The Rubric Everyone Passed
Ninth-grade World History runs three sections of 30. The common document-based question asks students to explain how Roman roads and aqueducts supported imperial expansion, using five documents. The department rubric has four criteria at four points each: contextualization, thesis, evidence use, and sourcing/analysis.
Across the teacher's three sections, 27 of 30 scored 3 or 4 on evidence use; 22 of 30 scored 1 or 2 on contextualization; 11 of 30 never addressed sourcing at all. A colleague teaching the same course, with the same prompt and rubric, reports a class mean twelve points higher. Two students with 504 plans for extended time took the task in a resource room and scored at the top of the distribution. One student asks, “Is contextualization the background paragraph we did in September?”
1. What do these numbers tell you, and what do they not tell you?
Read the worked answer
What they suggest: the profile is not a student deficit. Students can select quotations (27 of 30 at 3–4 on evidence) but cannot build a frame that makes the quotations mean anything (22 of 30 at 1–2 on contextualization). That combination usually means the sourcing-and-context routine was never made concrete - students were taught to find quotes, not to establish what a reader needs to know.
What they do not tell me: whether 22 of 30 represents weak instruction or a rubric that two teachers read differently. The student's question is the most diagnostic data point in the whole set. If a strong student thinks contextualization is “the background paragraph from September,” the criterion has never been operationalized for anyone - that is a task-design and rubric-language problem, not a comprehension problem.
The twelve-point gap is a validity flag, not a verdict. It could be different student populations, different instructional emphasis, or different scoring severity. Assuming the colleague is easy or that I am ineffective are both premature. The 504 scores are also informative: with extended time and a quieter setting, those students reached the top of the distribution, which points to pacing and working-memory load rather than to understanding.
What I would do before reteaching anything: pull ten scripts blind across both classes, score them independently with the colleague, and see where we diverge. If we diverge on contextualization, the rubric is the intervention.
2. Design the next instructional cycle - reteach, reassess, and rubric revision - and explain what each piece is meant to detect.
Read the worked answer
1. Calibration before instruction (30 min, common prep). Both teachers independently score the same ten essays, then compare criterion by criterion. Purpose: separate a scoring problem from a learning problem. If our contextualization scores differ by two points on the same paper, no reteach will fix that.
2. Rewrite the criterion in student language. Replace “contextualization” with a concrete obligation: Before quoting, tell the reader what was happening at the time that makes this quotation matter. Pair it with one model and one non-model paragraph - a quote dropped into a paragraph with no frame, versus the same quote preceded by two sentences of situation. Purpose: eliminates the “background paragraph from September” confusion without lowering the demand.
3. Reteach on new content, not Rome. Use a short sequence on a different case - Mali's trans-Saharan trade, or Han infrastructure - so that what I assess is transfer rather than memory of the reteach lesson. Purpose: distinguishes a routine students own from a routine they can only apply to the text in front of them.
4. Reassessment with adjusted conditions. Give the DBQ in two timed segments with a required planning box; keep extended time as the default for all students on this task type rather than a 504 accommodation, since the resource-room data suggests pacing is a general constraint.
5. Look at sourcing separately. Eleven of thirty ignoring sourcing is an instruction gap; add a two-question sourcing prompt (“Who made this, and what did they want?”) that students must answer on every document before writing.
3. How do you raise the twelve-point gap with your colleague without it becoming a competition or an accusation?
Read the worked answer
Lead with the instrument, not the scores. Opening with “your class scored twelve points higher” invites either defensiveness or false modesty. Opening with “I'm not sure we're scoring the same thing, and I'd like to test that” invites calibration. The anchor-paper exercise is the vehicle: it produces evidence about the rubric rather than about either teacher.
Be genuinely open to two findings. One is that scoring differs. The other is that his students really did learn contextualization better, in which case the useful question is what he did - and I should ask to see it rather than assume leniency. I would go in saying explicitly that I do not know which it is.
What I would avoid: comparing class means in a department meeting; implying that high scores mean low standards, which is a common and corrosive assumption in PLCs; and going to an administrator before talking to him, which ends the collaboration I need for the rest of the year.
If calibration shows real instructional differences, that is a win for the PLC - we have a strategy worth spreading, and the data to justify it. If calibration shows scoring drift, the fix is anchor papers kept in the shared drive and a norming session each quarter, applied to both of us.
The stake is student-facing: a rubric that two teachers read differently means a grade that depends on which room a student sits in, which is indefensible in a common course.