Article Summary: To understand how sample sizes can be so small in qualitative research, it helps to understand the concept of “data saturation” which is when you start to get diminishing return by adding more participants, since the same themes keep coming up in the research.
One of the most common questions we get when new clients reach out to us is “how many participants should our study have for qualitative research?” When we tell them, the number is often much lower than they expect, and the next follow-up question is typically “is that really enough?”
And the answer is yes. And the reason that qualitative research needs so few participants is the principle of “data saturation.” Think of it as the law of diminishing returns. We’ll explain.
First, off, let’s discuss sample sizes in qualitative research
In qualitative research, we’re not testing how many, or how much. That’s the realm of quantitative research, which requires a large enough sample size to have a high confidence interval. Typically, sample sizes start at 200 and go up from there in quantitative studies.
In qualitative research, we’re testing opinions, ideas, observing how people use products, and we’re doing exploratory research to learn. We don’t go in with pre-filled assumptions, which is required in quantitative research (in order to create a multiple-choice survey, for example, you first have to know which options to test). Qualitative research is the realm of learning and exploring human behavior, and it is far more in-depth than quantitative research. For this reason, qualitative research relies heavily on personas and segments. Instead of taking a population and asking everyone the same questions, the population is first segmented into similar characteristics. Researchers then interview these segments, typically with different discussion guides, to explore topics.
Quantitative research takes large amounts of a population, and through statistical analysis, finds common patterns. For this reason, quantitative research can include heterogeneous populations – meaning, they don’t have to be the same segment. However, it’s the opposite in qualitative research: we’re actually trying to create homogeneous segments, with similar characteristics.
Therefore, in qualitative research, the sample sizes can be much smaller. In fact, with a homogenous population, you can have as low as 7 up to 30 participants, per segment.
So that covers qualitative sample sizes. Now, let’s talk about data saturation, and how this applies to qualitative samples.
What does “data saturation” mean in qualitative research?
Let’s say that we’re doing a qualitative study with a homogenous sample size of 15 participants (keep in mind that typically there are numerous segments, so the total study population may be 30 or 45 if there are 2-3 segments). For the 15 participants, they’re all asked the same questions, though the researcher will explore new topics that come up and probe on responses, leading to insights along the way. The researcher starts to notice that after about 10-12 participants, the answers and themes explored all start to sound very similar. In other words, in a homogenous population, there is a lot of overlap in how these participants frame an idea or experience a product, and not a lot of new information is gleaned. By 15 participants, the same themes keep coming up, and no new insights emerge. If the researcher were to interview 5 more people from this segment, this pattern would hold, and the same themes would come out. So essentially, adding 5 more participants is not going to yield new information; it would just confirm these same themes.
That’s what we mean by “data saturation”: there is diminishing return in adding more participants once the key themes emerge and are confirmed repeatedly by participants. Adding more people doesn’t yield greater results.
So once a study reaches data saturation, the researcher can conclude that these are the key learnings from that particular study, and the report analysis process can begin.
Wait, but doesn’t this same thing happen in quantitative research?
Technically, you can also say that “data saturation” happens in quantitative research, say when “90% of a population says that X chose ___” But in this case, you’re comparing statistics to uncovering insights and opinions. Oftentimes, to quantify themes from qualitative research – if you really want to ensure that the themes learned apply across a bigger population – then quantitative surveys are used to ensure that the learnings from qualitative research do indeed make statistical sense. However, qualitative research is first used to explore what these themes are. The learnings are then packaged into a quantitative survey and tested with a much bigger sample size.
The key principle of qualitative research is to ensure that you do indeed have homogeneous segments. This depends on very specialized qualitative recruiting and screening methods. With a proper sample of qualitative respondents, you won’t need a huge sample size, and you should start to reach data saturation after about 10-15 interviews, depending on the topic.