there is some solutions regarding what are your data and how you want to
aggregate your data.

Classic example:
Car 1 --> n Property n <-- 1 Sales
(1 car can have 1 or more property)
In this type of situation, you don't want to have 10 * sales if your car has
10 properties.

Parent-child dimension with data for non-leaf members is 1 option
2 cubes (1 for the detail and 1 for the aggregate) merged into 1 virtual
could also solve the problem.

but, can you explain what are your expected results?

[quoted text, click to view]