Distributions
The Distributions section of the Model report contains four subsections: Outcome by Treatment, Calculated (Stabilized) Weight by Treatment, Treatment Distribution, and Residual Distribution.
Outcome by Treatment
When the treatment is binary and the outcome is continuous, the Outcome by Treatment section contains a stratified box plot, which shows the distribution of the outcome for both levels of treatment. When the treatment is binary and the outcome is binary, the Outcome by Treatment section contains a mosaic plot that shows the number of observations that fall into each of the four possible categories (treatment is 0 or 1 and outcome is 0 or 1).
When the treatment is continuous, the Outcome by Treatment section contains a scatter plot, with the treatment on the horizontal axis and the outcome on the vertical axis.
Calculated (Stabilized) Weight by Treatment
Note: This section appears only if a treatment model is specified.
When the treatment is binary, the Calculated Weight by Treatment section contains a stratified box plot, which shows the distribution of the inverse probability weights for both levels of treatment. If the Stabilized Weight check box is selected in the Model Launch control panel, the corresponding Model report produces this plot so that the vertical axis represents the stabilized weights. If the check box is not selected, the corresponding Model report produces this plot so that the vertical axis represents standard, non-stabilized inverse probability weights.
When the treatment is continuous, the Calculated Stabilized Weight by Treatment section contains a scatter plot, with the treatment on the horizontal axis and the stabilized weights on the vertical axis. Weights are automatically stabilized when the treatment is continuous.
Treatment Distribution
When the treatment is binary the Treatment Distribution section contains a table that shows how many observations fall into each treatment level. If the Show Tips and Interpretations option is checked in the Fit Causal Treatment red triangle menu, an alert message appears under the table if there are any missing values in the data table.
When the treatment is continuous, the Treatment Distribution Section contains a Treatment Histogram, which shows the number of observations falling into certain intervals of the treatment variable. The Treatment Histogram also displays the standard deviation of the distribution in the top right corner.
Residual Distribution
When the treatment is binary, the Residual Distribution section contains two Regression Adjustment Model Residuals scatter plots. The plot on the left shows the Target Level information and the plot on the right shows the Control Level information. These plots show the residuals of the adjusted regression model. The horizontal axis represents the predicted outcome under the given treatment level and the vertical axis represents the residual value. Although only one level of the treatment is observed for each row in the data table, the adjusted regression is used to predict the outcome for each observation for each treatment level. Therefore, the number of points on each scatter plot corresponds to the number of observations in the entire sample.
When the treatment is continuous, the Residual Distribution section contains an IPW Response Model Residuals scatter plot. This is the predicted versus residuals plot for the marginal structural model with IPWR. The horizontal axis represents the predicted outcome under the given treatment level and the vertical axis represents the residual value.