Applies to: @RISK 7.5 and Up

Contribution to Variance in @RISK

During regression analysis, the sequential contribution to variance technique measures how much additional variance in an output is explained as inputs are added to the model.

Stepwise regression determines both which variables to include and the order of entry.

When inputs are correlated, the regression may assign variance to one variable while ignoring others, so caution is needed in interpretation.

@RISK performs stepwise regression on an output to assess sensitivity to input distributions.

Inputs enter sequentially based on their significance.

If an input loses significance after others enter, it is removed.

Next, @RISK conducts a forward regression using only the variables retained in the stepwise regression. 

• The change in R² when each input enters represents its percentage contribution to output variance.

These contributions are shown in the Contribution to Variance tornado graph or can be listed in a worksheet.

• Total contributions equal R², though the graph may show fewer inputs than the total.

For correlated variables, some may be excluded from the stepwise regression if their variance overlaps with others. 

Excluded variables have zero contribution in the forward regression.

• Which variable is excluded can vary slightly between simulations, but the overall effect on the model is minimal.