What is it about?

Sensitivity analysis provides information on the relative importance of model input parameters and assumptions. It is distinct from uncertainty analysis, which addresses the question ‘How uncertain is the prediction?’ By reviewing about 300 papers mentioning modelling and ensitivity analysis we show that most modelling papers report a sensitivity analysis which is erroneous. Sensitivity analysis in mathematical modelling has many possible purposes, inter alia that of ensuring that the model behaves as specified. We show that many sensitivity analysis are 'perfunctory', i.e. they claim to explore how the model changes in relation to changes in its inputs but in fact remain blind to what the model is actually doing.

Featured Image

Why is it important?

Testing how the model behaves in relation to changes of its input is a due diligence obligation for modellers. Failing to do so may leave coding errors undiscovered, uncertainties underestimated, and jeopardize model calibration.

Perspectives

In our opinion, the problem with sensitivity analysis is partly attributable to the fact that mathematical modelling is not a discipline in its own right, and every branch of science and technology approaches modelling following its own culture and practice. Elsewhere (https://arxiv.org/abs/1712.06457) we suggest that one solution would be to put statistical modelling under the roof of the discipline of statistics, at least as far as the methods for model quality control are concerned.

Professor Andrea Saltelli
University Pompeo Fabra, Barcelona School of Management

Read the Original

This page is a summary of: Why so many published sensitivity analyses are false: A systematic review of sensitivity analysis practices, Environmental Modelling & Software, April 2019, Elsevier,
DOI: 10.1016/j.envsoft.2019.01.012.
You can read the full text:

Read

Resources

Contributors

The following have contributed to this page