More than numbers: Sampling

So, medical school taught us all about the rules of sampling in  research – generally more is better, if you want to be more accurate then do a power calculation (although sometimes this may be akin to picking a number out of the air). And we all know that randomisation is good practice too – right?

Wrong. These principles hold true for lots of quantitative research, where you are going to use your results to calculate test statistics, and to answer questions about causation or relationships using numbers. However, remember back to the introduction to this series – in qualitative research, the questions, methods, reasoning and results are different to what we’ve been ‘brought up on’! This is also the case in sampling.

If you are trying to find out about experiences, perceptions and opinions, then one of way to learn more about these is to try to find out about all the possible different viewpoints, including the extremes. Therefore you might select the people who you suspect would have different viewpoints about the situation. For example – if you want to know the experiences of the services you provide on your general paediatric ward, you might ask some children, some older teenagers, some people who stay for just a short time, some of your more regular patients, some parents, maybe you’ll even ask the nurses, the doctors, the allied health professionals and the dometics. You might even ask the family who just put in a complaint! Choosing to specifically speak to people because their experiences are likely to be different is called ‘purposive sampling’ and, as you can see, is a justifiable approach in many aspects of qualitative research.

Another sampling approach used in qualitative research is that of ‘snowball sampling’. This technique is particularly good at helping to gain access to groups who might not normally take part in research. Within child health, this might involve talking to teenagers who are members of gangs. Can you imagine trying to find teenagers on your wards, in your clinics or even at local schools who would openly come forward to talk about being in a gang? Would you even know who to approach? Snowball sampling can help in this kind of situation, where the identification of just one or two participants can then be used to help identify other potential participants in their social network. In our example – finding just one key gang member who will agree to take part can then influence their contacts to also come forward. These can be even easier if the researcher themselves is part of the hard-to-reach population. (Note – we do not advocate the joining of gangs, even for research purposes. But if your research is into experiences of minority ethnic groups, you may find it easier to access members of the same ethnic group – do you see what we mean?)

Next, it seems like we must talk about the issue of numbers. Except that in qualitative research, numbers are less important. Please be aware that does not mean they are unimportant – they certainly have a role and should not be completely ignored but in qualitative research, there is no quick and easy ‘power calculation’ to decide how many people to include in your work. Instead, consideration must be given to the research question being asked and the aims of the study. To review perceptions of your service as described above, you might chose to interview three people in each of the categories given so that you have diversity of experiences. However, if you wanted to explore the experiences of a gang leader, even a single in depth case study may provide you with so much rich data to inform theories that you do not need any further participants.

So, minds blown – you can now see that there are many issues surrounding sampling in qualitative research – and many textbooks explore these from very different perspectives.

But what about you – a sampled minority – what do you think?

(Visited 76 times, 1 visits today)