Standard errors of the estimates

Purpose

This plot displays a histogram with the relative standard error of each parameter estimate, computed with the Fisher Information Matrix. It is only available if the standard errors were previously computed, and it provides a graphical representation of the information already available in the column “R.S.E (%)” in the Pop.Param tab of the Results frame.

Example

In the following example, the parameters of a one-compartment model with first-order absorption, linear elimination and a delay areĀ estimated on the warfarin data set. The figure shows the relative standard errors of all population estimates, computed via linearization and importance sampling methods, with the exact values written in front of each bin.
This figure allows here to highlight beta_Cl_tsex_F as an estimate associated with a high standard error. This suggests to check the relevance of this covariate effect, by looking at the estimate for this parameter and the result of the Wald test in the statistical tests.

Settings

  • General. Add/remove the legend or grid.
  • Methods. Add/remove histograms bins for values computed by linearization or stochastic approximation, if they have been computed.