Social science research methods fall into two basic categories; quantitative methods and qualitative methods. Quantitative methods are about quantifying things – counting them – while qualitative methods are about discovering the qualities of things – telling stories about them. All the methods I’m using in this project fall into the ‘qualitative’ camp. Today, let me share a little about interviewing.
A couple of factors are important when you are using interviews to gather data:
- Sampling and recruitment – how interviewees are selected and recruited and how many interviewees you engage;
- Context – including both the physical context of the interviews (e.g., location) and who the interviewer is;
- Content – what questions you ask and how you ask them; and
- Analysis – how you interpret what people are telling you.
In a quantitative project, it’s very important that your sample is random and representative to ensure your statistics are valid. While it’s still a good idea to get a good spread of people to participate in your qualitative project, there are slightly different measures of quality in play.
Often, you just start where you can – with someone who is willing to talk to you – and then ask them to refer you on to others who would also have something to contribute to the project. This is called ‘snowballing’ because of the way your project grows in size and momentum as it (metaphorically) rolls down hill (like a snowball). Often, there’s no set sample size. Rather, the researcher aims to achieve ‘saturation’. That is, they are aiming to do enough interviews that each successive interview is introducing them to less and less new information – reaching a kind of saturation point. When you hit saturation varies from project to project – it could be ten interviews, it could be fifty, it could be more.
Where you have interviews and who does the interview can be pretty important in setting the tone for the conversation. In most research, the interviewer is perceived as having more power than the research participants (usually because they do). This can then affect the types of things people are willing to tell you. It’s my experience that researchers are encouraged to redress this perception by letting research participants choose the location for interviews, by dressing in a way that matches/mimics/appeals to research participants (e.g. wear a suit to interview business executives, or jeans with uni students etc.). I want to go a bit further and suggest that it shouldn’t just be that research participants perceive the power imbalance to be minimised, but that it really should be minimised! (You might like to see some previous posts on this kind of thing here and here.)
Interviews are basically a conversation with an agenda. Sometimes they are highly structured with an exact set of questions that are asked of each interviewee – unsurprisingly, these are called ‘structured interviews’. Sometimes there is a loose set of questions that the researcher uses as a guide, but they are free to expand on if the conversation veers into interesting territory that they hadn’t expected – ‘semi-structured interviews’. And sometimes, the researcher just asks a question to get things started and then lets the participant guide the conversation – ‘unstructured interviews’.
The benefit of structuring things more loosely is that people tell you things you didn’t expect – and this is often where the most interesting insights come from. But unfortunately, the less you structure things the more time you spend sifting through information that is irrelevant. As a result, I have always tended to embrace the semi-structured approach.
The idea that you can reach ‘saturation’ in your data collection assumes that, while each individual is unique, there are going to be commonalities between peoples’ experiences and opinions. In collecting lots of data from interviews we are seeking to find patterns – variations on a theme (or themes) – in peoples’ accounts and reflections. It is identifying these patterns (and their variations) and theorising how or why they might be occurring that is the job of the researcher.
Interviews are most often analysed thematically. Having done the interviews and transcribed them, the researcher develops a list of topics they think people have been bringing up. They then go back through the transcripts and identify peoples comments according to the themes. This process is iterative – you might need to edit or add themes as you go back and engage with that data. And it might happen multiple times. The researcher then reflects on the material and uses their theoretical knowledge to say something useful about it all. Often the most interesting things have already been said by participants and the researcher’s job is figuring out how to tease that out.
So, how do you feel about doing an interview, now?