Qualitative samples: unashamedly skewed
This article is part of a series by Katie Harris, principal at Zebra Connections (if you missed the other parts, start here).
This post is about getting the sample right in qualitative research. It’s an important issue because if you don’t get the sample right, then even the most brilliant research techniques in the world won’t get you anywhere.
Sample. Funny word if you stare at it long enough. Which I have. This is my fifth attempt at writing this particular post. But I’ve now identified the problem; I’m trying to nutshell theory and I’m boring myself to distraction.
So I’ve decided to change my approach. Rather than bleat sampling theory at you, I’m going to pick three things that I think are important at a practical level. If you are interested in the theory, check out Wikipedia.
In my new and improved post, a word on:
- Sample structure
- Sample size
- Sample skew
Sample structure refers to the way the research sample is put together. This is important because it has implications for the research dynamic. For example, sometimes it works well to include both males and females in the same focus group, sometimes it doesn’t. Sometimes grouping by lifestage makes more sense than by age. Or vise versa. Sometimes the strength of the customer/brand relationship, or a customer’s product usage, will be key considerations in how the sample is split.
There are no hard and fast rules here; the best way to put a sample together varies according to the research context. The key thing is to be aware of, and manage, the research dynamic to get the best out of your research participants.
For a quantitative research study, sample size is paramount. Generally, the bigger the sample, the less measurement error, and the greater confidence you can have in the results.
For a qualitative research study, sample size is important, but for completely different reasons. We’re not going for a head count, we’re going for a read of the heart. We’re looking to capture and explore, not measure, a range of attitudes and perceptions. Diversity, breadth and depth is the name of the game.
There’s no definitive ‘science’ to choosing a qualitative sample size. It simply needs to be big enough to make sure that:
- All segments of interest are included
- You talk to enough people, within those segments, to get a wide range of response
In research fairyland, where the funds flow freely, you wouldn’t worry too much about sample size. You’d just keep talking to people until you decided you weren’t getting anything new.
In the real world, a qualitative sample size is more or less determined by budgetary considerations. If anyone reading this disagrees, and has research fairyland budget, call me. I want to work on your projects.
Another ‘real world’ determinant of sample size is the researcher and/or clients’ personal point of view on how many segments to include and what constitutes ‘enough people’. This point of view will be based on experience, their risk profile, and the characteristics of the particular market of interest.
What if the sample is skewed?
Relax. It’s supposed to be. Qualitative research samples are purposive. They’re unashamedly skewed. Remember, we don’t need, or want to, include everyone. We have specific questions we want to ask of specific people. So long as we get a good range of people within our specific sample, we’ll get useful information.
Check out my next post covering issues relating to researcher bias, subjectivity, etc.