Introduction

New opportunities and technologies for large-scale data linkage and datamining have rendered databases one of the key resources for many areas of public life. In fields as varied as medical research, marketing and government, databases have become indispensible resources for generating knowledge and innovation. They are sources of great social benefit, and also of concern. Ethical and regulatory challenges arise from the virtual ubiquity of databases and the increasing digitalisation of datasets, and also from increasing possibilities to link and search digital datasets (e.g. Mittelstadt and Floridi 2015; NCoB 2015; Pasquale 2015; Prainsack 2015a). Moreover, many of the datasets established and used today lend themselves to both aggregation and ‘drilling down’, namely the automated searching for information throughout different hierarchies of records or files. This means that it has become much easier to go back and forth between the individual and aggregate level. As sociologist Lawrence Busch (2014) points out,

Because different datasets can be linked together much more easily, it is becoming increasingly common for insights obtained at the aggregate level to be used to make probabilistic ‘predictions’ about individuals. An example for this is the practice of consumer scoring (Dixon and Gellman 2014). Here, perfectly innocuous looking pieces of information about a person – such as the frequency of her grocery purchases, for example, the length of time she has been with her current employer or the type of car she drives – are used to assign a score to a person based on the statistical correlation of these pieces of information with relevant behaviours or characteristics, such as defaulting on a mortgage, suffering from a chronic illness or causing a car accident.