12 February, 2026 | Jimmy Henks | 21 views

As a transcriptionist of several years (over a decade), I have noticed so many things that researchers/data collectors do during data collection exercises which compromise the usability of their data. Based on these observations, I submit these guidelines for qualitative data collection, with emphasis on the interview and Focus Group Discussion methods. The guidelines are to help qualitative data collectors make the most of their data collection exercises in order to obtain quality data. 

1. Location Matters: I cannot stress enough how important the location you choose for your data collection exercise is. What I have noticed is that many data collectors do not think much about their interview location. After all, they’re not the ones who will transcribe the data. Thus, some choose locations as noisy as a marketplace (no kidding) with many people talking in the background and music blaring loudly. Some also opt for venues where heavy trucks or other loud traffic passes by intermittently. With others, doors get banged continually.

What data collectors fail to realise is the fact that whilst they may not be the ones who will transcribe the data, in choosing a noisy location, they’re not only punishing the transcriber, but the lead researcher as well. Because at the end of the day, the noisy environment renders the audio recording unclear or inaudible resulting in unusable or lost data.

So, in selecting your location, avoid places with excessive noise, for even Artificial Intelligence (AI) cannot help to recover lost data in such cases.

2. Avoid Crosstalk: Prior to the interview/Focus Group Discussion (FGD), have a discussion with your subjects about not talking over you or other speakers. Sometimes, interviewees/discussants are so eager to answer questions that they cannot wait for the interviewer to finish asking questions, and they talk over them. In cases where they are not the only subjects, they cannot wait for others to finish their submissions, and they talk over them. In most cases, crosstalk results in loss of data. Therefore, it should be avoided – even by the interviewer.

3. Diminuendo: The interviewer/moderator and the interviewees should project well enough for the recorder to capture their voices clearly. As they may forget, the interviewer or moderator should keep reminding interview subjects or discussants to speak up so the recorder can pick up their voices. They should also avoid resorting to diminuendo – fading out the volume of their voices.

4. Be a Good Listener: You are there to collect data, not to let your interview subject know how brilliant you are. So, minimize interjections, and listen to your interviewee’s answers. Even when you know the answers to the questions you’re asking, resist the urge to supply the answers, especially when your subject is finding it difficult to put their thoughts into words. Give them time to think and gather their thoughts so they can answer your question from their perspective.

5. Equanimity: Closely connected to the aforementioned, learn to be calm and composed. Don’t be upset or frustrated when your interview subject is not getting your question or doesn’t have an answer to your question. Take your time to explain it to them in a way they will understand. Then again, make sure your questions are clear enough. Some interviewers don’t go straight to the point when asking questions. They circumlocute so much that by the time they’re through with their submission, the interviewee is just confused.

6. Respect Your Subjects: Some interview subjects/discussants may give answers that may sound ridiculous, but don’t laugh at them. That is disrespectful. But, of course, you can laugh with your subjects about something humorous.

7. Ask Open-ended Questions: This is not a survey. It’s an interview/FGD. Therefore, don’t ask close-ended questions that will elicit yes/no answers. Not only does this make the interview tedious to transcribe, but most importantly, it results in scanty data. Craft and ask open-ended questions to elicit elaborate answers.

8. Be Flexible: Don’t be in a hurry to run through the interview guide given to you. This goes especially to data collectors (field assistants) who may not be the lead researchers. As the name implies, an interview guide is just meant to guide you through the interview or discussion, not to restrict you to a number of questions. Therefore, seek clarifications when needed by asking follow-up questions.

9. Identify Your Subjects: Moderators of FGDs should make it a point to consistently identify their discussants or require them to state their names/assigned pseudonyms before each contribution. I cannot stress this enough. It is surprising how some moderators fail to identify speakers before their submissions, thinking that the transcriber will figure it out on their own – perhaps by the voice differences of the speakers. But that is almost impossible to do with accuracy – even for AI. If you really want to know who said what, identify your speakers.

This also means that moderators will have to let their discussants understand that they must only speak when called. This is to prevent the situation where everyone speaks at the same time.

10. Discourage Side Conversations: Moderators of FGDs should ensure that there aren’t pockets of discussions going on within the main discussion. Therefore, FGD participants should be explicitly asked to avoid whispering or holding side conversations with others when someone else is speaking. Such behaviour creates additional background noise, which often results in poor audio quality and inaudible sections in the final recording.

So, if a participant disagrees with a speaker or has a different opinion about a submission being made, let them share it within the main discussion after that speaker has finished making their submission.