mmatzek,

I guess it depends. Do you have a single dimenion that has both boards and components? If so, then the granularity is probably components but the two fact tables tie to it at different levels of granularity, correct? If you then are at the board level, you should be able to do the calculation. If you then drop to the component level, the value would disappear.

This woudl be similar to the Adventure Works example with one of the Sales fact tables and the Sales Targets fact table. The Sales Targets has the Time dimension but only goes down to the quarter level. So if you were viewing facts from the Sales Summary measure group and the Sales Target at the same time, you'd have values in each at the Year and Quarter levels. Drop to months or days, and the Sales Target values would disapper. So you can create a "% of Target" measure that is Sales Target / Actual Sales, but it only has a value at the Year and Quarter levels.

Does that make sense?

--Craig

Here's the general idea:  A board is run through inspection and each component on the board is checked for defects.  The board itself is judged as Pass or Fail and this result is stored in a table at board level.  If the board fails (any component fails) then each component failure (call) is recorded in a table at the component level.  There is a different program that runs on the inspection machines for each distinct product.  So my first dimension is a program dimension that has a Program --> Component hierarchy.  I also have other dimensions: Time, InspectionResult (Pass / Fail) for the board, Machine, Fault Mode (detail of component defect).   If my measure is the # calls (1 for each record), then if I added the board ID to the measures table, is there a way to somehow calculate the [# of calls]/[DISTINCT # of board ID's]?  Or is there a much better way to acheive the calls per board calculation?