There are a number of sampling methods to choose from, but one will fit your survey better than the others. Usually, the purpose of your survey and your target audience will help determine which method you should use.
Sociologists believe that the number of people that a given individual can actually know is around 150.
These aren’t Facebook friends that you can add with the click of a button. These are the people you know well and whose minds you understand well enough to predict what they may think or how they may feel about something.
Zoom out to a 10,000-foot view of society, where the United States alone has over 300 million inhabitants, and there are over 7 billion people on the planet. Making it virtually impossible to know what everyone in a target market thinks about a topic, or about your goods and services. This is where surveys come in handy.
“Researchers usually cannot make direct observations of every individual in the population they are studying. Instead, they collect data from a subset of individuals—a sample—and use those observations to make inferences about the entire population.”
—University of California, Davis, Psychology Department
Probability Versus Nonprobability Sampling
The question then becomes, how to find the right subset of people for your survey? There are two primary sampling techniques that researchers use: probability and nonprobability sampling.
Probability sampling, or random sampling, identifies a target population and individuals are chosen at random. Every one in the target population has the same chance of being selected. So, if you’re conducting a phone survey, it doesn’t matter who in the house answers the call.
Nonprobability sampling is technique where samples are gathered using a methodology that applies additional criteria. So that not everyone in the target population has an equal change of being selected. In this case, that same phone survey has a specific person it’s trying to reach. Not just anyone will do.
Random vs Systematic vs Cluster
So what, then, are some of the pros and cons of each collection method?
If you want to give everyone an equal chance of being able to respond to your survey, a simple random sampling will do. Random sampling is subject to the vagaries of chance. Your sampling could be skewed heavily towards one or more segments of the population, depending on who gets chosen and who responds.
Systematic sampling uses a structure for the selection process, but it’s an arbitrary structure of your choice. By implementing a consistent selection scheme, you can refine your sample. However, because the schemes you apply to the selection process are arbitrary it’s also possible to end up with a sampling that isn’t representative. For example, if you choose to poll every 100th name in the phone book (remember those?!) you still might not get representative data.
With cluster sampling, you can narrow your search by selecting sub-groups within the larger population. You can narrow your selection pool by any demographic factors available to you; age, gender, income, etc. For example, you may be interested in polling stay-at-home dads instead of all stay-at-home parents.
The best strategy for you will depend on what questions you need answers to, and who constitutes your target population.
Stratified sampling takes a lot of thought and planning, but using more intricate methodology produces more accurate results. If you can identify sub-groups within a larger population, the idea behind stratified sampling is to poll those groups independent of each other.
Because stratified sampling yields data specific to specific sub-populations, if you know the sub-group’s representation within the entire community, you may be able to use that information to draw broader conclusions about the community as a whole.
When identifiable sub-groups exist this type of survey makes a lot of sense because the number of individuals is more manageable, and thus they can be easier to locate and contact for surveying purposes.
However, if there are no clear sub-groups in your population a stratified approach can be time and resource draining.
None of these approaches are better than any of the others. They’re just different. Figuring out what information you’re after and who you need to target should dictate which sampling model you use. Each has its own pros and cons, so it’s up to you to determine which one works best for your survey needs.