Part 6 – Sampling and generalising
The point of this series is not to cover what’s already in dozens of qualitative health research textbooks, but to offer some ‘back room’ insights into the possibilities and limitations of this underused resource.
So far I’ve looked at where QHR came from, the concept of criticality, the ‘insider’ or emic perspective, power, and last week, your role as the researcher.
Today I want to focus on one of the most common questions people pose about QHR, and that is “How can you possibly generalise from a study when you only have six participants?”
This is a great question. Not only because it strikes at the heart of one of the important differences between qualitative and quantitative research, but also because when qualitative health researchers get it wrong it’s a red flag for a poor study.
A common give-away that the author(s) didn’t really understand QHR is when you read in the limitations that “the study’s findings should be read with caution because of the small sample size”.
So what’s wrong with this?
Well firstly, qualitative research isn’t trying to be representative of a background population.
It starts from the premise that each person is unique (a view shared by most physiotherapists, as it happens). No two people experience the world, never mind health, in the same way.
Qualitative health researchers believe that any attempt to shoe-horn individuals into groups of convenience to the researcher is an act of wanton violence. (Note: the word ‘Individual’ means ‘cannot be divided’, so the idea of ‘like minded individuals’ is an oxymoron).
No, your qualitative methods should work hard to accentuate people’s individuality, not crush it.
So sampling in qualitative research isn’t about finding people who are alike, but finding people who are not. You want breadth, not conformity.
So the quality of the study has nothing to do with the sample size, and one person can provide as much rich data as 1,000.
To illustrate my point, consider this scenario:
Imagine you are the owner of a large practice and you decide to sample 1,000 of your clients to ask them whether they were satisfied with your service.
856 of them (85.6%) report being very happy.
A further 141 report being basically satisfied.
But three are not. One reports being verbally abused by one of the reception staff. One thought the clinic rooms were unhygienic. And one talked about the inappropriate touch of a therapist of the opposite sex. They don’t want to make a complaint, but thought you should know.
Now clearly you wouldn’t dismiss these three on the basis that they were statistically insignificant. Because you know that these individual views are particularly powerful, and you don’t need it reported by a minimum of 10 people to act on their concerns.
QHR works in the same way, but here’s an important difference.
Good QHR is not just about reporting what people said. That’s journalism. Neither is it about cobbling a few similar comments together and calling them ‘themes’. That’s the equivalent of using pie charts in an RCT, and I talked about this in an earlier post.
Good QHR builds theories and concepts, and it builds them sometimes from very particular, individual comments that – like the one-in-a-thousand comment from one of your clients – speaks volumes.
This process is hard to explain in a brief blog, but I’ve written about it in the articles attached below if you’d like to know more.
So if you ever see ever read in a qualitative study that the research was ‘under-powered’ or wasn’t generalisable because it only involved five participants, you’ll know this is a sample you should ignore.