Statistical modelling to make predictions encounters various challenges, including deal- ing with model uncertainty and limited sample size. Model uncertainty arises from the fact that we usually do not fully pre-specify a model before we fit it to a data set. 69, 101 An iterative process is often followed with model checking and model modification. On the other hand, standard statistical methods assume that a model was pre-specified. In that case, parameter estimates such as regression coefficients, their corresponding standard errors, 95% confidence intervals, and p -values are largely unbiased. When the structure of a model was at least partly based on findings in the data, bias may occur, and we underestimate the uncertainty of conclusions drawn from the model.
If you want to change selection, open document below and click on "Move attachment"
- (no access) - BOOK_steyerberg_Clinical prediction models.pdf, p30
|status||not read|| ||reprioritisations|
|last reprioritisation on|| ||suggested re-reading day|
|started reading on|| ||finished reading on|