Statistics when your models are wrong
How can we do more with less data?
- Peter Grünwald
- NWO Vici 2009
In collaboration with Centrum Wiskunde en Informatica (CWI) .
Scientists make considerable use of practically worthwhile but clearly errant models. For example, non-linear relationships are modelled as linear ones or dependent variables as independent ones, such as in DNA sequence analysis. However, existing statistical methods assume that the models used are correct. The researchers shall develop new methods that do not make this assumption. This will enable us to do more with less data.