In her dissertation ”Robust analysis of uncertainty in scientific assessments”, doctoral student Ivette Raices Cruz, who recently defended her dissertation at Center for Climate and Environmental Science (CEC), has investigated methods for expressing uncertainty in scientific assessments. Analysis of uncertainty aims to assess the importance of uncertainty in a conclusion caused by lack of knowledge. In her dissertation, Ivette Raices Cruz describes how robust Bayesian analysis can be used to combine statistical analysis and quantitative uncertainty analysis. She demonstrates the method by showing how it is applied to examples of three existing, as well as one new, analysis.
How to quantify uncertainty in a conclusion
In one of the studies, Ivette Raices Cruz applied the method to an already published risk assessment regarding children's intake of aluminum via chocolate. She showed how to quantify uncertainty in a conclusion of an assessment with probability, either precise or with intervals on the probabilities. In this case, the analysis confirmed the previous conclusion of the risk assessment, that given the information in the assessment, it is safe for children to eat chocolate. However, Ivette Raices Cruz demonstrates that the choices you make when you specify a model can have a great impact on the conclusion. To demonstrate the use of robust Bayesian analysis in evidence-based decision making, she applied the method and quantified uncertainty in various types of meta-analyses.
"Scientific assessments rely on scientific methods and it is therefore important that you also handle uncertainty in a systematic and clear way that can be defended. If the assessments form the basis for important decisions, it is especially important to understand the importance of uncertainty", she says and continues:
"Uncertainty is not something we learn to communicate when we become researchers, but this type of research can hopefully contribute to it becoming part of the scientific method. If you look, for example, at the IPCC climate report or Cochrane's evidence syntheses, they have established ways of dealing with uncertainty in conclusions."
Hopes that the method will be more recognised
According to Ivette Raices Cruz, not all scientific studies aim to express uncertainty, but she hopes that her research can help make the quantification of uncertainty more scientific. She also hopes that the method she has used will be more recognised and that it will be addressed in organisations that develop principles for making scientific assessments when quantifying uncertainty.
"Uncertainty is usually quite unknown to many. It creates some misunderstandings and one such is that probability precludes uncertainty. It is difficult to avoid uncertainty, even as a researcher. The philosopher Voltaire once said that 'uncertainty is an uncomfortable position, but certainty is an absurd one'."