The art of interviewing
I quite do a lot of interviews. It’s actually not a task I particularly enjoy, nor something that necessarily comes entirely naturally to me. So, I figured I’d come up for some tips for myself on how to do them better, and thought I’d share.
Interviewing is not really a science, but more of an art that can be crafted. I chose the above photo — pulse interactive — because it conveys some of the dynamism of an interview (a kaleidoscope of lights in a room of mirrors), and if you look carefully I’m part of it. That’s intentional, because the interviewer is not merely a passive, clinical, extractor of information, they’re a key part of what is an interactive exercise.
Some aspects of interviewing such as sample size seem more like science. There are several “magic numbers” out there for saturation point: 12, 24, 40. I recently asked the designers of the Qualitative Impact Protocol (QuIP), Fiona Remnant and James Copestake, why 24? Fiona told me, “it’s not really a magic number (of course), but a reference point to riff off.” Similarly, James’ response was that it depends on the primary evaluation goal — is it exploratory or confirmatory? What we’re looking for, pragmatically, is minimal justifiable effort, but leaving the door open for the possibility of further studies to add depth or breadth later, if needed. Generally, there are diminishing marginal returns after a certain point, and even if there isn’t really a magic number there are some rules of thumb and fuzzy thresholds that give us somewhere to start.
However, what we’re looking for from an interview isn’t simply a representative picture, we’re seeking valuable answers. Key informant interviews, for example, are really seeking key information, and this depends more on the quality of the interview rather than how many interviews you’ve done. Simply adding more poorly designed and implemented interviews adds close to nothing. So, let’s think a bit about quality.
Part of quality relates to how you structure the information you’re looking for and the questions you ask. For example, Margaret Roller has a useful funnel which I often use as a general schema.
As Roller notes, you ‘begin broadly and progressively narrow the topic area to the subject matter of greatest importance to the research objectives.’
I’ve found this is a generally good guide, but experience from conducting surveys taught me not to wait too long before asking your most important questions. Wait too long and your key informant or programme participant might give you a worse answer because they’re tired, bored, or need to go. I interviewed an MP recently who came back and forth to the interview as they were literally voting in parliament. Had my colleague and I stuck to the funnel, we would have wasted the (very) limited time we had without getting the key answers we really needed. So, it’s well worth considering beforehand which questions are the most important.
Relatedly, don’t ask too many questions. The more questions you ask, the less time respondents have to respond to them, so your answers will be shorter and less rich. I did an online search on the assumption that I would quickly find empirical studies demonstrating interview response fatigue after a particular threshold of questions. Attention spans are about 47 seconds (some estimates are considerably less), and over the past couple of decades, people’s attention spans have shrunk considerably. So, it stands to reason that long surveys or long interviews will likely trigger fatigue, unless the person you’re talking too is really interested in the topic. 10 questions is a reasonable rule of thumb. But, in practice, it depends on the interviewee — some people give long answers, other short answers. So, again, there isn’t a perfect number that will suit all respondents.
There’s also a balance to be struck with the breadth of questions you ask. You both want a respondent to have the freedom and liberty to respond comfortably (in this way interesting new information you hadn’t thought of can emerge) because they’re more likely to speak authentically (and, maybe, tell you the “truth”). But you also want the respondent to hit the target and answer the main area(s) of inquiry you have.
There are also plenty of studies and advice out there on why we shouldn’t ask leading questions, why we should avoid jargon, among others. But, beyond technical criteria, I wanted to ofter a few rules of thumb (for myself) as to how to behave well during an interview. Interviews are an interaction and a relationship between the person asking, the person answering, and the environment in which the questions are asked and answered. AI interview software is becoming increasingly sophisticated, and perhaps, in time, it will become effective in mimicking human traits, as Yuval Noah Harari argues in Nexus. However, the human interaction still seems to elicit a different type and quality of answers, whether done well or poorly. So, here are some thoughts based on my trial and error of trying to do better.
1. Show gratitude
No-one owes you their time. So, the bare minimum you should do is thank the respondent for their time and what they’re prepared to share with you. Don’t assume that they’ll give you everything you want, you don’t have any right to a persons deepest, darkest secrets, any more than they do to yours. Not all interviewees will provide you with valuable information. Sometimes they provide totally useless information and waste your time (your time is valuable too, I know). But, you should still act cordially and demonstrate to the respondent that what they share with you is valuable.
2. Build a connection
Another key aim in interviews is to build rapport. Even in standardised surveys, building rapport is important. One study found that it often ‘involves extensive unscripted conversation between the interviewer and the respondent.’ What you’re trying to do is to build a connection with the hope that the respondent will given you valid and useful answers. Anthropologist Bronislaw Malinowski was famously lied to by communities he visited. If you see photos of him with communities and how he talked about himself and others, this is hardly surprising.
Larry Bremner’s argued at the European Evaluation Society Conference last year that we should share something of ourselves with those we interview. I noted the following in a recent article:
‘The idea of evaluators reciprocating — ‘sharing part of us’ evidently challenges traditional interviewing techniques and styles and would likely make interviewers who view themselves as being “objective” and “independent” uncomfortable, opening up all kinds of concerns about response bias.’
Perhaps we can go too far, but reciprocity and relationality are important in the evaluative process itself. We shouldn’t shy away from this. So, don’t be afraid to share something of yourself and make personal connections. Interviews don’t need to be (and shouldn’t be) a one-way extractive process.
3. Be transparent
Thirdly, be honest and transparent with those of whom you’re asking questions. Sometimes you don’t know all potential information uses beforehand, but you should at least give interviewees a sense of how you plan to use their information. Remember, it’s not your information, it’s theirs unless and until they give you their consent to use it. They have have the right not to respond to any questions they prefer not to answer, and they should be allowed to change their minds whenever they so please. This can be difficult and frustrating at times (colleagues and I have a dilemma of evolving information uses presently), but how you share information with interviewees shapes the information they provide you. Hiding information from them, at the very least, will dispose them to hide information from you.
4. Be present, and listen
Perhaps the most important and most difficult part of an interview is really being there. There’s potentially a lot of anxiety entering into an interview — what do I need to ask? Did they really answer my question? How much time do we have? Etc. Despite all these real-time distractions, it’s important to try to be present and listen as carefully as you can to what the respondent says, because only then are they likely to share interesting and useful information with you. If you’re disengaged and distracted, why would they bother sharing?
If it’s an interview on a screen, put your camera on. If someone has generously given you their time, at least show them who you are. But, presence goes beyond whether someone can see you, it’s about whether you are actively listening to what they have to say. They need to see that — it should be written on your face (e.g., you nod), evident in your gestures (e.g., scratch your chin), and in your sounds (words of affirmation, agreement, where appropriate). I’m not saying this is natural to everyone. I have a nervous tick when I’m listening — a sort of Rodin thinker gone wrong — which I didn’t know I had until I saw myself on screen recently. However, it’s important to try to be engaged and to show it.
Another challenge of being present is trying to force things. Leading questions are an Achilles heal in interviews because the respondent should be given the space to say what they think, as much as possible, rather than anticipating or repeating what you think. Early on in my career, I was once described as being Paxmanesque during an interview in Uganda, presumably being too insistent to get an answer (though it was really because I couldn’t hear what the person was saying). On another occasion in Bolivia, I was keen to get a clearer response from a researcher and my respondent answered “soy muy gris (I’m very grey).” I needed to be comfortable with ambiguity. It can be helpful to feed the interviewee back some of that they said to sense check with them. An interview isn’t a debate, it’s a discussion. You should also consider offering respondents choices and options (i.e., was this good or bad, were there positive changes or maybe not?). Interviewees should always feel that they have the option, because when they don’t they will likely revolt, or exit.
5. Be flexible
And perhaps underpinning most of the above, you need to be flexible. This is one of the hardest things to be when you are short of time and believe you need certain information. I can recall a case where I commissioned a company to do an interview for me in Rwanda a few years ago and was very careful in providing interview guidance. However, likely short of time, the interviewer was so desperate to hear answers that linked responses to programme activities that they forced responses. I couldn’t use any of that information.
It is important to be flexible in the interview rather than insisting that every question is answered to your satisfaction. It’s important to provide sufficient room for the interviewee to given you their view and for insights to emerge. You should feel free to change the order and wording of questions, cut questions, and introduce (a small number of) follow up questions. This real-time editing can sometimes be really difficult, even more so if you’re not working in your first language. For instance, I tend to be more structured in Spanish than I am in English as I don’t have quite the same linguistic dexterity, and I have to listen harder in Spanish than in English.
But, more than the technique or question editing, what counts is that the person you’re talking you feels that you’re listening to what they actually say and responding to that, rather than simply regurgitating a script. At the end of the day, you’re trying to get the best out of a conversation.
I selfishly wrote this blog as tips to myself, but hope those tips are of some use to others.
