The ethics of transcribing qualitative interviews

Transcription is fundamental to (most) qualitative research. The transcript is (usually) the actual thing you analyse as data, rather than the audio recording of an interview or focus group. And yet it’s curious how little attention it gets. I run a qual methods course and I barely mention it. I’ve done a quick online search and I can’t find much in the methodological journals; ditto the textbooks I have on my shelf.

Nor does it get much attention from procedural ethics. This is pretty odd given the audio recording is arguably the most identifiable research material we generate – it is the one piece of research material with all the details in it, the names, the places, the details you’ve promised the ethics committee you’ll change in order to protect the participant. And it is in the participant’s voice. I’ve never been asked and have not seen any advice that I should address who’ll be transcribing, whether they signed a confidentiality agreement, how they will securely store the material, or if they will destroy the audio recording and the transcription.*

New PhD students get told to do their own transcribing. It’s good experience blah blah; it’ll get you closer to the data. To be fair, it can be a great way to get novice interviewers to really engage with their technique – there’s nothing like *hearing yourself* talking too much, repeatedly interrupting or umming, ahhing and stumbling your way through a tortured question. But give a qual researcher some discretionary cash and I’ll bet they spend it on transcription. It is the most boring and onerous of research tasks. I’ve seen PhD candidates do paid work just so they have cash for transcription. I’ve seen projects flounder because the researcher is totally over transcribing. I’m sure there are researchers who’ve left the last one or two or three interviews un-transcribed in order to get on with things.

And why not pay someone to do it? There’s is no guarantee transcribing your own material will produce a more accurate transcript – paying a professional means getting someone with superior skills. You may get there in the end but it will take several extra hours of your precious time. Chasing up a couple of unclear words marked up by the transcriber will still be quicker than doing the job yourself. You can even anticipate some of this by providing a list of jargon, abbreviations, medical terminology or common phrases used by the participants. Have you got the message that I think transcribing is dull, onerous work, and exceedingly time consuming so best left to the experts?

But, those experts are expensive. I’ve recently been quoted a basic rate of AUD$2-3 per audio minute. The price goes up for more than two speakers, speaker identification, bad sound quality, sometimes for English as second language and fine-grained transcribing (eg for conversation analysis), and a shorter turnaround. That’s AUD$150 for an hour long interview with 1 native English speaker, great sound quality and a week turn around. 10 interviews AUD$1500, 20 interviews AUD$3000, throw in a real (i.e. diverse) sample and well, costs spiral. I was once involved in a grant application that budgeted AUD$50,000 for transcription costs. That’s coming close to a full time research assistant. I  will hold my hand up to making decisions that should be methodological – how many interviews do I need, interviews or focus groups, how long do the interviews need to be – based on reducing the budget. There are projects I’d like to do but don’t because I haven’t got the cash for transcription. I’m sure I’m not alone here. Transcription is an entry level barrier to qual research.

Let’s review – transcription is vital to qualitative research, it’s a methodological and ethical blind spot, an onerous task, and it’s expensive. Hang on – it’s not expensive anymore! A North American company rev.com will transcribe for US$1 per min of audio (AUD$1.3), with a 12 hour turn around and 99% accuracy. And, they don’t charge extra for anything: multiple speakers – free, speaker id- free, superfast turnaround –free, and apparently no sales tax. To make this concrete, that AUD$50,000 budgeted for transcription – rev.com would have charged us AUD$13,000.

My research is generally funded by the public purse. I should reduce costs where I can (right?) and rev.com offers one hell of a saving. I can spend that cash on something else (like research translation) or not ask the public purse to give me so much money in the first place. I don’t know how big the research transcription market is but should all publicly funded research be using rev.com?

I do have a slight misgiving about Australian public funds leaving the economy.

I also wonder about the transcribers. rev.com promote the fact that the transcription is done by people, not robots or computers. Their transcribers are unlikely faster than those employed by my local company. A good transcriber is looking at 3-4 hour’s work for an hour of audio – or longer if it is multiple speakers, crappy recording quality, lots of jargon, etc (i.e. real research data). My local transcriber charges me AUD$150 for their 3-4 hours of work; rev.com will charge me half this. I don’t know what my local company pays their freelancers (Google tells me typists earn AUD$28-35 per hour), but a rev.com transcriber gets US$0.4-0.65 per audio minute, that’s at best US$39/AUD$50 for their 3-4 hours of work. Really, I have no idea the margins for my local company or rev.com but it might be half the pay for exactly the same task undertaken by pretty much similarly qualified people.

My uneasiness is about paying so little with an expectation of a high quality product; a product that is fundamental to our work as qualitative researchers. Maybe if you won’t do it yourself or can’t afford to pay minimum wage, then you simply can’t afford to do qualitative research?

Be interested to hear what my qualitative research colleagues think. You using rev.com?

*I found one archived document from Newcastle University (2007, amended 2015) that stated “Transcription should usually be undertaken by the researchers. (ii) If it is to be undertaken by other than the researchers (eg, a transcription service or research assistant) participants must be informed as this potentially compromises data security and participants’ privacy.” I have never heard of a research project informing a participant their data may be comprised because a transcription service is being used.