Calculated Risks: The Teacher as Big Data Producer and Risk Analyst

Abstract

The availability of data sets and development of complicated analytics is changing the decision-making processes in a multitude of disciplines. Purely data-driven approaches have become very successful in executing mechanistic processes, and offer valuable information in more complex, human-driven situations. (Spiegelhalter, 2014). The influence of big data and the decisions influenced by its analysis cannot be ignored in education. Teachers feel this impact in a number of ways, but none more immediate than third-party, large-scale examinations designed to satisfy an unending thirst for accountability and surveillance. While these examinations do collect data in a narrow sense (a detailed discussion of their validity is outside the scope of this discussion), they cannot compete with teachers as data producers and analysts of complex data. Here, I explore a shift of lens that frames teachers as the driving force in the collection and inquiry of big data rather than the inert consumers of third-party data. With a specific focus on mathematics education, my intent is to highlight two important roles of teachers in the emerging age of big data. A brief look at the existing literature on risk analysis highlights the importance of context and the existence of multiple rationalities in risk-based reasoning. This places teachers as the quintessential producer of big data and analyst of the pedagogical risks therein.Teacher as Big Data ProducerThe role of teacher is one steeped in context. If one were to frame the profession of teaching as a research study aimed at gathering data and testing various parameters, it would look much different than the data that institutions collect today. The multiple interacting forces of the classroom make it difficult to isolate variables and detennine causality. Even when testing claims to decipher trends, they must do so in a contextual vacuum, because any deviance from the typical student in the typical classroom in the typical school (the generalization continues) is weakened by specificity. The result is a tendency to whitewash students into categories with those falling outside tidy lines labeled, rather fittingly, as 'at risk'. Large-scale testing, such as state or provincial math efficiency examinations, produces large n, p data (Spiegelhalter, 2014, p. 264). The participants (n), although not always willing, are many while the parameters to be measured (p) are in number. In the case of mathematics achievement tests, the participants might include an entire age group while the subject of the testing would include isolated curricular concepts. The aim of such examinations is to produce a snapshot in time of the understanding of a specific subset of individuals. This snapshot is purposely contextless, and collapses the incredibly complex world in which teaching and learning occur. Student history, teacher style, school culture, and divisional priorities are necessarily ignored. The contextual void eludes to a quantitative objectivism in the subsequent risk decisions to be made based on the data.The recent research on risk shows that risk decisions are based on a gamut of information and rationalities that involve tacit and implicit knowledge (Pratt, Ainley, Kent, Levinson, Yogui, & Kapadia, 2011). This stands in stark contrast to data collected with the purposeful eradication of context. In keeping with Spiegelhalter's (2014) definitions of statistical problems, education can be termed a small n, p enterprise where teachers encounter pockets of students (ri) and collect vast amounts of tacit and explicit data (p) (p. 264). This data is linked to the context from which it emerges, but this is no longer considered a deficiency. Teachers encounter real problems that are extremely complex in their context-dependence, and generally, dependent in reflexive ways on the subjective perceptions of different participant groups (Pratt et al., 2011, p. …