What's the issue?
Sampling for qualitative research is essentially different from sampling for quantitative research. When sampling for quantitative research, we usually have in mind the representativeness of our sample, to be able to make generalisations about the population. In qualitative research, however, our aim is not generalising but explaining the phenomena as comprehensively as possible, focusing on specific meanings and practices. It is not the purpose of our qualitative study to determine how typical a phenomenon is for the population. Usually, we do not want to make inferences beyond our sample.
Children for focus groups or interviews, and sites for our observation, are sampled based on researchers' decisions about what characteristics are important for our sample.
We can draw a sample from a quantitative sample by asking children at the end of survey whether they are willing to participate in focus groups or interviews as well.
If we only do qualitative research, we can sample children at schools, through our own or our children's social network, through parents if dealing with younger children.
Whether doing online or offline qualitative research about peculiar or specific topics (e.g. focus groups with young IT experts), it is easier to sample at web discussion forums focused around that particular topic. This way, it is easier to sample from specific populations which are difficult to 'recognise on the street'.
Try to be as specific as possible about the sample of children you include in the qualitative study because that allows you to be more exploratory.
For conducting focus groups with a broader age range (e.g. 8-18), we invite children of similar age (e.g. 8-9, 10-11, 12-13, etc.) to be in the same groups.
The size of a sample for qualitative interviews is good enough if ranging from 20-40 (if we need to compare findings, we can double it). When dealing with a very specific group of children, the sample can be even smaller. Anything beyond 50 can only mean putting in extra effort, which can be better used to be much more careful about the consistency of interpretation and analysis.
Pitfalls to avoid
One common pitfall is to insist on representativeness when sampling for qualitative research. No matter how accurately we sample to ensure a representative sample, our efforts will not pay off in qualitative research. We will never be able to do a big enough number of qualitative interviews or focus groups to ensure a sample large enough for generalisations, which we are not aiming for in the first place. Always try to bear in mind that we are not aiming for generalisations. We are not trying to tell how many people think that, but why they think as they do and what are the reasons behind that thinking. We can always follow up the qualitative part of our study by a quantitative survey to test for generalisations.
Questions to consider
How many qualitative interviews do we really need? How long should the observation of a specific site (e.g. a school yard) take? How many focus groups do we need, considering that we need a series of them if that is the main method of data collection? Based on which criteria will the focus groups be divided?
Kuzel, A.J. (1992). Sampling in qualitative inquiry. In B.F. Crabtree & W.L. Miller (Eds.). Doing qualitative research (pp. 31-44). Newbury Park, CA: Sage
Mason, J. (2002). Qualitative researching (2nd ed.). London: Sage.
Lincoln Y. S., & Guba, E. G. (1985). Naturalistic inquiry. Beverly Hills, CA: Sage.
A researcher's mistake
In a qualitative research project investigating, through interviews and observations, why only some middle class households adopted cable television, our research team contracted a recruitment company to locate 10 households with, and 10 households without, cable. We stipulated that the households should be from the London area for, though less than ideal, it was convenient for the research team since the project timeline was short. Mistakenly, as it turned out, we assumed that the agency had a database from which to draw a sample from all over London. Instead, we received a sample entirely based in Potters Bar, a small town just north of London, where a large proportion of residents commute into London each day. Worse still, we discovered later that one recruiter had gone from door to door in a particular part of Potters Bar, while the other approached people shopping on a Saturday morning, both thus producing rather homogenous samples. The lesson to learn is to ask the recruitment agency how they work, to specify in the contract that the sample should, as much as possible, reflect the diversity in the population sampled and, if concerns remain, to check with the interviewees themselves just how they were recruited. (Leslie Haddon, UK)