Quantitative research typically focuses on a defined target population. This is the group of people, events or items that you intend to study and to draw conclusions about, and this might be very broad or quite specific. For example, if you want to examine whether physical activity influences mental health outcomes in adults, your target population might be people aged 18 and over living in Western Australia, or it might be people aged 18 and over studying at Curtin University (keeping in mind that your choice of target population will influence how generalisable your findings are).
Even if your target population is relatively small, it is generally not possible or practical to obtain data from everyone (or everything) in it. Instead, you will usually collect data from a smaller subset of this population. This is known as your sample, and the process used to select it is called your sampling method. There are two broad types of sampling methods:
Probability sampling: involves the random selection of participants (or items) from the population.
Non-probability sampling: involves the non-random selection of participants (or items) from the population.
Probability sampling methods are typically required in quantitative research, as the usual goal is to use the sample to draw conclusions about the target population. The sample must be representative of the population in order to do this accurately, and probability sampling helps achieve this by ensuring that everyone (or everything) has a chance of being selected.
This page outlines five commonly used probability sampling methods, before concluding with some information about sample size.
Simple random sampling is a probability sampling method where every individual in the target population has an equal chance of being selected. Similar to drawing names out of a hat, each participant is chosen purely by chance.
This method is most suitable when the target population is relatively small, making it feasible to select participants entirely at random. You would choose this over other probability sampling methods when you want the simplest and most straightforward approach, with minimal risk of introducing bias through the selection process.
Systematic random sampling is a probability sampling method in which participants are selected at regular intervals from a list of the target population. For example, every 10th person on an alphabetical list might be chosen.
This method is useful when the population is large. In this case, it provides a straightforward and efficient way to select participants evenly across the population, as long as there is no hidden pattern in the list that could introduce bias.
Stratified random sampling is a probability sampling method in which the target population is divided into subgroups, or strata, based on characteristics such as age, gender, or location. Participants are then randomly selected from these, ensuring that each subgroup is represented in the sample.
This method is most suitable when you want to include specific subgroups of the target population in proportion to their size, in order to ensure that important characteristics are reflected in the sample.
Cluster sampling is a probability sampling method in which the target population is divided into groups, or clusters, such as schools, neighbourhoods, or hospitals. You then randomly select some clusters and include all individuals within the chosen clusters in your sample.
This method is most suitable when the population is large or widely spread out, making it impractical to list or reach every individual. Cluster sampling allows you to collect data efficiently while still maintaining a probability-based approach.
Multistage sampling is an extension of cluster sampling. You first randomly select clusters from the target population, and then take a further random sample from each cluster. This second-stage sample can be selected using simple, systematic, or stratified random sampling, depending on what is most practical for your study.
This method is most suitable when the population is very large or widely dispersed, as it allows you to collect data from a manageable number of individuals while still obtaining a representative sample.
If you would like to consolidate your understanding of the sampling methods outlined above, you may find the following activity helpful:
In addition to deciding on a sampling method, you also need to think carefully about how many participants to include in your study - in other words, your sample size. This is typically the total number of people (or events or items) in the study, but it can also include the number of measurements or observations made on each person.
In general, larger sample sizes are preferred because they allow smaller effects (that is, differences between groups, changes over time, or associations between variables) to be detected. However, there is usually a limit to how many people can be sampled due to resource availability, and careful consideration of sample size is therefore important to ensure the research question can be answered reliably and ethically. Statistical considerations, ethical concerns, and context-dependent tolerances for error (for example, between a high-risk and low-risk intervention) all need to be factored in.
From a statistical perspective, there are two types of potential errors that could be made when analysing your results:
Type I error: concluding there is an effect when there isn’t.
Type II error: concluding there is no effect when there actually is one.
Researchers usually select acceptable levels for these errors (for example, 5% for Type I and 20% for Type II), and then calculate the sample size needed to detect a meaningful effect. The probability of detecting a true effect is called the power of the test, and this is equal to 1 minus the Type II error rate. Therefore, researchers generally aim for this to be 80% or higher.
For more information on determining the appropriate sample size for 80% power, you may like to watch this Power video. In addition, you may like to download the free tool G*Power to calculate your required sample size.