Sample Sizes for Qualitative Research

From Market Research
Jump to: navigation, search

Sampling for qualitative research is completely different and it is not unusual for qualitative studies to have tiny samples. To understand the impact of sample size on qualitative studies we will briefly pretend they are quantitative (see Determining The Sample Size for a discussion of sample sizes for quantitative surveys). Imagine a study where we are evaluating two new products, and the first of these products is a clear winner with 75% of people preferring it. With a single focus group of eight people we can compute the confidence interval as 35% to 97%. If we conduct two focus groups the confidence interval becomes 48% to 93%, so we still cannot even be sure that the product that was the clear winner is actually most preferred. With four focus groups, and a sample of 32, the confidence interval is 57% to 89%.

The true confidence intervals for qualitative research are actually much larger than this, as researchers rarely count the number of people to prefer a concept, instead relying upon their perceptions of the overall vibe of respondents, which adds more noise. Further, the samples that are recruited for qualitative research can be quite unrepresentative, which adds yet more uncertainty to any result.

The poor precision of qualitative studies means that a greater level of discipline is required when interpreting qualitative research. Qualitative studies should not be used for estimation. That is, they should not be used to pick winning concepts or to understand what people like (which are both estimation questions). Qualitative studies can, however, identify bad ideas: if nobody likes a concept then it is very unlikely to become popular with a bigger sample.

As discussed in Introduction, Theory and an Overview, market research involves estimation and explanation. What qualitative research lacks in rigor for estimation it makes up for in its improved ability to explain things (i.e., to create new models). Of course, estimation and explanation are not independent and a degree of caution is required when obtaining explanations from qualitative studies as the problems of precision make it impossible for there to be much in the way of certainty about the accuracy of any models derived from qualitative research. For this reason, it is common to use quantitative research to confirm any important findings from qualitative research (although many firms rely solely on qualitative research when conducting branding and advertising research, as often quant is seen as being poor for such research).