Every spring, New Jersey districts receive NJSLA subclaim data alongside the overall performance levels: Reading broken into Literary Text, Informational Text, and Vocabulary; Writing into Written Expression and Knowledge of Language and Conventions; Math into Major Content, Additional and Supporting Content, Reasoning, and Modeling.
Most districts review these categories once and then move on to the overall performance level for reporting and board presentations. That is a reasonable response to a real constraint, not a failure of diligence. The overall level is a single, well-defined number. Subclaim data is harder to summarize, and it rarely arrives in a form that's ready for instructional use.
Why Subclaim Data Gets Set Aside
Overall performance levels compress a test into five ordered categories, which makes them straightforward to chart and present. Subclaim scores don't compress the same way. Cambium, the vendor administering New Jersey's assessments since the NJSLA-Adaptive transition, reports subclaims as categorical flags: below expectations, near expectations, above expectations, rather than scale scores, because each subclaim is measured by only a handful of items. There is no scale score to average, and the difference between "below" and "near" doesn't correspond to a fixed numeric gap.
Treating these categories as if they were interval data produces a specific, avoidable error. Averaging "below" (1) and "near" (2) into 1.5 assigns a level of precision, and a value, that doesn't correspond to anything the assessment actually measured. Subclaim data is ordinal. Reporting it as continuous data shows up in board presentations more often than it should.
The Limits of Subclaim Data at the Individual Level
A subclaim built on eight to twelve items carries a wide Conditional Standard Error of Measurement (CSEM) at the individual student level. In practice, this means a single student's subclaim category can shift from one administration to the next without any meaningful change in that student's underlying skill. A "significantly below expectations" flag on one test date and a "near expectations" flag on the next are both plausible outcomes for a student whose actual performance hasn't changed much at all.
That doesn't make subclaim data useless. It makes it unreliable as a diagnostic tool for one student and more reliable as a description of a group. Aggregated across a grade level of thirty or more students, the measurement noise from individual items tends to cancel out. As a hypothetical illustration: if a large majority of students in a grade, not a handful, land in "significantly below expectations" on Reasoning, that pattern reflects something real about instruction rather than test noise, even though no single student's flag in isolation would support that conclusion.
The pattern still needs to be described precisely. Comparing this year's fourth grade to last year's fourth grade measures cohort drift, since it's a different group of students. Longitudinal growth requires tracking the same students across years. Districts sometimes present cohort comparisons as if they were growth data, and the distinction changes what conclusions the data actually supports.
Building Instructional Groups from Subclaim Data
Start with the subclaim category itself, not a derived number, matched to the class roster for one grade and one subclaim at a time. Sort students into three groups: below, near, and above expectations. The result is a grouping list grounded in an actual assessed item set, rather than recall of which students seemed behind earlier in the year.
Build the grouping at the classroom level, where it can inform intervention scheduling directly. A district-level subclaim summary is appropriate for a board presentation, where the audience needs a general pattern, but it won't tell a teacher which students in a given class need support on Informational Text. Build it per subclaim rather than by overall performance level as well: two students at the same overall level can land in different subclaim categories, and grouping only by overall level puts students with different instructional needs into the same intervention block.
Where the Workflow Breaks Down
Matching subclaim data to a current roster depends on accurate student identification. NJSMART assigns each student a ten-digit State ID (SID) intended to follow the student across districts, but a summer transfer that hasn't been resolved by the fall enrollment snapshot can still require a manual match. Students with 504 accommodations, testing refusals, or absences during part of the administration window further reduce how completely "every student in the grade" maps onto the data file. None of this invalidates the approach. It means the resulting groups should be treated as a starting point that a teacher can adjust based on direct knowledge of the students involved.
Subclaim-level grouping lists that identify individual students are personally identifiable education records under FERPA, not aggregate reporting. Distribution should be limited to the teacher responsible for that roster, not routed through a shared folder with broad access.
A Practical Starting Point
If last spring's subclaim file is still sitting unused, test this approach on one grade and one subclaim before considering a broader rollout. Compare the resulting grouping list against a teacher's own assessment of student needs. That comparison, more than the grouping list itself, indicates whether the subclaim data is adding information the teacher didn't already have.