Chapter 4. Fieldwork Study One: Long-Term Use of Personal Informatics
4.3. Study Design
4.4.3. Making Sense through ‘Data-Work’
Having established ways in which our participants encountered a quantified past, I now turn to how people made sense of it in-situ, and accounted for their data in relation to broader personal narratives, and the how they remembered the past.
As previously highlighted, a specific discursive analysis was directed towards those parts of the interviews where participants were seen to very directly interact and make sense of their data. These interactions were defined by a tension between what was remembered, and what the data implied. This manifested as points of negotiation. Here participants were attempting to both communicate the meaning of their data for the interviewer, and crucially, to subjectively interpret it, in such a way as to construct a coherent narrative and achieve their own sense of verisimilitude.
There are various rhetorical means people used to achieve this: asides and embellishment; pauses and explanations; self-reflective commentary; surprise and questioning of data. Together, I describe these as ‘data-work’. By this, I mean the language and ‘work’ that is done to qualify and make sense of one’s data. In this case, the data was made accountable to the past as participants knew and remembered it, and such that it was reasonable and presentable in the context of the interview. Parallels for this are to be found in design ethnography of the local interaction surrounding photo narratives (Crabtree et al., 2004) and similar ethnographic work on the situated organizational practices surrounding ‘home-mode’ photos and videos (Kirk et al., 2006, Kirk et al., 2007). Most
contemporaneously, data-work offers another lens on Taylor’s conceptualization of ‘data- in-place’ – a description of how data is necessarily ‘entangled with wider forms of life’ (Taylor et al., 2015). While the term ‘data work’ has arisen rather briefly before (specifically on work about data infrastructure in collaborative research environments (e.g., Jackson and Baker, 2004; Karasti and Baker, 2008), it is adopted here to describe how individuals interact and make sense of personal data, in-situ.
Data-work here is hence shown to achieve contextualisation, negotiation with data, and a sense of verisimilitude. The extracts below are typical, and give a flavour of the rich and personal narratives of participants.
Data-Work as Contextualisation
In several cases, data-work took the form of contextualisation, to identify oneself within the data, and develop its potential personal significance within a broader context.
Consider these two extracts, from participants looking back at their running data. “Oh, this is funny, so this is... the day before Tim was born. His birthday is the 16th. So that's [pointing at the map] the flat that Jill and I moved to so… how funny... that's a really short route. Oh it's not that short. I kind of went down into the Dene – this bit in the middle is Jesmond Dene, and so I always try and kind of work a run through there.” (Tony, SportsTracker)
“Yeah here is just exactly when I moved to Newcastle. This is the week that I moved here. So this is the first time that I... I live really close by Leazes Park so that's why everything starts changing now, because it's in the park, I can’t run at night
anymore, because it gets weird and the birds are getting weird noises [sic] and it's really scary. But this is the first time that I ever ran in Leazes Park.” (Tanya, Nike+)
The quotations above are both about a run in the park, and yet both have special
significance for the participant, which is instantly recognized and explained. Seemingly anonymous data is marked out to ‘talk to’ – “Oh this is funny” – and elevated to
something more personal. Rhetorically, this data has been put above data about many other runs. Important details, like the date, or one’s home are highlighted. Tanya in particular emphasises the novelty of this particular run; that it was the first time she ran in Leazes Park. In both cases, the data is located temporally in terms of bigger life events: a child’s birth, or moving to the UK. Overall, in both cases, the data becomes embellished with personal commentary – about birds, places, family etc. – which goes beyond what the data alone can show. All of these are ways of personally contextualising the data, remembering and showing how the data relates to them.
Data-Work as Negotiation
Beyond simply contextualising the data, data-work was also a means of negotiation with the data. This formed an effort to resolve a tension between what the data appeared to represent, and one’s own memory and narrative. Consider two further extracts, relating to food intake on MyFitnessPal, and music listening on Last.fm.
“The 14th of September, I apparently had no tea that day as well – which I don't believe – porridge for breakfast, and more pasta for lunch and some prawn cocktail crisps, a horrible mugshot thing and some grapes and I did loads of walking, which doesn't feel like very much on that day either.” (Leanne, MFP)
“See, I would say, it's probably not that I've only listened to seven songs, I've maybe, I dunno… Or maybe I did only listen to seven songs. Or maybe I just didn't scrobble them somehow. I'm not exactly sure, but it is kind of odd because there's sort of, a consistent number of over 100 plays each week and then it is this gap.” (Darren, last.fm)
The underlined portions indicate a number of different techniques to play down the meaning of the data being discussed. In the first case, the validity of the data is undermined by conditional words like ‘apparently’ or ‘probably’, and an outright rejection of the data, which is reduced to a question of belief. Quantities are qualified to give them meaning – “loads”; “not very much”; “only”. Darren compares the questionable data to another week where there is a “consistent number” of plays, in contrast to the ‘gap’ he is now seeking to explain. The efficacy of the tracking can be brought into question - “maybe I just didn’t scrobble them somehow”. As Becky suggested, in the case of her Moves data: “Sometimes my phone died, or the battery’s gone and it turns Moves off.” These are all efforts at explanation but also ways to render the account more flexible, and to make it fit more easily within the current exposition.
Of course, what exactly has happened is often simply not clear, and some participants then sought to probe their data more deeply for an explanation. Consider another extract from Darren:
“I guess this was maybe… this was maybe when I stated getting into running… so there’s a slight chance that when I was listening to Chemical Brothers, if… I wonder if it will tell me what songs I listened to… because it’s further…. Ah ok, Escape Velocity, yeh that’s the one I was thinking of, yeh that was one of the tracks that I sort of put on to run to.” (Darren, Last.fm)
Here he has in mind a period of time when he was running, and remembers specific songs he would run to. He then browses the data around this time for these songs, to identify and temporally locate this section of his music data. Note also how these extracts could be data or narrative-led. Whereas in Leanne’s example above she is led by the data – she literally recites her data with commentary – here Darren is using the data to back up his narrative.
Data-work as verisimilitude
Darren’s example above shows the way a resolution might be reached. Overall,
participants clearly sought a sense of verisimilitude – that their account was close to real life as they experienced it; that it seemed right. This was found somewhere in between what was remembered and what was recorded, as participants sought coherence between the two. Participants like Leanne above were quick to disregard or undermine data if it was unaccountable to their own remembering. However, even when the data was perceived as inaccurate – especially where there were gaps and errors – it remained highly interpretable and could be spun into a narrative. While participants often sought and found affirmation in their data, on occasion it refined their narrative – adding specific details, or curbing inflated claims.
“I can see there... how I went from 30 minutes swimming in the morning, just a casual swim, to 60 minutes, at least forty-fi...at least 40 minutes.” (Joanne, Excel) This tension between past-as-remembered and past-as-recorded is evident, and not easily resolved.
“Because obviously, I don't take [the data] as a, you know, ‘this is what happened.' But at the same time, your memory doesn't always remember things in the correct way either.” (Becky, Moves)
Becky highlights this tension directly in this quote. It is ‘obvious’ that she interprets the data in light of her own memory of what it portrays, but she remains open to the
possibility that one’s own memory could be flawed and is hence flexible. In certain contexts, people placed more trust in the data, or their own memory. Joanne above, a fitness addict who fastidiously and actively records her activity in an Excel spreadsheet, claimed total confidence in her data and stood corrected by it. Darren suggested that “in my mind, I probably listened to as much music that week’. However, he could attempt to explain and reverse engineer perceived errors within last.fm to support his doubts. His data was still informative, but not always authoritative.
The above analysis reveals the significant work required to qualify and contextualise one’s data - the way in which “data morphs into selves” (Davis, 2013). Secondly it suggests that tensions arise in doing this, and there are seams in the data which people are able to pick at to construct a coherent narrative. An important limitation should be
sounded out here. While it did appear quite natural for people to talk, rhetorically, about their data in this way, one ought to question why or when this kind of data-work might be necessary or important. Bartlett (1932) might remind us here that literal recall is
“extraordinarily unimportant” (p.204) in the course of everyday life, in contrast to extreme circumstances, like a witness stand in a courtroom. The data-work here took place in the context of an interview. The present activity towards which participant’s remembering with data was oriented likely had a number of ends. First, they were making sense of their data for themselves; they then tried to present this in a coherent way to the interviewer. This sense making was interwoven with broader stories they had to tell: about being a ‘fitness addict’, becoming a parent, losing weight etc. Perhaps participants sought to please or impress the interviewer. This is simply to argue that just as
remembering is situated, so is data-work; and hence the data-work represented here, is representative of the particular local needs to account for the data in a particular way. Data-work in a doctor’s office or in a boardroom might well take on a different character. Such narrative work and tensions also surround retrospective interpretations of other media like photographs or social media posts. However, particularly in the context of the aforementioned ‘dataism’ (van Dijck, 2014) – the belief in the objectivity of data – there is something to be learned from the way that participants constructively and flexibly interact and remember with data. Personal informatics tools are deliberately employed to
provide ‘objective’ measurement, differing from other recording tools or historical markers, and create a historical record often as a by-product of their everyday use. Clearly, that record cannot simply be read or taken as-is.
Nonetheless, however accurately, and for whatever purpose, it is clear from the study findings that personal informatics, for these participants, did not miss the mark in capturing some essences of people’s lives. The next section of these findings explores ways in which these essences might be perceived or constructed as meaningful for participants, and of broader value in their lives.