Once the basics of market research sampling are understood, specific techniques can be explored. Each research project objective is best served with a certain type of market research sampling. Straight talk from your sample provider can help set a research project up for success. Understanding the options available can also speed up the process, helping deliver a shorter time span to first value for any study. As research design begins, often stakeholders are included in important discussions. But these stakeholders may not have experience in the field or be familiar with respondent sampling options. Likewise, survey design may be impacted by the choice of sampling technique, so having a short primer can be helpful. What follows is an overview of six techniques frequently used for market research sampling.
True random sampling is the gold standard for probabilistic studies, but it may not be attainable with certain limitations. The most common method of random sample selection involves assigning numbers to the population of available participants and the use of a random number generator for selection. This technique is used so that each respondent within a pool has the same probability of being selected for feedback. In this way, it aims to remove bias by allowing chance to dictate research participants.
While the advantages of removing bias are desirable for many research projects, random sampling is not always attainable as it requires access to a larger database of willing survey participants. With respondent screener requirements, creating a large pool is costly and time consuming. Unless a significant database is immediately accessible, the time involved in achieving representative numbers makes this technique difficult to get projects done within budget and time constraints.
Systematic sampling is the more common cousin of true random sampling. It allows for a methodical approach to retain the benefit of removing bias that comes from random sampling. Systematic sampling is based on the size of the population to be surveyed. The actual number of that population is then sampled according to the radio needed for representative sample size. Because this sampling method does not require such a large pool of ready participants, it is easier to achieve the required sample size.
Example:
If a list of all of St. Louis households numbered 139,594, and the requested representative sample was one percent, systematic sampling would start with a random spot on the list and then select every 100th household until the 1,395 households are selected.
A modified approach to systematic sampling is called stratified sampling. This method is often preferred in an attempt to reduce the chances of a misrepresentative population sample. If respondent data is first separated according to strata of demographics, or segments. Participants are then selected systematically from these strata.
Example:
That same list of St. Louis households could be broken down into strata such as those living in single family owned homes, single family rental units, multi-family housing, etc. The systematic sampling method is then started to ensure the final sample pool adequately represents each strata.
Quota sampling is a technique which takes stratified sampling to greater detail by requiring quotas to be reached within each of the strata. In this way, it is a non-probabilistic version of stratified sampling. The downside to quota sampling is that with an inadequate sample pool, quota sampling can force a hand-picking of participants which gives an opening for bias. Also, as the selection is no longer random, the margin of error cannot be calculated.
When a large population, especially one over a wide geographic area, needs to be sampled, cluster sampling can be a good option. In research design, the population of study is broken down into smaller groups, referred to as clusters. Individuals are then selected randomly from the clusters to create the study sample.
Example:
Five large St. Louis neighborhoods are selected as clusters and individuals are selected randomly from each cluster to represent opinions from every neighborhood in the city more accurately.
A large number of small clusters to sample from is preferred to a small number of large clusters. This is best practice is used to eliminate cluster sampling bias (CSB) which can occur when some clusters in a given territory are more likely to be sampled than others.
Area sampling is a random sampling technique applied with geographic boundaries. Area sampling is a common technique used for ecological or geographic sampling. It is not often used in the study of humans or human behavior as the potential to end up in areas with no observational instances, such as a desert, or sparsely populated area is possible. If area sampling is used to observe humans, selecting random areas within a known populated region can ameliorate these issues by combining some standards of cluster sampling with this technique.
Before choosing a sampling technique for any study, the objectives and measurement protocols should be determined. The goals of the study, the available resources and the amount of time allotted for study completion and analysis are factors that may immediately eliminate some techniques and guide the decision for the best possible outcome.
Once the sampling type is confirmed, screening potential participants begins. To take the next step, review our guide, Survey Design Best Practices and Screening Question Template. This free guide offers an easy-to-follow and comprehensive resource for getting started.