One of the research methods we will be using in the DSS project is data ethnography. In a basic sense, this involves doing a lot of ‘hanging around’ the digital data that we find in each school. This will involve sifting through software code and data logs, looking at system architectures and mapping data infrastructures. This will also involve engaging in a variety of ‘analog’ ethnographic techniques such as interviews, observations, extended field notes, and document and policy analyses to gain a detailed, ‘thick’ sense of how people encounter and experience digital data. In this sense, we aim to develop ‘’thick descriptions’ of the institutional processes and individual practices surrounding data use in each school.
However, digital data do not ‘speak’ for themselves. Instead, the ways in which data are used to ‘say things’ are culturally and institutionally framed, and often rely upon older tropes and associations. We are particularly interested in the ways in which the data-driven narratives that we might find in our schools are in tension with cultural understandings and institutional conventions. In this sense, we need to pay particular attention to the role of human decision-making in resolving these tensions.
In this post I explore three aspects of datafication to help us think through the different aspects of our data ethnography: data points, data flows and data narratives. Using these three concepts will help us to frame some of the key questions that need to be asked about data in schools.
So let’s begin with data points. Data points are discrete ‘things’ (e.g. elements, instances, moments or facts) that can be extracted from online and offline life and then represented as data. So if we think about data in schools, all the following things can be reducedto a data point: a test result; a child; a sibling; a parent; a school yard fight; a child arriving late to school; a student’s home address; and a username.
However, as soon as any element of ‘real-life’ becomes encoded as a data point it is decontextualized from the time and space in which it originally existed. Moreover, it isdecontextualized from its original social and historical context. In the case of school data, this might include the stripping away of any contextual understanding of the school and neighbourhood, the individual student’s emotional state and/or family circumstances. Key questions to ask here, then, include issues of representation and reduction. For example, what is marginalised (and what is amplified) in the translation of social elements into data points? What elements of school and schooling cannot be translated satisfactorily into data form at all?
These individual data points then become mobilised through data flows and drawn into an array of institutional and corporate enclosures in ways that are often unknown to the individual or object that generated it. An individual data point joins various data flows and therefore becomes abstractedinto an economy of data points. A key question is how we can trace and make sense of data flows in and around schools.One goal of our research then, might be to ‘consider the breadth of the heterogeneous ensemble through which [data] flows’ (Cote, 2014, p.135). This will help us to identify how human decision-making shapes these flows as well as developing a sense of moments in this process at which data might be liberated.
In this way, making sense of data points and data flows requires a kind of storytelling, or what Dourish and Cruz (2018) call narration. On one hand, data prompt individuals and institutions to generate new narratives. Yet on the other hand, it is important to acknowledge the ways in which data are also incorporated into old narratives. While we are often told that data are ‘objective’ and ‘neutral’, Dourish and Cruz (2018) demonstrate how data can be used to bolster existing stories, beliefs or opinions rather than actually bringing about new insights.
Dourish argues that the progression from data point to data narrative needs to be understood in terms of two scalar moves. The first move is the creation of data sets – thereby involving a move from small to massive (i.e. massification). This move is based on the assumption that a combined mass of data points are ‘equivalent’ and representative of a single comparable phenomenon. The second move is then back down from massive to small. This move is based on the assumption that features in the data ‘correspond’ to a specific feature in the world: i.e. “an item of interest in the domain about which the data ‘speak’” (Dourish & Cruz, 2018, p.2). In many respects, both of these moves involve the construction of data narratives.
Indeed, every stage of this process of scaling-up and (re)scaling-down involves crucial decisions being made (often by humans) that effect how people narrate and are narrated by data. In this sense, errors and discrimination can occur at many different points throughout the datafication process. For example, disparate data points can be assumed to be equivalent even though they are not. Data can be judged to relate to a phenomenon in the world even when it does not. Alternatively, extraneous data points can be drawn into narratives unnecessarily. Ultimately, multiple actors are engaged in these narrative acts, but these are different acts for different purposes.
In short, the idea of data narratives draws attention to the host of assumptions, judgements, opinions and decisions that lie behind the datafication process. Digital data do not naturally or spontaneously appear fully formed. Data analyses are not objective calculations with a ‘true’ result or unambiguous answer. Instead, digital data are written and narrated into existence by a variety of different authors. For us, this raises a host of critical questions to consider. Here are a few issues that our research needs to grapple with:
- How are data points created in schools? What apps, digital tools, platforms, sensors, cameras are in schools and who is responsible for checking the Terms and Conditions of these? Dept of Ed? Teachers? Principals? Parents?
- What happens to data that is collected in schools? Can we ever know the ways in which it is reused, particularly in regard to commercial apps and digital tools in schools?
- Can students really opt out of digital tools and educational platforms if they are integral to the curriculum? How can students opt out of facial recognition technology in schools if the camera has to register that student in order for them to be to opted out? What happens to that data?
- When it comes to making sense of data, who has the power to write the narrative? And who or how are these narratives answered? Big data sets are dependent on the logic of equivalence – that is data need to be considered similar enough to sit within the same data set. But who decides whether this data is equivalent, how it should be categorised and what should it be deemed correspondent to?
- How does datafication change the processes of learning? Data-driven epistemologies lead to standardisation, pattern recognition and predictive analytics. What does this mean for how teachers ‘see’ their students and assess their learning?
These questions will be difficult to answer conclusively. However, not only are these key guiding questions for our research, but most importantly, they highlight issues that have profound consequences for young people’s lives. It is therefore essential that schools and educational authorities are also aware of these issues, so they might begin to develop and implement ethical data management strategies.
Cote, M. (2014). Data motility: The materiality of big social data. Cultural Studies Review, 20(1), 121-149.
Dourish, P., & Cruz, E. (2018). Datafication and datafiction: Narrating data and narrating with data. Big Data & Society, July – December, 1-10.