Build a Quiz and Rubric That Match What You Taught
Most quiz questions come from wherever's convenient, a textbook bank, a template from last term, a quick search. That's fast, but it also means the questions can quietly test things you never actually covered, or grade against criteria that don't match how you taught the material. A teacher who spent a week on primary-source analysis can hand out a bank question that tests date memorization instead, and not notice the mismatch until the grades come back strange.
Warning
A question pulled from a generic bank tests the bank's version of the topic, not necessarily the version you taught. A student can know your lesson cold and still miss a question written for a different angle on the same subject.
An assessment should test what you taught, not what a question bank assumed you'd teach.
Why borrowed questions drift from the lesson
A question written by someone else was written for their framing of the topic, their emphasis, their assumed prior knowledge. Even on the same subject, that framing rarely lines up exactly with what happened in your specific session, and the gap shows up as students who clearly understood the material still missing points. The drift compounds when the rubric comes from a different source than the questions, since a scoring guide written for one framing of a topic doesn't reliably score answers written against a different one.
Picture a high school history teacher who spent a class period on the specific economic causes of World War I, alliance obligations, arms races, and colonial competition, using a set of primary sources students annotated together. A bank question that asks students to "list the causes of World War I" in general terms will get answered correctly by students who read the textbook chapter on their own and skipped the in-class discussion entirely, while a student who engaged deeply with the actual lesson but phrases things differently than the answer key expects can lose points for a technically correct but unexpected answer.
Before: pull questions from a bank or a past assessment, add a rubric separately, hope the two line up with what was actually taught.
After: give Claude the actual lesson content and what you want to assess, and get back questions and a rubric built from the same material, aligned from the start.
A process for a rubric-aligned assessment
- 1
Give Claude the actual content, not just the topic name
Paste your notes, slides, or a summary of what was actually covered. "The causes of World War I" as a topic name leaves too much room for drift; the specific content you taught, including which primary sources or examples you used, narrows it to what students should actually know.
- 2
Ask for a fixed number of questions at a stated difficulty
Request a specific count and level, five questions at an intermediate difficulty, rather than an open-ended set. That's what keeps the assessment scoped to a realistic session length instead of ballooning into something that takes longer to grade than it did to teach.
- 3
Ask for the rubric alongside the questions, not after
Request scoring criteria for each question at the same time it's written, tied to what a correct answer should demonstrate. A rubric written afterward tends to describe the answer key instead of the actual skill being assessed, which makes it harder to fairly credit an answer that's right but phrased differently than expected.
- 4
Check each question against your lesson content directly
Read every question back against what you actually covered. A question that requires knowledge from outside the lesson, even accidentally, is worth cutting or revising before it reaches students, since it's testing whether they did outside reading rather than whether they learned what you taught.
Build a 5-question quiz on the causes of World War I for a high school history class, with a rubric that scores each answer on accuracy and use of at least one specific example.
”For a lesson built around a specific activity, a lab, a primary-source annotation exercise, a group discussion, it helps to tell Claude what actually happened in the room, not just the underlying topic, so the questions test the version students actually experienced.
This lesson was built around students annotating three primary sources in small groups, not a lecture. Write questions that ask students to reference specific evidence from those sources, not just recall general facts about the topic.
”
Inside Claude Tutorial
Building the test and the grading criteria from the same source is transferable.
Writing an assessment and its rubric together, from what was actually taught, instead of assembling them from separate sources, applies well beyond quizzes. The app has a full lesson on it, with practice that carries over to any evaluation you're designing.
When the rubric needs to flex for partial understanding
A strict right-or-wrong rubric can undersell a student who understood the core idea but missed a detail. Building in partial credit deliberately, rather than deciding case by case while grading, keeps the standard consistent across every paper. This matters most on open-response questions, where two students can write genuinely different answers that both demonstrate real understanding of the underlying concept.
Tip
Ask Claude to add a partial-credit tier to the rubric for each question, describing what a partially correct answer would include. Deciding that standard before grading starts is what keeps it consistent from the first paper to the last.
For each question in this rubric, add a partial-credit tier describing what an answer would need to show to earn half credit, not just full credit or none.
”The same flexibility question comes up in reverse for a student who technically hits every rubric point but shows no real depth, an answer that's correct but feels mechanical. It's worth deciding ahead of time whether the rubric should reward genuine insight beyond the minimum bar, rather than discovering mid-grading that two very different answers earned the same score.
Add a top tier to this rubric that distinguishes a technically correct answer from one that also connects the concept to something else covered in the unit, so genuine depth is rewarded above the baseline.
”Continue reading
- Writing a Lesson Plan That Fits the Time You Actually Have: building the session this assessment is designed to test.
- Turning Quiz Results Into What to Reteach, Not Just Who to Flag: what to do with the results once this assessment has been given.
- Giving Feedback on Student Work Without Softening the Real Issue: turning a rubric score into feedback the student can actually act on.