The spread of ‘business intelligence’ (BI) tools and predictive software is not unfamiliar to those in policing. With shrinking budgets and increasing pressure to demonstrate ‘effective/efficient’ performance, it is not a surprise Police Services are drawn to implement technologies claiming to facilitate “agile, fact-based decision making” with “24/7, user-friendly data access to make faster, more informed decisions”. The promotion of quantitative data as capable of generating, single truth, actionable information is, at the core, the “mythology” of big data: belief that large data sets offer a higher form of knowledge that produces insights/predictions that were previously impossible. Armed with this mythological belief, there is a significant motivation toward using quantitative data as the only legitimate tool to ‘prove’ effectiveness, while abandoning qualitative data as ‘just anecdote’. The ‘Achilles heel’ in this approach is that the human context is largely or entirely missing from the BI systems being used, creating a crucial disconnect in data-driven decision making.
Organizations envision that predictive use of data will enable them to move away from the reactive, and not only predict things that are going to happen, but actually intervene before they do. In the vein of ‘information is power’, BI systems gather, cleanse, link, and analyse large data sets and organize into user friendly dashboards which link users with a vast amount of aggregate data and access to individual ‘drill downs’. Clicking a button on the Crime Analysts Toolbox, or the S/Sgt’s Monthly Stat Generator is intended to remove anonymity of big data and ultimately represent the human(s) metaphorically beneath the data. However, to act on aggregate data views on behalf of individuals requires a balancing act capitalizing on the predictive capabilities of BI tools, while also respecting the lived experiences of the humans ‘beneath’. Yet, the drill down data, believed to capture human context, is largely or entirely missing. It is frequently not even found in digital form. This is the case when senior administrators are counting officers’ sick days. BI will provide the aggregate, e.g. Officer X missed 45 hours in the last 30 days and location Z has had 13 entries, on M and W between 12 and 1:30. However, BI cannot generate the qualitative reasons for why Officer X was absent, i.e., prolonged illness, job related stress, personal challenges, etc. Similarly, when the break and enter team is looking for the number of recent entries in a given location, BI will not provide an understanding of the fact that location Z actually has geographical/environmental factors that increase the risk of break and enters. Such disconnects can lead to ineffective personnel management or costly allocation of investigative resources.
The relationship between being data driven and being actionable depends on interplay between quantitative and qualitative data. While efficiencies and effectiveness of the organization are supported by aggregate views of quantitative data, being actionable is supported by individual level, qualitative data, important for responding to the unique circumstances of individuals. To avoid disconnects between the human context and data driven decisions will require BI tools, and the police services that use them, to re-envision how to collect and interpret robust quantitative and qualitative data so that subsequent action is truly representative of the humans beneath the data.