The Data That Agrees With You
Seventh-grade ELA. Your argument unit post-assessment shows a class average of 78 percent, up from 71 percent on the pre-assessment. Item analysis: 82 percent correct on identifying a claim; 41 percent correct on judging whether evidence is relevant to a claim; 28 percent correct on explaining the reasoning that links evidence to claim. The eight students who scored above 90 were nearly all in your advisory group, which met daily during the unit. Five of the eight students who scored below 60 missed six or more days during the unit, including two who were out for a suspension. Your PLC meets Monday and the shared pacing guide puts the next unit - narrative writing - starting Tuesday.
1. What does this data actually support, and what would you need to know before treating the seven-point gain as evidence your instruction worked?
Read the worked answer
The seven-point gain is real but it is not evenly distributed, and an average hides that.
What the item analysis supports: the gain is almost certainly concentrated in the lowest-complexity task. Identifying a claim is largely recognition; a 71-to-78 jump with 82 percent on claim identification and 28 percent on reasoning means students got better at the part that requires the least. That's not nothing, but it isn't the standard.
The 28 percent on evidence-reasoning is the headline. When a single skill collapses that far below everything else, the first hypothesis should be instructional, not student: did the unit actually teach reasoning, or did it teach claim-plus-quote and assume the bridge? I'd audit two weeks of my own materials for whether I ever modeled the sentence that connects evidence to claim.
The advisory confound matters. If the high scorers were the students who saw me daily, the gain may reflect additional time and relationship, not the core instruction - which is useful information, not a reason to feel good. The attendance confound cuts the other way: five of eight low scorers missed six or more days, so their scores partly measure absence. Before I use this data to change instruction, I'd want to know what those five can do on a cold read of a short argument - a ten-minute task, not a unit test.
2. Monday's PLC wants to start narrative writing Tuesday. How do you make your case with evidence, and what is your fallback if the team decides to move on?
Read the worked answer
I'd bring three things and no adjectives.
- The item-level data, printed. One column per item, one row per student. It takes the team ninety seconds to see the 28 percent.
- Two anonymous student responses - one strong claim with a quote and no reasoning, one that explains the link. Ask the team which one meets the standard. That question usually does more than my argument would.
- A costed proposal. Not 'we should slow down,' but: 'Twenty minutes a day for four days inside the narrative unit. Students write a claim about a character's motive and defend it in the narrative's margins. The standard doesn't disappear because the genre changed.'
The realistic risk is that I lose the vote. If I do, my fallback is not quiet compliance or a private parallel unit. Narrative writing is genuinely rich territory for this skill - a character's choices are claims, and a scene is evidence - so I'd embed the reasoning sentence as a required reflection in the narrative task and report back at the next PLC with the same item analysis. If the 28 percent doesn't move, I have a stronger case and better standing.
3. Name the one instructional change you would make, and design the two-week check that would tell you whether it worked.
Read the worked answer
The change: explicit, repeated work on the reasoning sentence - the one that explains why a piece of evidence supports a claim. I would stop accepting quoted evidence as a complete answer, and I'd model the sentence structure, contrast weak and strong examples, and require revision until the link is stated.
The check, two weeks out: five argument items, three of them reasoning items written in a new topic domain so I'm measuring transfer rather than recall of unit content. Administered cold, ten minutes, ungraded for the students' sake.
Decision rules set before I look:
- Reasoning items above 55 percent: the move is working; continue.
- Between 35 and 55: partial; I'd examine whether students can do it with a sentence frame but not without, which tells me the frame became a crutch.
- Flat or below 28: my hypothesis was wrong, and the problem is likely upstream - students may not be able to identify what a claim even needs, which sends me back to the reading side of the unit.
I'd also check the five students who missed six or more days separately rather than folding them into the class number, so I don't mistake their absences for my failure or my success.