The Data Wall That Doesn't Move Instruction
Your school requires all teachers to post a 'data wall' showing student scores on the fall interim assessment, color-coded by proficiency band. You teach two sections of Algebra I. Section 1 has 30 students: 8 in 'below basic,' 14 'basic,' 8 'proficient.' Section 2 has 30 students: 6 'below basic,' 10 'basic,' 14 'proficient.' The interim test had 40 items, mostly procedural, with only 4 items assessing modeling. Your principal wants to see 'growth' on the spring interim, which uses the same blueprint. You have noticed that three students in Section 1 who scored 'below basic' routinely give strong verbal reasoning in class but freeze on timed tests.
1. Critique the interim assessment as a measure of what you actually teach, and identify what the data wall can and cannot tell you.
Read the worked answer
What the interim measures: With 36 of 40 items procedural and 4 modeling items, the test is heavily weighted toward procedural fluency. If my instruction targets modeling and reasoning (as the standards do), the test is a misaligned proxy - it will understate students who reason well but compute slowly, and overstate students who compute fluently but can't model.
What the data wall can tell me:
- Broad band placement on procedural skills - useful as a screening signal.
- Section-level differences in procedural fluency (Section 2 is stronger here).
What it cannot tell me:
- Whether students can model - 4 items is too few to support any inference about a student, let alone a subgroup. With 4 items, a single item is 25% of the modeling score.
- Whether the three verbal-reasoning students in Section 1 are actually struggling or are test-format casualties. The wall collapses them into 'below basic' and hides the distinction.
- Anything causal about my instruction. A color change from fall to spring could reflect test familiarity, timing, or item exposure - none of which is learning.
The honest summary: The wall is a screening tool for procedural fluency. Treating it as an evaluation of instruction or as a full picture of student learning is a category error, and I should say so - diplomatically - when the principal asks about growth.
2. Design a plan to gather better evidence on the three verbal-reasoning students and on modeling more broadly, without adding to your grading load.
Read the worked answer
The constraint is no added grading load, so the evidence has to come from work students already do.
For the three students:
- Structured oral defense (embedded, not extra): During the next modeling task, pull each of the three for a 2-minute 'convince me' conversation while the class works. Use a simple 3-point rubric (explains the model, justifies a choice, connects to context) recorded on a half-sheet. This is formative, not graded, and takes 6 minutes total.
- Format comparison: Give one modeling problem twice - once timed, once untimed - as a warm-up and exit ticket on different days. Compare. If untimed performance is markedly better, I have evidence of a format effect, not a knowledge gap. This uses existing warm-up/exit-ticket slots, so no new grading.
For modeling broadly:
- Re-purpose existing assessments: Add a single modeling item to the next unit test and score it with a 4-point rubric that already exists in the department. One item, one rubric - minimal load.
- Sample, don't grade everything: Collect modeling work from a rotating third of each section per week. Over three weeks I have full coverage with a third of the grading.
- Student self-coding: Have students label their own work with the modeling cycle (identify, formulate, compute, interpret, validate). I spot-check the labels rather than re-reading every solution.
How it feeds back: The oral-defense notes and the timed/untimed comparison give me a defensible answer to 'are these three struggling or mis-measured?' - which the data wall cannot provide.
3. The principal says, 'Just show me growth on the spring interim.' How do you respond while maintaining a productive relationship?
Read the worked answer
I'd agree with the goal and reframe the evidence, not resist the request.
What I'd say, in substance: 'I want to show growth too, and I'll bring the spring interim data. I also want to bring two other pieces so the picture is honest, because the interim only has 4 modeling items - if I report only that, we'll be making decisions on a narrow measure.'
What I'd bring to the meeting:
- The spring interim data - I'm not withholding it.
- A modeling-specific measure - the unit-test modeling item and the rotating-sample data, reported at the standard level, not as a single score.
- The three students' timed/untimed comparison - a concrete, short story that shows why the interim alone would mislead. Principals respond to specific students more than to abstract measurement arguments.
Framing that protects the relationship: I'd avoid saying the interim is 'invalid' or that the data wall is 'meaningless.' Instead: 'The interim is a good screen for procedural fluency; here's what it doesn't capture, and here's a low-cost way to see it.' I'd offer to present the modeling evidence at a department meeting so the principal gets credit for supporting richer data, and so the conversation becomes about improving the measure rather than defending my teaching.
Bottom line: Comply with the request, supplement it with better evidence, and keep the framing collaborative - 'here's more, not instead.'