The Keyword Chart the Data Do Not Support
Mr. Delgado teaches Grade 3 in a school six weeks from the state mathematics assessment. On the most recent common formative, 21 of 24 students correctly solved three-digit addition and subtraction problems, finishing most in under two minutes. On eight two-step word problems, only 9 of 24 were correct. The dominant error is operation choice: students add whenever they see more or total, even in comparison problems that require subtraction. Six students got correct answers but could not explain their steps when asked. The instructional coach recommends fifteen minutes of daily timed fact drills and a posted keyword chart (more = add, left = subtract). Meanwhile, Lina used compensation (398 + 257 as 400 + 257 − 2) and was marked wrong because the rubric requires the standard algorithm. Her father has emailed asking why. The team meets Thursday to finalize the next unit.
1. Using the numbers, explain what the data do and do not support in the coach's proposal. State what you would accept and what you would reject, including what to do about Lina.
Read the worked answer
What the data support. Fluency with the algorithm is a strength (21 of 24, most under two minutes). Fifteen minutes of daily timed fact drills targets a skill most students already have and would displace problem-solving time. Accept five minutes of spaced retrieval practice, not fifteen.
What the data do not support. The keyword chart. The error pattern is exactly what keyword rules produce: more triggers addition even when the problem asks how many more. Nine of 24 correct on two-step problems, with operation choice as the dominant error, points to representation and relationship-building, not vocabulary. Posting the chart would ratify the misrule.
The six who cannot explain. Correct answers without explanation indicate fragile procedural knowledge or answer-getting without a model. They need bar models, number lines, and a paraphrase routine: What quantities do we have? What is the relationship?
Lina. Compensation is a valid, efficient strategy that reflects place-value understanding. The rubric is measuring algorithm fidelity, not addition, which is a construct mismatch. Accept and discuss her method; separately require one standard-algorithm item so the rubric has a legitimate target. Marking a correct, explained strategy wrong teaches that mathematics is rule-following.
Decision: accept five minutes of fact practice, reject the keyword chart, and spend the remaining time on compare problems and representation.
2. Design a three-week instructional plan that addresses the word-problem errors while keeping fluency strong.
Read the worked answer
Week 1: Daily 10-minute number talk using compensation and decomposition (398 + 257; 500 − 276). Introduce a three-step paraphrase: read, restate in your own words, name the quantities. Use one compare problem per day with a bar model; students label the known and unknown parts before computing.
Week 2: Schema-based pairs: compare problems (how many more, how many fewer) and two-step problems. Students sort six problems by structure before solving, then solve two and explain to a partner. The nine students with operation errors get a small group with concrete bar-model strips and teacher think-alouds.
Week 3: Mixed practice with faded scaffolds. Students write their own two-step problem and trade. Lina's compensation strategy is featured in a number talk; she explains why subtracting 2 works. The standard algorithm is taught alongside, with the message that both are tools chosen for the numbers.
Formative checks: Tuesday error-analysis exit tickets; Friday one two-step problem scored for representation, computation, and explanation. Keep a running error tally to see whether more = add is fading.
Alignment: these moves target the standards for two-step problems and place-value strategies, not just test format. Fluency practice stays short and distributed.
3. How should grading and feedback change so that Lina's work is judged fairly without abandoning the algorithm standard?
Read the worked answer
Separate the standards on the report card: computation (NBT), two-step problem solving (OA), and mathematical explanation. Lina's computation score should reflect correct answers; her explanation score should reflect a clear, valid strategy. A rubric that only awards points for the standard algorithm conflates the two.
Change the task, not just the wording. If the goal is algorithm fluency, ask for the algorithm on a separate item with clear directions. If the goal is problem solving, accept any valid strategy and score the representation and reasoning. Never mark a correct, explained compensation strategy wrong on a problem-solving task.
Give feedback that names the strategy: You used compensation; explain why 400 + 257 − 2 is equivalent. For the six students who cannot explain, ask for a drawing or bar model before a number sentence. For the keyword group, ask: Which two quantities are being compared?
Communicate with Lina's father in plain language: the class accepts more than one valid strategy, and the rubric will be revised to score reasoning separately from the algorithm. Keep a copy of the revised rubric and the item analysis, because this is the kind of decision a parent may challenge.