Model based graphics

The data based graphics include the

• Equations
• Residuals
• Confidence intervals

Click here to download the data and model setup (.mum) file for this example.

Fitting a Poisson model to the data

In this example, only the Configuration, Variables and Advanced tabs are used.

Configuration tab:

Variables tab:

Advanced tab:

Select Analysis, Run from the main menu bar.

The fixed part of the output is shown below.

This is followed by the random part of the output (3 random effects).

Model-based graphs

Equations

If the analysis has been completed successfully, select the Model-based graphs, Equations option from the File menu.

If Drug is used as a "group" variable, a graph is created for each level of Drug. By choosing Drug as the "Mark" variable, overlay graphs for each level of drug are obtained.

The figure below shows that the intercepts and slopes vary substantially across drug types.

Residuals

Select the Residuals... option to create a residual plot.

The residual plot indicates a number of outliers.

By clicking in the graphing area, a spreadsheet containing the observed, fitted and residual values may be displayed.

Confidence intervals

In the dialog box shown below, the Empirical Bayes estimates for the intercepts, marked by Drug, are requested.

The estimates and confidence intervals demonstrate a wide variability across the level-2 (Drug) units.

Data used to make the confidence interval graph is obtained by clicking in the graph area.