www.phidot.org

The capability to plot the real parameter estimate as a function of individual covariate values has been implemented. An icon on the Toolbar and a menu entry in the Results Browser under Output | Specific Model Output | Individual Covariate Plot provide access to this capability. A plot of the function and 95% confidence intervals are provided, with the capability to set other individual covariates to specified values. The values can also be downloaded to Excel to produce publication quality plots, or more complex plots. See the help file "Individual Covariate Plot" for details.

Gary

gwhite wrote:The capability to plot the real parameter estimate as a function of individual covariate values has been implemented. An icon on the Toolbar and a menu entry in the Results Browser under Output | Specific Model Output | Individual Covariate Plot provide access to this capability. A plot of the function and 95% confidence intervals are provided, with the capability to set other individual covariates to specified values. The values can also be downloaded to Excel to produce publication quality plots, or more complex plots. See the help file "Individual Covariate Plot" for details.

Gary

Reminder to folks that this nifty new capability (which also exists in RMark) plots the function for a given model - it does *not* plot a model averaged function, if such a function can be generated...

There is a function in RMark

Code: Select all

covariate.predictions

that will generate model-averaged predicted values (and CIs) for a parameter across a specified range of covariate values. You can ask it to use just a particular model, but you can also have it compute model-averaged predictions across all models in a set.

See Ch 16, pg 77 of the Appendix for RMark in GIM.