MMARS *always* reports the grade level of the students as their grade level when the test was administered -- not their rostered grade level. This is because the data and scores represent a grade-level test. Example:
* When your 5th grade students took the CELDT test in 2013-14, they were attending their elementary school. Per your (new) 2014-15 rosters, those students have matriculated to the middle school. Of course, they are now 6th graders, at the middle school, but their DATA is from a 5th grade test. Therefore, their 2013-14 CELDT results will be reported under "Grade 5", shown in the new 2014-15 rosters at the middle school, and shown under the 2013-14 rosters, as Grade 5 students, at the elementary school where they took the test.
* Likewise, When your 8th grade students took the CELDT test in 2013-14, they were attending the middle school. These students do not exist in your (new) 2014-15 rosters, because they have matriculated to a high school, and are no longer in your district. These students cannot be reported with your new 2014-15 rosters, but are shown under the 2013-14 rosters, as Grade 8 students, at the middle school where they took the test.
If you're reporting by classroom, you must select the appropriate "customer-provided" roster. If you simply want the results by school, you can always use the "Per State Test File" roster -- which is exactly how the data was organized when downloaded from the publisher, without any added local knowledge about each student's school and classroom(s).
The important point here is that MMARS results will never show the students "current" grade, when the test was taken in a prior year. The grade-level always represents the grade-level when the test was taken. I realize this can be confusing, from either perspective. We have chosen this paradigm, because the representation of the grade-level DATA will always be accurate.