Why data need ethics / The more data affect people, the more ethics matter

Tracing ethical issues in data projects with the Data Ethics Decision Aid (DEDA)

Data, algorithms and models are not simply neutral. On the contrary, they can constitute significant social impact and affect the individual livelihood as much as larger parts of society. Recently the Unemployment Agency of Michigan has been criticized for using a software application that wrongfully identified users as illegal recipients of social welfare. And in the Netherlands the press disclosed that datasets representing 11 million taxpayers were used for finding novel ways of analysis. The problem: access to the data was not monitored and the use was not regulated or controlled. The tax revenue service cannot confirm that data has not been stolen, or used in a way that breaches privacy regulations. Residents of cities in the Netherlands were alarmed when they learned that municipalities allowed corporations to use the wifi signals of their mobile phones to analyze how people move in urban public space.

The unprecedented access to data resources and the application of novel means of analysis produces valuable insights for improving public administration and corporate management. However, collecting data, applying models and developing algorithms for analysis and decision-making, raise questions concerning compliance and accountability. Especially public management is under scrutiny of critical citizens and journalists and subject of political deliberation. Careless use of data and its applications can cause public outrage.

Often data practices are developed with best intentions and the best service for residents in mind. However, often the social impact is difficult to foresee, unintended consequences are overlooked and risks are not identified and hence not limited. The law regulates some aspects, such as privacy and data loss. But in many instances the law simply does not apply. This is where stakeholders have to think of ethics, commonly shared values and institutional codes of conduct.

The Data School developed a process to trace such ethical issues already in early stages of a project. Developed in close cooperation with data analysts from the City of Utrecht (NL), the Data Ethics Decision Aid (DEDA) allows the various participants of a data project to reflect on their case and the possible implications of their actions. It addresses questions concerning data collection, data management, bias, algorithms and models in order to reflect on how they are used and which impact they might constitute on the analysis process itself and on the targeted populations when applied. An introductory workshop guides the participants through the various stages of a data analysis project and the various ethical issues that need to be considered. Implementing DEDA into the work process will then raise awareness for ethical issues and stimulate participants in data analysis projects to address responsibilities. An online questionnaire documents the deliberation process and serves as documentation for future projects reviews.