I think this needs to consider two sides:

Transcribed interviews can help you as a researcher to better code the data and to find/organize illustrative examples of code pieces. If you can code audio files to some degree, in my opinion it makes no differences as long as you can work with the data and organize your research transparently and efficiently. A reason why one should transcribe interviews is that you can read much more faster than listening. So scanning capability (e.g. when you need to go back in your data for some reason) plays also a role. So transcribing is pain, but for intensive working its better.

From a reader's perspective it might be a bit different, because it is much more convenient to read a transcribed quote in a text instead of switching to a different medium and searching for earphones to listen to an audio file (and still you need an internet connection to stream the audio files - but maybe not the biggest problem). An other issue might be to think about search engines that index your text and not the audio files.

But maybe there are some interesting tools which I do not know. Do you have some examples? So, I would tackle the problem also from an other solution perspective: How can you automatize transcribing interviews? (I know researchers who use youtube that transcribe audio files). Maybe others have some ideas or practises on this...

Thank you very much so far for the very interesting answers!

I can see the advantage of a transcribed document when coding, especially from an efficiency perspective (reading faster than listening); though I am unsure whether this justifies/outweighs the effort of transcribing?

In my opinion, sharing interviews could be done with an audiofile; with consent.

For a publication, I suppose, one could transcribe selective passages for quotes and/or - at least in in online publications - link to a specific point in an interview  (similar to songs on spotify).

In the software I am using (NVivo), it is possible to code audio files too. I am sure other softwares can do that too.

I also see the disciplinary perspective though I wonder if that is more a cultural practice than a methodological necessity.

Can you think of any methodological reason for not using audiofiles for coding?

@Martin: Good point about the automated audio to text software; perhaps worth a new question?

Thanks again

Dear all,

indeed interesting answers. I am working with MaxQDA which also contains the implementation of audio files. But I think the strenght of this function is that you as a researcher in a coding process can shortly switch to a specific point in the interview. Sometimes - and this would be the strenght of audio files - you can hear underlying tones (polemic or irony). In this case (similar to Katherine) it is important now WHAT the interviewees says but also HOW they say it.

In my opinion it is absolutely necessary to trascribe your interview. The transcription process can give you some first insights in teh content and form of the narrative.

If the goal is to be systematic in one's analysis of qualitative data, there are several software programs that allow you code audio data without transcription. Even so, I think there is a major element of convenience here because audio data cannot be skimmed through or searched. Not transcribing will thus save money but not time.

As for the unique advantages of listening to the spoken word, there is nothing that stops you from listening to recordings once they are transcribed. And, if you want to go the effort, several of the CAQDAS program let you link the audio and text versions of files and move back and forth between them (basically involves a lot of "time stamping").

Also, there are already several discussions on RG about automated transcription and the simple answer is that it doesn't work for interviews. Instead, it is optimized for one person speaking in a dictation format.

In qualitative stuides, there is also an issue to consider called "theoretical saturation" (saturación teórica), which indicates that at some point in the qualitative research concepts are repeated, or new information or perspectives can not be heard in the what the informants say. Beyond the technical issues of the automated transcriptions,  codings, etc., it is possible that when we find saturation a complete transcription of interviews may be inefficient (with a time consuming - cost benefit perspective) , for academics, marketing researcheres, contractors or clients.

Transcription is really helpful to a researcher not only for the coding process, but some instances it helpful for later references.

I think the questions is "Do you have to transcribe verbatim"? There is a tradeoff in technique. Verbatim transcription is expensive and time consuming- Do you have the money and time? On the other hand it may be faster to read transcripts (during analysis) than listen to them. I saw several comments that verbatim transcription is easier to code- but easier than what? I never saw a comparison. I have done coding successfully with the listiening technique. One hazard that few people mention related to the verbatim transcription is accuracy- If you hire someone to transcribe for you, their typing skills are not the only skill needed. It has been suggested that because of the many errors typically found in transcription, the researcher will need to check about 30% of the transcripts against the original recording. This means additional time also. If errors are found, then more recordings will need to be audited.

However, if you have the time to transcribe your interviews personally, there is an advantage because the researcher can actually begin to do analysis as they reflect on the interviews they are transcribing. This advantage may disappear if there are too many interviews to make this practical.

It is understandable that this question occurs, as there is little in the literature that provides guidance. In a quick review of recent qualitative articles, I have found that researchers typicall say "Interviews were transcribed verbatim and checked for accuracy by the researcher", but this tells the reader little about the actual approach. How was accuracy assessed? By whom? Who did the transcription? When the investigator audited the transcripts (compared them to the digital recording), how many errors were found? How were errors treated? These issues all speak to the rigor of the study methods.

Although verbatim transcription is often identified in the methods section of qualitative articles, there is little evidence (only opinion, because that is the way we have always done it) to suggest it is an imperative. Thus, like many things in research, the investigator needs to examine the strengths and limitations of each method (or combination of methods) and make decisions based on what is or is not feasible, given the available resources (time and money).

Sandra L Siedlecki hi Sandra, I noticed that you have coded successfully without transcription, I was wondering whether you had any sources that demonstrate this at all?

I'm at undergraduate level and have been told not to transcribe, and that i can code using listening technique but all of the literature tends to support the transcribing....

Try this article from 2006

Halcomb, E.J., & Davidson, P. M. (2006). Is verbatim transcription of interview data always necessary? Applied Nursing Research, 19(1), 38-42. https://doi.org/10.1016/j.apnr.2005.06.001

Thank you so much Sandra L Siedlecki I also found this : http://www.qualitative-research.net/index.php/fqs/article/view/1564/3249

I believe it is important. IT means the transparency of the research work, accesible in any time to annyone (if it is needed). Audio tracs can be lost or some kind destroyed...For me it means it is done, it is documented...It also means /provides accuracy in every point of you research ...it is written down, you cn also check the statements nd in that way it means a higher level of quality...It is "safe", it is easier, more "visible"...But we hconsideration that people are different and there are different ways of communication that suits us...I hope you understan...I think it should be audio and transcript. Best regards...

that's a really great question, and although i am quite late to this chain ---- i'd say no, it's not necessary to transcribe interview data, and here is why: https://ocean.sagepub.com/blog/whos-disrupting-transcription-in-academia