Standards-Based Grading: A Practical Teacher Guide

Standards-based grading reports what a student can currently demonstrate on specific learning standards instead of hiding every task inside one assignment average. A useful system needs four things: clear targets, a defined proficiency scale, comparable evidence over time, and a written rule for reassessment and final marks.

There is no universal 1–4 formula. District policy still controls report cards, credit, accommodations, and grade conversion. The safest way to start is one six-week unit—not a whole-school switch—and test whether two teachers can look at the same evidence and reach the same proficiency decision.

This guide gives you the scale, worked example, evidence log, reassessment policy, family note, and pilot plan to do that.

What Is Standards-Based Grading?

Standards-based grading, often shortened to SBG, records achievement by learning target. An entry says something like “W.7.1: supports an argument with relevant evidence — Proficient” rather than only “Essay 2 — 84%.” Practice, timeliness, participation, and academic achievement may all matter, but they are not blended into one unexplained number.

The Washington State Board of Education describes several common features: grades are organized around learning goals, success criteria are available in advance, achievement is separated from effort or behavior, selected evidence is used for grading, and recent evidence receives emphasis. Those are design principles, not a mandatory national model.

Standards-based grading is not:

• changing an 82% to a “3” while leaving the rest of the system untouched;
• giving unlimited retakes without new learning;
• grading every worksheet against every standard;
• declaring homework, deadlines, or collaboration unimportant;
• letting software make the final proficiency decision; or
• requiring every student to reach a target on the same day.

It is a reporting decision: the academic mark should describe the strongest defensible evidence for a named skill at the reporting date.

Standards-Based Grading vs. Traditional Grading

The important difference is not numbers versus words. It is what the grade claims to represent.

• Main unit: A traditional gradebook centers an assignment, test, or project. A standards-based gradebook centers a standard or learning target.
• Typical entry: Traditional reporting might record “18/20 on quiz 3.” Standards-based reporting might record “Proficient on solving two-step equations.”
• Practice: Traditional systems often include practice in the average. SBG usually records it separately or gives it little or no academic weight.
• Behavior: Traditional marks may include behavior. SBG reports it separately when local policy allows.
• Later learning: Traditional systems often average later evidence with early attempts. SBG interprets it through a stated evidence rule.
• Student question: A points system prompts “How many points do I need?” A target-based record prompts “Which part of the target can I demonstrate?”

A traditional gradebook can still be accurate, and a standards-based one can still be inconsistent. SBG becomes more informative only when each target is small enough to assess, each level has observable criteria, and teachers apply the same evidence rule.

Start with student-friendly learning targets. If a target says “understand ecosystems,” neither the student nor the scorer knows what evidence counts. A target such as “use evidence from a food-web model to predict the effect of removing one population” is assessable.

Build a Four-Level Proficiency Scale

Many schools use four levels, but the labels and calculation rules are local decisions. The North Carolina Department of Public Instruction’s standards-based reporting materials show why teams must define both the scale and the final-grade rule locally rather than assume one conversion.

Use this starting structure, then replace the generic language with descriptors for the actual standard:

• 4 — Extending: Applies the target in a new or more complex context. Evidence shows independent transfer, justified choices, and all Level 3 requirements.
• 3 — Proficient: Meets the stated grade-level or course target. Evidence shows accurate, independent performance on the complete target.
• 2 — Developing: Shows important parts but not the complete target. Evidence is partly accurate, inconsistent, or successful with prompts or support.
• 1 — Beginning: Shows limited prerequisite or target evidence. An entry point is visible, but substantial support is still required.
• NE — Not enough evidence: There is not yet a defensible proficiency decision. Evidence may be missing, inaccessible, or incomparable; this does not automatically mean zero mastery.

Three guardrails prevent common scale problems:

1. Level 4 is not “more work.” Ten correct problems do not prove deeper transfer than five equivalent problems. Define a more complex application.
2. Level 3 is the target, not mediocrity. If 3 means the student independently meets the standard, describe it to families as success.
3. NE is not a disguised 0. Missing evidence creates an action—collect evidence—rather than a claim that the student knows nothing.

Do not convert 1, 2, 3, and 4 to percentages by intuition. A 3 on a four-level proficiency scale does not inherently mean 75%. If your school requires a letter or percentage, use its approved conversion and explain what information is lost in the conversion. For arithmetic checks after a policy is set, use the appropriate grade-calculation method; do not let the calculator define the policy.

A Worked Standards-Based Grading Example

Suppose a seventh-grade team is assessing an argument-writing target based on CCSS.ELA-LITERACY.W.7.1: write arguments with clear reasons and relevant evidence.

The team narrows one reporting target to this:

Target: I can support a clear claim with relevant evidence and explain how the evidence supports my reasoning.

Its scale might read:

• 4: Meets Level 3 and transfers the reasoning to a new source set, addresses a meaningful counterclaim, and explains why one piece of evidence is stronger than another.
• 3: States a clear claim, selects relevant evidence from the provided sources, and explains the connection between evidence and claim independently.
• 2: States a claim and includes evidence, but the evidence is partly irrelevant, underexplained, or selected with teacher prompts.
• 1: Restates the topic or copies source details without a defensible claim-and-evidence connection.

Now record evidence without pretending every entry is equally informative:

1. Sept. 9 — Exit paragraph: The student works independently with one familiar source. The result is Level 2: the quote is relevant, but the explanation is incomplete. Next, model an evidence-to-reasoning sentence.
2. Sept. 16 — In-class response: The student works independently with two familiar sources. The claim, evidence, and explanation meet Level 3. Next, test the same target with a less familiar source set.
3. Sept. 24 — Revised argument: The student attends a conference before revising, then writes the final response independently. The explanation is stronger and remains Level 3. Keep Level 3 and offer a transfer task if Level 4 evidence is needed.

The defensible report is Level 3, not because 3 is the highest number and not because the mean of 2, 3, and 3 is 2.67. The later samples are comparable, independent evidence of the complete target, and they form a consistent pattern. If the September 24 response had been completed line by line with a teacher, it would not carry the same weight.

This distinction matters: standards-based grading should favor better evidence, not automatically favor the newest or highest score.

Choose an Evidence Rule Before Students Submit Work

An evidence rule tells teachers, students, and families how multiple observations become one current proficiency level. Write it before the first assessment.

Use this four-question protocol for each reporting decision:

1. Alignment: Does the task actually measure the full target?
2. Independence: What prompts, scaffolds, collaboration, or reference materials were available?
3. Comparability: Is this attempt as rigorous as the earlier evidence?
4. Pattern: Is there enough consistent evidence to make the decision stable?

Then choose one documented aggregation approach:

Most recent comparable evidence

Use the latest valid sample when targets are assessed repeatedly under comparable conditions. This responds well to learning, but one rushed or unusually easy task can distort the mark.

Consistent body of evidence

Use the level supported by multiple recent samples. This is often more stable, but scorers need a shared decision rule. For example: “At least two independent samples at Level 3, including one summative transfer task, with no newer contradictory evidence.”

Recency-weighted calculation

Some gradebooks weight later evidence more heavily. This is easy to automate but can create false precision. A 2.74 still needs an educational interpretation, and a weighted mean cannot decide whether the tasks were aligned or comparable.

Whichever approach you use, moderate a small sample with a colleague. Score three pieces independently, compare the evidence behind each rating, and revise descriptors where the disagreement came from the scale—not the student.

Write a Reassessment Policy That Protects Access and Rigor

Reassessment should create new evidence after new learning. It should not become either an unrestricted test loop or a privilege available only to students who can stay after school.

Adapt this policy template:

Reassessment policy: A student may submit new evidence for a target after completing a brief correction or practice plan. The reassessment measures the same target at comparable rigor with different prompts or items. Students may use the scheduled class flex block; an after-school option is additional, not required. The new result replaces or updates earlier evidence according to our published evidence rule. Missing work, work habits, and academic proficiency are communicated separately.

Define these details:

• Readiness: What shows that the student addressed the gap?
• Access: When can every student reassess during the normal school day?
• Scope: Is the student reassessing one standard or repeating an entire test?
• Rigor: How will you avoid an easier second version?
• Deadline: How late can new evidence arrive before reporting closes?
• Replacement rule: Does new evidence replace, supplement, or receive greater weight than old evidence?
• Accommodations: How will IEP, 504, language, and accessibility supports carry into the new attempt?

A test-correction routine can prepare a student for reassessment, but a guided correction sheet is not always independent evidence of proficiency. Use a new item, explanation, conference, performance, or short transfer task to verify learning.

Set Up a Standards-Based Gradebook That Teachers Can Maintain

Do not begin by importing every state standard. A gradebook with 80 targets and one weak observation per target creates less clarity, not more.

For one pilot unit, give each record these fields:

• Standard code: W.7.1.
• Student-facing target: Support a claim with relevant evidence and reasoning.
• Proficiency descriptor: The complete Level 3 language from the shared scale.
• Evidence source and date: Source-based argument, Sept. 24.
• Conditions: Independent; text annotations allowed.
• Current and previous levels: Current 3; previous 2.
• Next step: Compare the strength of evidence in a new source set.
• Work-habit note: Submitted after the deadline.

Keep the evidence source and conditions. A gradebook that stores only “3” cannot tell a future teacher whether the student wrote independently, used a scaffold, or demonstrated the whole target.

Use formative assessment examples to decide what guides instruction without becoming a permanent mark. Use summative assessment examples for the more independent evidence that supports a reporting decision.

For rubric-heavy assignments, the free rubric generator can help draft criterion language. Treat the output as a draft: check the standard, developmental fit, scoring distinction, accessibility, and point logic before students see it. In an AI grading workflow, a teacher should still review criterion scores and comments before returning them.

Run a Six-Week Pilot Before a Larger Rollout

Week 1: Select the learning

Choose three to five priority targets for one unit. Unpack each into observable student actions. Remove targets you will not have time to teach and assess well.

Week 2: Build and calibrate the scale

Write Level 2, 3, and 4 descriptors using real sample work. Ask a colleague to score the samples without your explanation. Revise any descriptor that needs oral translation.

Week 3: Teach students to read the scale

Show one anonymous sample at two proficiency levels. Ask students to highlight the evidence for each descriptor. Let students practice identifying one next step before they receive a mark.

Week 4: Collect low-stakes evidence

Use exit responses, short explanations, conferences, or problem sets. Record conditions and feedback, but do not rush to a final proficiency decision.

Week 5: Collect independent evidence and reassess

Use a task that covers the complete target. Apply the published reassessment policy to one standard so you can test scheduling, access, and gradebook behavior.

Week 6: Audit the system

Review five questions:

1. Could students explain what Level 3 required?
2. Did two scorers apply the descriptors similarly?
3. Did the gradebook preserve evidence and conditions?
4. Could every student access reassessment?
5. Did family questions reveal a policy gap or only unfamiliar language?

Expand only after the pilot answers those questions. A smaller reliable system is better than a school-wide launch built on undefined levels.

Explain Standards-Based Grading to Students and Families

Send a concrete example before the first report. Avoid leading with philosophy.

You can adapt this family message:

This unit’s academic report is organized by three learning targets. “3 — Proficient” means the student independently meets the target. “2 — Developing” means the evidence is partial or still needs support. Work habits and deadlines are reported separately, so the academic level describes learning. Students can provide new evidence after targeted practice using the reassessment process attached. If a letter grade is required, we use the school’s published conversion—not a direct 1–4 percentage calculation.

Show one assignment beside its standards record. A paper may provide Level 3 evidence for claim and evidence, Level 2 for counterclaim, and a separate work-habit note for timeliness. That is clearer than saying only “84%,” but families need to see the mapping.

Report both the current level and the next action. “Developing in proportional reasoning” is incomplete communication. Add: “Can build equivalent-ratio tables; next, explain the constant of proportionality in an equation.”

Common Standards-Based Grading Mistakes

• Relabeling percentages. Converting 90–100 to 4, 80–89 to 3, and so on keeps the old point system while making the labels harder to interpret.
• Tracking standards that are too broad. “Writing” or “algebra” cannot produce actionable evidence.
• Making Level 4 mysterious. Students should know the transfer or complexity requirement before the task.
• Using only the highest score. A high result from an easier or heavily scaffolded task may not represent current independent performance.
• Averaging every attempt. Early practice can overpower later valid evidence and make improvement mathematically irrelevant.
• Treating missing evidence as no learning. Use NE, collect evidence, and follow local missing-work policy.
• Offering reassessment outside school only. That can make access depend on transportation, work, caregiving, or internet availability.
• Mixing behavior back into the academic level. If timeliness matters, report it clearly rather than quietly lowering a math or writing standard.
• Automating judgment without review. A gradebook or AI tool can organize evidence; it cannot know whether the task truly measured the intended standard without teacher validation.

If student work is processed by a digital or AI service, follow district approval and data-minimization rules. The U.S. Department of Education’s Student Privacy Policy Office maintains FERPA and data-security guidance. Remove unnecessary identifiers, confirm retention and training-use terms, limit access, and do not upload protected records to an unapproved tool. GradeWithAI’s student-data privacy guide provides a practical vendor-review checklist, but district policy remains controlling.

Frequently Asked Questions

Does standards-based grading mean no letter grades?

No. A school can report proficiency by standard and still issue a required course letter grade. The conversion must be a published local rule. Do not assume that a 3 automatically equals 75%, 85%, or any other percentage.

Is a 3 good in standards-based grading?

Usually, a 3 means the student meets the stated standard independently—but labels differ. Publish the descriptor, not just the number. Families should not have to infer whether “3” means on target, average, or almost there.

Should teachers use the most recent or highest score?

Use the evidence rule your team publishes. Recent evidence is persuasive when it measures the complete target at comparable rigor and independence. The highest score is not automatically the best evidence, and the newest task can be anomalous.

How many pieces of evidence are enough?

There is no universal count. One complex performance may cover a target better than five exit tickets. For a classroom pilot, seek at least two recent, independent, comparable samples for important targets and document exceptions.

Does every assignment go in a standards-based gradebook?

No. Record enough evidence to support decisions. Practice can receive feedback without becoming a permanent academic mark. Tag only the targets an assignment genuinely measures.

Can students reassess indefinitely?

Not necessarily. A fair policy can require evidence of new learning, comparable tasks, a reporting deadline, and one or more scheduled opportunities. The key is that access and replacement rules are known in advance.

Can AI determine standards-based grades?

AI can help draft rubrics, organize criterion evidence, or suggest feedback. A teacher should verify alignment, accommodations, scoring, and comments before a proficiency result affects a student record. Use only school-approved tools for identifiable student work.

Sources and Next Step

This guide uses the Washington State Board of Education’s comparison, North Carolina DPI reporting materials, the Common Core writing standards, and the U.S. Department of Education’s student privacy guidance. These sources show legitimate models and constraints; they do not create one national grading formula.

For your next planning period, choose one target, write the Level 3 evidence first, add a Level 2 misconception and a Level 4 transfer task, then test the scale on three real samples. If the samples do not fit, improve the scale before changing the students’ grades.