As part of your research, you will need to identify "who" you need to recruit or work with to answer your research question/s. Often this population will be quite large (such as nurses or doctors across Victoria), or they may be difficult to access (such as people with mental health conditions). Sampling is a way that you can choose a smaller group of your population to research and then generalize the results of this across the larger population.
There are several ways that you can sample. Time, money, and difficulty or ease in reaching your target population will shape your sampling decisions. While there are no hard and fast rules around how many people you should involve in your research, some researchers estimate between 10 and 50 participants as being sufficient depending on your type of research and research question (Creswell & Creswell, 2018). Other study designs may require you to continue gathering data until you are no longer discovering new information ("theoretical saturation") or your data is sufficient to answer your question ("data saturation").
Why is it important to think about sampling?
It is important to match your sample as far as possible to the broader population that you wish to generalise to. The extent to which your findings can be applied to settings or people outside of who you have researched ("generalisability") can be influenced by your sample and sampling approach. For example, if you have interviewed homeless people in hospital with mental health conditions, you may not be able to generalise the results of this to every person in Australia with a mental health condition, or every person who is homeless, or every person who is in hospital. Your sampling approach will vary depending on what you are researching, but you might use a non-probability or probability (or randomised) approach.
Non-Probability sampling approaches
Non-Probability sampling is not randomised, meaning that some members of your population will have a higher chance of being included in your study than others. If you wanted to interview homeless people with mental health conditions in hospital and chose only homeless people with mental health conditions at your local hospital, this would be an example of convenience sampling; you have recruited participants who are close to hand. Other times, you may ask your participants if they can recommend other people who may be interested in the study: this is an example of snowball sampling. Lastly, you might want to ask Chief Executive Officers at rural hospitals how they support their staff mental health; this is an example of purposive sampling.
Examples of non-probability sampling include:
Probability (Randomised) sampling
Probability sampling methods are also called randomised sampling. They are generally preferred in research as this approach means that every person in a population has a chance of being selected for research. Truly randomised sampling is very complex; even a simple random sample requires the use of a random number generator to be used to select participants from a list of sampling frame of the accessible population. For example, if you were to do a probability sample of homeless people in hospital with a mental health condition, you would need to develop a table of all people matching this criteria; allocate each person a number; and then use a random number generator to find your sample pool. For this reason, while probability sampling is preferred, it may not be feasible to draw out a probability sample.
Things to remember:
Further reading:
Babbie, E. (2008). The basics of social research (4th ed). Belmont: Thomson Wadsworth
Creswell, J.W. & Creswell, J.D. (2018). Research design: Qualitative, quantitative and mixed methods approaches (5th ed). Thousand Oaks: SAGE
Salkind, N.J. (2010) Encyclopedia of research design. Thousand Oaks: SAGE Publications
Vasileiou, K., Barnett, J., Thorpe, S., & Young, T. (2018). Characterising and justifying sample size sufficiency in interview-based studies: systematic analysis of qualitative health research over a 15-year period. BMC Medical Research Methodology, 18(148)