I have a measure that counts the rows of a table ignoring the filtering on the variable post-secondary enrollment:

count_ignore_ps = CALCULATE(COUNTROWS(highschool_recent_year), REMOVEFILTERS(highschool_recent_year[ps_enr]))

Here is it used in a clustered column chart as the y axis. With the x axis being school and post-secondary enrollment. When used in this way the measure should be the total number of rows for the school. This works as intended on the left graph but does not on the right graph.

The left graph has filters:

• Race 1 or Race 2 or Race 3
• no filter on school

The right graph has filters:

• Race 1 or Race 2 or Race 3
• School A or School B or School C

There are only three schools so no filter on school and the union of all the schools should be the same. The measure is producing different result.

I think this is because there are 10 race 1 students who are enrolled and 0 race 1 students not enrolled in school A. So for some reason the measure is not counting those students in the other condition.

Does anyone know what could be causing this issue or a solution?

Here is a link to the toy data: https://drive.google.com/file/d/143r4Nlmz819kdvaBCz6KFdkxAVP0R5ov/view?usp=sharing