself. In this instance, if the restaurant were to just ask random people about their favorite color, rather than those within its target audience, the conclusions it made might not be as reliable. In other instances, creating a pure, simple random sample (SRS) might be more beneficial. Before conducting sampling research, it is important to identify what conclusions you hope to draw and who you are hoping to survey. Once these things have been adequately identified, you’ll be able to use small samples to help you draw big conclusions about almost any topic. 

Researchers use sampling because it helps them efficiently learn about a group in general, without needing to survey the entire group. During an election, for example, it would be impossible to survey every likely voter about who they plan to vote for. Instead, a researcher would ask a specific group of voters about their preferences and attempt to draw broader conclusions from the responses they receive. While this sort of polling certainly presents its own unique challenges, it can still provide valuable—and actionable—insights for all involved.

Sampled surveys can be used to answer many different questions. Learning about how people typically live their lives, how people view the world, and how people use a product or service can help businesses develop better strategies and methods for reaching their target audience. There are many different types of sampling and each of these methods can be effectively applied in different situations, for different market research needs.

The various types of sampling methods will generally fall into one of two categories. The first category is random sampling while the second category is representative sampling.

A random sample, as the name suggests, is a sample of randomly selected individuals, designed to represent the population as a whole. Simple random samples can help companies and other organizations draw broad conclusions about people in general. If a company is trying to sell a product that essentially everyone might use, such as toothpaste, a simple random sample can help them draw broad conclusions. What flavors of toothpaste do people typically prefer? When do people typically brush their teeth? What type of toothbrush do most people use? These are questions that can be effectively answered by asking a wide range of people for their opinion, rather than limiting the survey to a deliberately narrow group.

In contrast, researchers using representative sampling don’t want a random sample of all people. Instead, they want a random sample of people that are representative of a specific group. For example, if a company is selling a product that only some people use, such as skiing equipment, they’d want a sample of individuals that actually use that particular product.

Representative samples can be broken down in myriad different ways. In the example above, “people who ski” could be a distinctive group that helps filter the broader population. In other instances, you might consider breaking the population down by age, demographics, location, income, hobbies, profession, or other traits. As long as you can find enough survey takers to generate statistically significant conclusions, you will have a considerable amount of flexibility when creating a representative group.

The different types of sampling can also be distinguished as either probability sampling or non-probability sampling. Essentially, with probability sampling, every single individual within the target group (which can be either random or representative) has an equal chance of being selected for the survey.

With non-probability sampling, on the other hand, some people within this group will be more likely to be selected than others. For example, if the group you hope to draw conclusions about is American adults, but you conduct a survey at a mall in Missouri, you are using a non-probability sampling method for your survey. In this case, you are not randomly sampling American adults because your broad group has been filtered down to “mall shoppers in Missouri.” This particular type of survey is known as a convenience survey (more info below). While it is indeed possible that these mall shoppers might happen to produce results that are similar to the opinions of the American adult population as a whole, it is important to recognize which portions of the broad group are systematically excluded by your sampling methods. 

As suggested, probability sampling is a type of sampling in which every single member of a group has an equal probability of being selected for the survey. Probability sampling can still exist within a filtered group (such as American adults), as long as every representative of this subgroup has an equal chance of being selected.

There are four primary types of probability sampling methods. 

Simple random sampling is both simple and random. That means that within a group or subgroup, each member of the population has an equal chance of being selected as a respondent. There are many ways in which a simple random sample can be created. For example, every person within the group might be given a number and then a specific portion of these numbers is selected entirely at random (using a random number generator, drawing from a hat, etc.). Simple random sampling offers the benefit of a “pure” random data set, enabling researchers to draw sweeping conclusions. However, simple random sampling is also criticized for being relatively inefficient. 

Systematic sampling is a type of sampling that involves selecting a random starting point in the overall population and choosing sample members at regular intervals. For example, if a researcher has a list of every resident of a city with a population of 300,000, they might choose to generate a random sample of people by surveying every 100^th person featured on the list. In this instance, 3,000 people will be surveyed. 

As long as there is no hidden pattern in the list that might skew the selection process, systematic sampling creates a sample where members of the selected population don’t appear to have anything in common. Systematic sampling still provides most of the benefits of random sampling because, when properly applied, the population essentially is randomly selected. At the same time, this straightforward method requires considerably less effort than other sampling methods. 

Stratified random sampling randomly selects from several subgroups in order to create the final sample. Suppose the researcher wants to gain insight about the opinions of American adults. Rather than simply selecting 500 random adults, the researcher might select 10 adults from each of the 50 states to create the “random” sample population. If each of the subgroups has a lower standard deviation (possibility of error) than the total group, then the margin of error can be systematically decreased. 

Cluster sampling creates a sample by pulling people from multiple (but not necessarily all) subgroups of a population. Ideally, each of these subgroups, or clusters, will be a diverse representation of the population as a whole and will also be structurally similar to the other subgroups. Cluster sampling is one of the least expensive forms of probability sampling and is also ideal for sampling relatively large populations. To successfully use this particular type of sampling, it is crucial for the clusters to be consistently structured and for the selections within each cluster to remain random. 

While probability sampling can be used to draw conclusions from random (though sometimes slightly modified) groups, non-probability sampling uses groups that are a bit more deliberately structured. Non-probability sampling can help reduce random biases and, in many instances, ensure that key portions of a broader population are included within the sampled population. 

Quota sampling is a sampling method in which the researcher manipulates the sampling population in order to represent the population as a whole. This type of sampling is especially useful when the broader population includes many different types of people. 

For example, suppose the survey is designed to draw conclusions about American adults. Rather than risking a random sample in which one group (race, gender, age, geographic location, etc.) is either overrepresented or underrepresented, the researcher might deliberately select a proportioned number of individuals from each of the conceivable subgroups. So if Black Americans represent 13% of the population, the researcher would deliberately ensure that the sampling population is actually 13% Black—and adjust other populations to be proportionally representative as well. By doing this, they would avoid a less accurate simple random sample, which might only be between 5-20% Black. Quota sampling is typically used for large, clustered populations, such as the population of the United States. 

Convenience sampling, as you might guess, is a type of sampling that is done by surveying a group of people that is easiest to reach. This sampling is often the easiest to conduct and is often very affordable. During a convenience sample, a researcher might go to a crowded public area and ask people if they are willing to be surveyed. This population is by no means randomly selected, but depending on the type of data the researcher is hoping to gather, that might not really matter. Convenience sampling is often used during a pilot study in which a company is trying to learn about the feasibility or popularity of a proposed product. 

Snowball sampling is a non-probability sampling method that is designed to help reveal information about populations that are difficult to reach or are “hidden.” With snowball sampling, researchers will encourage their already existing population to reach out to additional members of the population in order to help bolster the underlying data set. While this does create systematic biases, it is one of the best methods for reaching populations that tend to avoid answering random surveys, such as individuals engaging in illegal activity. Snowball sampling is only occasionally used by market researchers, but though it might be problematic, it has helped deliver data where other sampling methods were proven to be ineffective. 

Purposive sampling is a type of sampling in which researchers will directly (rather than randomly) select a subpopulation that is supposed to be representative of the population as a whole. This type of sampling is often called “judgment sampling” or “expert sampling” because it involves the judgment of someone who is familiar with the group and its basic characteristics. Purposive sampling is often characteristic of other non-probability sampling, such as quota sampling, but involves an additional layer of human intervention.

Want to learn more about sampling best practices? Read our Ultimate Guide to Market Research.

Survey sampling with a market research panel, such as SurveyMonkey’s integrated global panel, can help researchers and organizations quickly access a large, random population. When using these sorts of panels, surveyors will have the freedom to control the questions they are asking, the populations they are drawing from, and the types of surveying they choose to use.

Populations can be divided in many different ways. Demographics, geography, professional profile, and more might all be actively considered. These panels can be used for valuable insights, including basic market research, product development, brand tracking, and consumer behavior. By using a panel to look at a specific group of people, businesses can draw crucial conclusions about their broader target audience. 

Every type of sampling method will have both pros and cons that come with it. For example, while a simple random sample can decrease bias and help you draw broad conclusions, generating a truly random sample can often be very inefficient. Furthermore, you might want to learn about a specific subgroup, rather than the population as a whole. At the same time, while convenience sampling can help you quickly generate data, these sample populations can be extremely biased and may cloud your final conclusions.

Clearly, there is no universally “best” type of sampling. To determine which type of sampling makes sense for your campaign, you will need to begin by determining what—exactly—you are hoping to learn by conducting the survey. From there, you will need to consider other relevant variables, such as time and cost constraints, the ways in which survey questions will be worded, and whether the population you want to survey can be accessed with ease.

By making an effort to better plan your survey, it will be easier to determine which type of sampling will be most useful for you. With a firm understanding of the different types of sampling and access to valuable resources, such as SurveyMonkey’s audience of more than 80 million people, you can learn a lot about a population and conduct better market research.