REASON

Legal Requirements for Statistical Modelling

Funding body: Province of Styria (Land Steiermark)

The interdisciplinary project “REASON – Legal Requirements for Statistical Modelling” traces the extent to which political decision-makers based the measures they took during the COVID-19 crisis on statistical modelling. Building on this, REASON examined which legal requirements for such modelling can be derived from the democratic, rule-of-law and liberal fundamental principles of the Constitution. The project’s results culminate in a legal policy proposal that is usable in practice and successfully reconciles the innovative potential of statistical modelling with the requirements of a liberal state governed by the rule of law.

Statistical models depict partial aspects of reality in simplified form in a mathematical equation. This makes complex relationships observable and makes forecasts possible. Such statistical models and the forecasts based on them therefore play a central role in, among other things, combating a pandemic. However, the quality and informative value of such models and forecasts always depend on the data, assumptions and methods on which they are based.

If these data, assumptions and methods, or any shortcomings of the models, are not transparent, however, democratically legitimated bodies may be unable to make an informed decision: the line between undisputable facts and politically relevant value judgements is already blurred before the modelling documentation reaches the decision-maker’s desk.

The question therefore arises which criteria such models would have to meet in order to satisfy constitutional requirements. To answer this question, REASON first analysed press conferences held at the beginning of the COVID-19 crisis. The analysis aims to show whether and how political decision-makers addressed the information deficit that existed at the start of the pandemic. In a further step, interviews were conducted with experts who were directly involved in decision-making, for example as members of the Coronavirus Task Force. This made it possible to establish what role statistical modelling actually played in decision-making.

In parallel, a comprehensive analysis of the relevant legislation was carried out. This identified existing provisions on statistical modelling in legislation and revealed possible gaps.

The project was completed on 31 December 2021; the final report will be published in the summer semester of 2022.

Interim report (PDF)

Participating institutions:

University of Graz

Graz University of Technology