To record interviews or not?

Recently I have had an on-going discussion with one of the team at Senatus about whether we should be audio-recording our interviews. This could be either by making an audio-recording on our mobiles phones,using the ‘dictate’ button on Word, or recording the MS Teams meeting (where we woudl have visuals as well as the audio).

I’ll admit, I’m old school. I’ve been taking notes at interviews and hearings since my very first job in employee relations over 25 years ago. I’d like to think that I’ve got quite good at it. I also think it’s not an easy skill to acquire – having to balance taking the notes, whilst actively listening to the interviewee.

The driver for this discussion is the persuit for an accurate, complete, set of interview notes. If we record the interview, we won’t have to worry about whether we missed an important point that the interviewee shared with us. Unless you undertake investigatory interviews frequently, you may not realise the extent that individuals do not speak in the same way as written text. People start sentences but quickly change their minds about how they want to express themselves; they might start a setence three or four times before they settle on the direction of their narrative. Individuals don’t use full stops; we carry on talking, using conjunctives such as ‘and’ and ‘but’. And in the moment, we understand a person because we also interpreting their non-verbal cues, but when you read the transcript it makes no sense at all because of the way that they presented their point. With transcripted interviews, technically, you do not need interviewees to sign-off the interview notes.

The expectation is that if you have recorded an interview, the participant is given a copy of the transcript. However, these transcripts will be full of the extraneous speech, the half-sentences, the unintelligible sentences. So, when you download the interview transcript, you have to edit it to remove any extraneous speech in order for the transcript to be meaningful. Even after you’ve undertaken this edit, these transcipts are often hard to read due to the interactions between the interviewer and interviewee. Each party might interrupt the other, finish each others’ sentence, talk over each other. As such, the transcripts do not support easy analsyis of the evidence. I’ve also learnt that some people don’t recognise their speech when you provide them with the transcript of what they have said – which throws up (unwarranted) issues of credibility on the part of the inteviewer.

With handwritten inteview notes, you will not be able to capture every word the interviewee has said, but you will automatically filter out the half sentences, or any comments that appear repetitive. We inform interviewees that it won’t be a verbatim document, so participants don’t expect us to capture every single detail. They are also informed that they can add to the interview notes when they review them. The format of the drafted notes will also be presented in a way that enables easy analysis.

From personal experience there is no ‘easy’ way to manage interview notes. Note taking can be a laborious and time consuming task. However, I subscribe to the belief that spending time drafting the interview notes after the interview enables you to get closer to the evidence, which both influences your interview prep for future interviews and helps you pull the investigation report together into a coherent and logical presentation.

Maybe this is just confirmation bias, but I recently came across an article by Rutakumwa et al. (2021) on this very subject that supports my personal view based on interviewing thousands of people over the last twenty or so years. I thought I’d share their findings. This won’t be a ‘spoiler alert’: their conclusion was that handwritten interview notes is considered to be the better approach in certain circumstances.

• Audio-recordings are considered to provide more accruate transcripts and therefore ‘better’ interviews (Green and Thorogood, 2009; Lee, 2004; Paulus et al., 2017; Tuckett, 2005).
• However, having a recording device in an interview influences the interviewee’s level of contribution (Rapley, 2004; Nordstrom, 2015). Whilst as investigators, we know that we will manage the recording in terms of confidentiality and integrity (as described to the interviewee), the interviewee may not feel the same. The interviewee may self-edit if they know that they are being recorddd and this will limit the amount of evidence they provide the investigator. Clearly, not informing the interviewee that we are recording them is unethical and would demonstrate a lack of integrity on our part if we did that.
• Furthermore Nordstrom’s (2015) research found that even the existence of a recording device will lead to some interviewees to self-edit.
• Nordstrom’s (2015) research considered the extent to which recording inteviews are normalised across different cultures, race, age, class, and politics. Whilst one interviewee might be happy to be recorded, the next will not. Or if they reluctantly agree to be recorded, they may self-edit.
• Even if an interview is being recorded, the interviewee may ask to go ‘off the record’. In those instances, the interviewer will need to make hand-written notes. And it is not always clear (to me at least), when the interviewee is happy to be ‘on the record’ again. Sometimes an ‘off the record’ comment can evolve into a 10 minute explanation, which may contain important evidence that you do want to capture as an interviewer.
• That an effective inteview arrives from an interviewee feeling comfortable talking to the interviewer about a particular topic (Oakley, 2016).
• Rutakumwa et al’s. (2021) research found that when interviewers were trained in hand-writing notes, the accuracy and quality of the notes was found to be similar to audio-recordings.

In essence, the issue of recording interviews is about trust, which in turn impacts on the overall quality of the interview. Whilst the transcript from an audio-recorded interview might be more accurate, it might lack depth in terms of evidence collated. If the inteviewee trusts the interviewer, then audio-recording might be beneficial and easier for the interviewer. But to what extent does the interviewee and the interviewer understand the level of trust between them when they meet for the first time to undertake the interview? Perhaps the safest approach is to keep to hand-written notes. This will generate a better outcome in terms of the quality and depth of evidence, which is crucial in any investigatory interview.

References

Green J and Thorogood N (2009) Qualitative Methods for Health Research. London: Sage.

Lee RM (2004) Recording technologies and the interview in sociology, 1920–2000. Sociology
38(5): 869–889.

Nordstrom SN (2015) Not so innocent anymore: making recording devices matter in qualitative
interviews. Qualitative Inquiry 21(4): 388–401.

Oakley A (2016) Interviewing women again: power, time and the gift. Sociology 50(1): 195–213.

Paulus TM, Jackson K and Davidson J (2017) Digital tools for qualitative research: disruptions and
entanglements. Qualitative Inquiry 23(10): 751–756.

Rapley T (2004) Interviews. In: Seale C, Gobo G, Gubrium JF, et al. (eds) Qualitative Research
Practice. London: Sage, 15–32.

Rutakumwa, R., Mugisha, J.O., Bernays, S., Kabunga, E., Tumwekwase, G., Mbonye, M. and Seeley, J., 2020. Conducting in-depth interviews with and without voice recorders: a comparative analysis. Qualitative Research, 20(5), pp.565-581.

Tuckett AG (2005) Part II: rigour in qualitative research: complexities and solutions. Nurse
Researcher 13(1): 29–42.

The limitations of memory in workplace investigations

Many investigations rely on participants’ recollections of events, and the information from an investigatory interview is a valid source of evidence.   However, there are times when participants make memory errors.    These are not conscious errors (as that would be lying, or potentially vexatious), but occur unconsciously due to a range of factors.

Memory errors can occur at two points:  The event itself, and at points after the event.   These are explored further below:

The event

Memory errors can occur at the very moment that the memories are being made.  The reasons for this could be:

1. The event is similar to other events that the individual has witnessed.   As such, when the individual is recalling the memory, they may confuse what happened at one event with what happened at another event.  A typical example is if you park your car in the same car park on a regular basis.  There will be some days when you forget where you have parked your car on that day.   This is technically called “source confusion”.

• The individual wasn’t paying attention at the time – either they were thinking of something else, or their attention was diverted to another event happening simultaneously.   As such, the individual won’t lay down a memory about the event.  It’s not possible to recall a memory when it hasn’t been stored originally.    But this doesn’t mean that their memory of the periphery events is any less accurate than other person’s memory.

• Similarly, an individual is more likely to remember something if it was the focus of their attention at the time. They are less likely to recall what was occurring on the periphery.

• Individuals are more likely to remember an event if it’s associated with a negative emotion.  However, if the event was traumatic, the individual may suppress the memory as a coping mechanism.

After the event

• The biggest challenge with memory recall is that memories fade over time (“memory decay”).   After 24 hours, memory is only 30% accurate.  After a month, it will only be 10% accurate.

• Memory errors can occur if an individual discusses the event with a second person, who shares new or different information.  Unconsciously, the individual may adopt this new information but stores it in their memories as if they had seen it themselves, particularly if that new information is plausible or they have learnt about this new information from someone they trust.  Therefore, when the individual is asked to recall the memory, they will include the new information, and truly believe that they witnessed this new information at the time of the event.  This is called “intrusion error”, and regularly occurs when witnesses discuss events between themselves prior to their investigatory interview.   It should also be noted that it actually aids the accuracy of an individual’s memory recall if the new information is accurate.  

• Our brain makes sense of the world by developing “schemas”.  For example, when you got into a restaurant for a meal, you know what will happen and in what order.   It’s not something that we are consciously aware of, and we will have numerous “schemas” for different situations.   When an individual is asked to recall a memory, they may “fill in the gaps” with information from their “schema”.   Again, this is an unconscious act, but occurs because the witness is thinking “this must have happened, because this is what always happens”.   For example, an individual may not remember that a colleague was late to a particularly meeting, but they know that the individual is late to every meeting.  Therefore, their “schema” will lead them to believe that they recall their colleague arriving late to that particular meeting. 

• “Imagination inflation”, is the term given to individuals who describe recollections which are partially imagined, rather than what actually happened.  

Factors to consider

• It is possible for witnesses to have different memories of the same event, and this could be attributed to what they were focussing on at the time the memory was being made. 

• Just because an individual’s memory is limited, it does not reduce the accuracy of that memory.

• Equally, if someone has a vivid memory “Flashbulb memory”, it does not mean that it is any more accurate than another person’s memory of the same event.  It just means that they remember more.

• It is not possible for an individual to know what a false memory is and what isn’t. If questioned, individuals will be adamant that their memories are accurate – unless they are proven otherwise by other more tangible and credible evidence.

• The level of an individual’s confidence of does not correlate to the level of accuracy in that individual’s memory of an event.

• Despite all of the above, individuals’ memories are more often accurate than not.

Therefore, in investigations, it is important that the evidence is triangulated, preferably with contemporaneous evidence such as audio recordings, CCTV, emails, diaries.   Failing that, the balance of probabilities, which is applicable for workplace investigations is used to determine what actually happened.

Judging the accuracy of witness evidence

I’m taking a detour in my reading at present into the world of memory. There is a considerable amount of literature out there in relation to police investigations, but not so much for workplace investigations.

One book that I’ve found particularly user-friendly is Daniel Reisberg’s The science of perception and memory: a pragmatic guide for the justice system. (The link is not to Amazon, but to Better World Books, which is a “greener” option when buying books).

The book has made me pause and think about the need for investigators to have a checklist to assess the potential for memory error in witness testimony. If such a checklist would exist, it would contain the following:

1. Whether neutral questions were asked by the interviewer. What may surprise you, is that asking questions such as “what were you paying attention to during this event?” will actually increase memory error.
2. The extent of the witness’ spontaneity in providing their memories. The less questions we ask as investigators, the less we will potentially contanimate their memories.
3. The extent that the event was clearly observed (e.g. viewing angle, amount of light, the complexity of the situation, the amount of attention that was being paid).
4. Contamination (otherwise known as source confusion): our memories may have been influuenced after talking to others about the event – particularly if it’s someone we trust (“intrusion error”); or we may have filled the “gaps” in our memory, unconciously.
5. The urgency in which the witness is asked to recall a memory: the more pressure we put on a witness, the more they will “fill the gaps” with erroneous memories. This has a technical term: imagination inflation.
6. Motivation: the drivers behind the witness’ desire for their recollections to be heard (which are not always noble)
7. Plausability: whether the recollection is believable or not (think: being abducted by aliens).
8. Retentional retrieval: it is harder to recall memories over time, and our memories fade quickly at first; we may also fail to recall a memory, as we’ve not been given the correct “retrieval cues”.
9. After the event memory: when someone recalls an event that they didn’t think was significant that time, but they do now. The chances are that they didn’t lay down a complete memory, and therefore unconsciously they will have “filled the gaps” as they sought to retrieve the partial memory.
10. Claiming a flash-bulb memory: when a witness claims that the memory is as clear as the day that it occured….it’s not.
11. The level of stress that a witness has or did experience: stress has a negative impact on memory.
    

Lack of consistency in recalling the details of an event, does not mean that a witness is not credible. Research shows that the other elements recalled by witnesses could be accurate. Equally, just because someone seems super-confident in their recollections, it doesn’t mean that their memories are more accurate than someone feeling less confident.

It should also be noted that many of the above (but not number 5) are “honest memory errors”. As such, witnesses should not be berated for their mistakes in accurately recalling an event, as human beings are unable to spot what is a “real memory” and what part of the memory has been contaminated, unconsciously. And witnesses won’t recall every part of an event – they will only recall what was relevant to them at the time.

I should quickly mention that Reisberg makes a disclaimer: that despite the concerns around memory error, memories are often more accurate than they are inaccurate. So, maybe I’m over-thinking the need for a checklist?