MDX Query using BottomCount to limit Median calculation returns null

Question

I'm building a new cube that includes some person age values that aren't useful when summed, but are
useful when the median is determined. The measure group that contains the measure for the age has a 1-1 relationship with a dimension in the cube because both use the same table as the source. This is important because I use the key attribute of the dimension
as the set expression in the Median function to prevent any summation before the median is found. Here is the code for the original median calculation:
MEMBER Measures.[Median Age] AS
MEDIAN(
[Placement Dimension].[Id Removal Episode Fact].Members,
[Measures].[Age At Removal Months]
This median naturally represents the half-way point in the series of values. My analysts have also requested
median-type values at the quarter and three-quarter points in the same series. I've been able to accomplish this for the three-quarter point by nesting the TopCount function in the set expression of the Median function to limit the set to the last half of
the records and then find the median point like this:
MEMBER Measures.[75th Percentile] AS
MEDIAN(
TOPCOUNT(
[Placement Dimension].[Id Removal Episode Fact].MEMBERS
,Measures.[Episode Count] / 2
,Measures.[Age At Removal Months]
,Measures.[Age At Removal Months]
However, my attempt to use the BottomCount function in the same way as TopCount to find the quarter point
in the whole data set by limiting the calculation's set to the first half of the data always returns null. Here is how I've formed the code:
MEMBER Measures.[25th Percentile] AS
MEDIAN(
BOTTOMCOUNT(
[Placement Dimension].[Id Removal Episode Fact].MEMBERS
,Measures.[Episode Count] / 2
,Measures.[Age At Removal Months]
,Measures.[Age At Removal Months]
And here is the query that returns the values:
SELECT
Measures.[Episode Count]
,Measures.[Median Age]
,Measures.[25th Percentile]
,Measures.[75th Percentile]
} ON 0
,[Date Begin].[Calendar Hierarchy].Year.&[2011]:[Date Begin].[Calendar Hierarchy].Year.&[2014] ON 1
FROM [POC Cube]
WHERE
[Age at Removal Mos].[Age in Years List].[Age Year].&[0]:[Age at Removal Mos].[Age in Years List].[Age Year].&[5]
I don't know why the end result is always null. I don't have any null values in the data for this measure, and I know what values I should be seeing because I've found the median records manually in results from a SQL Server query. I've tried using TopCount
and multiplying Measures.[Age At Removal Months] in the TopCount function by -1 to workaround the descending sort, but I still get nulls. I've also tried separating these queries out so the quarter point and three-quarter point calculations aren't run together,
but I still get nulls for the quarter point calculation.
I'm open to any help fixing this situation by modifying my current code or by using alternate methods, but the end result has to be dynamic enough to be used as a calculation in the cube. Thanks!

Answer 1

The links might helps.
http://technet.microsoft.com/en-us/library/ms144864.aspx
http://www.mssqltips.com/sqlservertip/3034/sql-server-analysis-services-ssas-2012-top-and-bottom-functions/
http://www.sqlservercentral.com/blogs/bradleyschacht/2012/03/12/mdx-functions-bottomcount/