Scoring a Milestones-Style Argument Response When a Source Is Misattributed
Three seventh-grade teachers at a south Georgia middle school are scoring a district argument-writing task built in the Milestones constructed-response format. The prompt asks students to argue whether the school day should start later, using two provided articles. The team's four-point rubric awards one point each for a clear claim, relevant evidence, explanation that connects evidence to the claim, and command of conventions. Darius writes: 'Schools should start later because scientists say teens need nine hours of sleep. The article says 73 percent of students are tired in first period. Also the CDC says start times should be 8:30 or later. This proves students will do better.' The 73 percent figure appears in Article 1; the CDC guidance appears in neither article. His commentary restates the evidence without explaining how sleep connects to achievement. His conventions are clean. Across the grade, 22 of 78 responses cite a source that is not in the texts.
1. Assign a score on the four-point rubric and justify it trait by trait. Then explain what the score alone fails to tell Darius's teacher and what feedback you would attach.
Read the worked answer
Score the traits separately rather than letting one problem color the whole response.
- Claim (1): The position is stated and arguable.
- Evidence (1): The 73 percent figure is present and relevant. The CDC attribution is inaccurate, but this rubric does not measure citation accuracy, so penalizing evidence here would score a trait the rubric never defined.
- Explanation (0): 'This proves students will do better' jumps from fatigue to achievement without a mechanism. The connection is asserted, not explained.
- Conventions (1): Sentence control is clean.
Total: 3 of 4 (or 3 of 4 with the accuracy note flagged separately).
What the score hides is the source problem. Giving a 2 would silently punish Darius for a skill the rubric did not name, and giving a 4 would reward unsupported elaboration. Because 22 of 78 responses show the same citation pattern, this is an instruction issue, not a Darius issue. Attach feedback that names the one revision needed: state what the sleep evidence shows about attention or test performance, then link it back to the claim. Flag the CDC citation in a separate accuracy comment so the score and the correction stay conceptually distinct.
2. The team has 25 minutes the next day before the unit continues. Design the re-teach and name the evidence you would collect to know it worked.
Read the worked answer
Twenty-five minutes cannot fix everything, so target the single trait that scored zero and treat the citation pattern as a two-minute side note.
Sequence:
- Project Darius's response anonymously. Ask students to underline the claim and circle the sentence that explains why the evidence matters. They will find no such sentence, which creates the need before the strategy is named (5 minutes).
- Contrast two commentary sentences: one that repeats ('This shows students are tired') and one that explains ('Because tired students lose focus in first period, later start times could raise the share who are alert during instruction'). Name the difference as repetition versus mechanism (5 minutes).
- Students revise only the commentary sentence in their own response, using the frame 'The evidence ___ matters because ___, which means ___' (10 minutes).
- Two volunteers read revised sentences; the class checks whether the sentence names a cause or consequence (5 minutes).
For sources, have students highlight each cited fact inside the articles. Anything unhighlighted is either a citation error or a paraphrase that needs a page reference.
Evidence of success: an exit slip with a new short passage in which students write only the commentary sentence. Success criterion: at least 80 percent name a mechanism without repeating the evidence wording.