Rubric Maker Prompt: Analytic, Single-Point, or Holistic

Prompt

Act as a K-12 assessment designer who builds rubrics from a teacher's own assignment and learning targets, writes level descriptors a student could check against their work, and makes sure the points add up before anything is handed out.

Inputs:
- The assignment, as students will see it: [AssignmentTask]
- Grade level and subject: [GradeAndSubject]
- Learning targets or standards this task assesses: [LearningTargets]
- Rubric type wanted (analytic, single-point, holistic, or "recommend one"): [RubricType]
- Number of levels, level names, and any weighting: [ScaleAndWeights]
- What students struggle with on this kind of task: [CommonMistakes]
- Who reads it (students, parents, co-teachers) and reading level: [Audience]

Generate:
1. Rubric type check: confirm RubricType or recommend one, in two sentences tied to the task (analytic for separate skills, single-point for feedback on drafts, holistic for quick overall judgments).
2. Criteria list: one criterion per learning target in LearningTargets, each named in student language. Flag any target the task does not actually let students show.
3. The rubric as a table, using ScaleAndWeights. Every descriptor describes what is visible in the work (counts, labels, uses of evidence), not effort or attitude. Each level differs from the next by one clear step.
4. Point math: maximum score, the score for a student who earns the middle level everywhere, and what one point is worth as a percentage.
5. A student-friendly checklist version of the top level, written for Audience.
6. A single-point version of the same rubric (criteria plus the proficient column, with blank "concerns" and "advanced" columns) for draft feedback.
7. Feedback comment stems: two per criterion, one for each of the two most common mistakes from CommonMistakes.
8. Calibration check: three questions the teacher answers after scoring three sample papers, to catch descriptors that do not separate levels.

Rules:
- Use only the task, targets, and scale given. If a level count or weight is missing, ask instead of guessing.
- No grade bands or district policies that were not provided.
- Keep student names and identifying details out of examples.

Instructions

Run this in ChatGPT (it also works in Claude or Gemini) with your own assignment sheet and learning targets. Replace every [bracket]. Do not paste student names, IDs, grades, or work samples that identify a student into a public AI tool; FERPA protects student education records, so follow your district's AI and data privacy policy. Review every descriptor before students see it.

Generated Output

This image was generated using the prompt above.

Rubric Maker Prompt: Analytic, Single-Point, or Holistic - Result

Examples

Example Input

AssignmentTask: Lab report for the investigation "How does water temperature affect how fast sugar dissolves?" Students test water at 10, 25, and 50 degrees Celsius, three trials at each temperature, and time how long one sugar cube takes to dissolve.
GradeAndSubject: Grade 7 science.
LearningTargets: (1) write a testable hypothesis, (2) identify independent, dependent, and controlled variables, (3) record data in a table and make a line graph, (4) use the data to support a conclusion, (5) explain one source of error.
RubricType: analytic.
ScaleAndWeights: 4 levels named Exceeds (4), Meets (3), Approaching (2), Beginning (1). Target 4 (conclusion from data) counts double.
CommonMistakes: conclusions that never mention a number from the data; graphs with no axis labels or units.
Audience: grade 7 students and their parents.

Example Output

1. RUBRIC TYPE
Analytic fits this lab report: the five targets are separate skills, and a student can have a strong graph and a weak conclusion. Use the single-point version (part 6) for draft feedback.

2. CRITERIA
- Hypothesis (target 1)
- Variables (target 2)
- Data table and graph (target 3)
- Conclusion from data (target 4, counts double)
- Source of error (target 5)
All five targets can be shown in this task.

3. ANALYTIC RUBRIC
Hypothesis
- Exceeds (4): If-then hypothesis names temperature and dissolving time and predicts a direction with a reason.
- Meets (3): If-then hypothesis names both variables and predicts a direction.
- Approaching (2): Names one variable or makes a prediction that cannot be tested.
- Beginning (1): No hypothesis, or a question instead of a prediction.
Variables
- Exceeds (4): Correctly names the independent, dependent, and at least three controlled variables, and says how each control was kept the same.
- Meets (3): Correctly names independent, dependent, and two controlled variables.
- Approaching (2): Names two of the three variable types correctly.
- Beginning (1): Variables missing or mixed up.
Data table and graph
- Exceeds (4): Table shows all 9 trials and the average for each temperature; line graph has a title, labeled axes with units, and the averages plotted.
- Meets (3): Table shows all 9 trials; graph has labeled axes with units.
- Approaching (2): Missing trials, or graph missing axis labels or units.
- Beginning (1): No table or no graph.
Conclusion from data (x2)
- Exceeds (4): States whether the hypothesis was supported and quotes the average time at two or more temperatures to explain the pattern.
- Meets (3): States whether the hypothesis was supported and quotes at least one number from the data.
- Approaching (2): States a conclusion but uses no numbers from the data.
- Beginning (1): No conclusion, or a conclusion the data does not show.
Source of error
- Exceeds (4): Explains one specific source of error, how it changed the results, and how to fix it next time.
- Meets (3): Explains one specific source of error and how it could change the results.
- Approaching (2): Names a vague error ("we messed up").
- Beginning (1): No source of error.

4. POINT MATH
Four criteria x 4 points = 16, plus Conclusion 4 x 2 = 8. Maximum 24.
Meets on every criterion: 3 x 4 = 12, plus 3 x 2 = 6, total 18 of 24 = 75%.
One point = about 4.2% of the grade.

5. STUDENT CHECKLIST (top level)
- My hypothesis says "If... then..." and gives a reason.
- I named what I changed, what I measured, and three things I kept the same.
- My table has all 9 trials and an average for each temperature.
- My graph has a title, labels, and units on both axes.
- My conclusion uses at least two numbers from my data.
- I explained one mistake, what it did to my results, and how to fix it.

6. SINGLE-POINT VERSION
Columns: Concerns | Proficient | Advanced. The Proficient column uses the five Meets descriptors above; the other two columns stay blank for written feedback.

7. FEEDBACK STEMS
- Conclusion: "Which number in your table shows this? Add it to your conclusion."
- Conclusion: "Your data shows a pattern across the three temperatures. Say what happens as the water gets warmer."
- Graph: "Label the x-axis with temperature and its unit (degrees Celsius)."
- Graph: "Add units to the y-axis so a reader knows the time is in seconds."
(Two stems each for Hypothesis, Variables, and Source of error follow the same pattern.)

8. CALIBRATION CHECK
- Did two papers you would call different land on the same level for any criterion?
- Did any descriptor need a judgment word ("good", "clear") to decide?
- Did the double weight on Conclusion change any overall score by more than one level?