6. Assigned Work Examples, Edge Cases, and Calibration Guide

Calvary Preparatory Academy — Teacher Grading Examples

Holistic Grading Model — Detailed Examples & Prompts

Page 6 of 6
Assigned Work • Edge Cases • Calibration

Assigned Work Examples, Edge Cases, and Calibration Guide

How to score Assigned Work completion, handle non-standard situations, and stay calibrated across the semester.

Assigned Work Completion — examples (0–10 per section)

This category measures completion status only — not quality. Quality is assessed in Learning Verification. Score based on what is actually submitted and ready at meeting time.

Score	What you see at meeting time	Example
10	All assigned work for the section is complete, organized, and immediately accessible. Student is ready to present without any setup time.	Student opens immediately to the completed worksheet, essay, and lab report. All three are finished. Meeting starts on time.
8	Most work complete. One minor item outstanding — perhaps a study guide or a practice activity. The significant assignments are all present.	Essay and quiz defense are ready. Student notes the study guide was not fully completed but the substantial work is all there.
5	About half the assigned work present. A meaningful portion is missing. Meeting is less productive because teacher has less to work with.	Two of four assigned activities are complete. Student explains they ran out of time on the other two and plans to finish this week.
2	Very little submitted. Perhaps one activity out of several. Student arrived at the meeting unprepared to present most of the section.	Only the first quiz is complete. Journal, essay, and two practice activities all missing at meeting time.
0	Nothing submitted. Nothing to review or present.	Student admits they did not complete any assigned work for the section.

When Assigned Work score is high but LV score is low

Assigned Work and Learning Verification work together by design. A student who submits AI-generated work earns completion points here (the AI completed the form of the work) but scores very low on Learning Verification (they cannot explain it in the meeting). The model is working correctly when you see this pattern.

Common edge cases and how to handle them

Situation	How to score it	Documentation
Student completes all work but it is clearly low quality	Score Assigned Work at 10 — completion is complete. Score Depth of Understanding appropriately based on what the meeting reveals. The two categories do different jobs.	No special documentation needed — the LV score captures quality.
Student presents portfolio work from a prior section	Award LV points for the section being defended. Enter in the section row that was defended. Note in col H which section the work came from.	Col H: "Portfolio — Section [X] material defended at Section [Y] meeting."
Student misses meeting; work was submitted	Meeting Attendance: 0/10. Assigned Work: score based on what was submitted at the original appointment time. LV: leave blank (no meeting = cannot score). Do not enter 0 for LV — blank means not scored, 0 means scored as zero.	Col H: "Meeting missed — LV not scored this section."
Student reschedules and has a makeup meeting	Meeting Attendance: per policy (0 for second+ reschedule, full for first). Score LV from the makeup meeting normally.	Col H: "Makeup meeting [date]." Document any point restoration under teacher discretion.
Student claims to have submitted work but teacher cannot find it	Ask the student to locate it during the meeting. If genuinely submitted and teacher missed it, score at 10. If student cannot produce it, score based on what is available.	Col H: "Submission dispute — student located file during meeting" or "Work not located."
Student scored very high in the digital textbook but meeting reveals a significant gap	Score LV honestly based on meeting performance. Set Teacher Conviction to 0 (discontinuity flag). Do not adjust the curriculum grade — that is Edmentum's number.	Col H: "Discontinuity flag. Curriculum grade [X]%. Meeting could not verify [specific concepts]. Follow-up initiated."

Staying calibrated across the semester

Calibration drift is the most common grading quality issue in a model that relies on professional judgment. Here are the patterns to watch for in your own scoring:

Drift pattern	What it looks like	Correction
Inflation over time	Scores gradually increase across the semester even for students who have not meaningfully improved. Section 15 looks like 90% for a student who earned 65% in Section 3 without clear growth evidence.	Review your score distribution at the midpoint of the semester. Compare Section 1–5 averages to Section 10–15 averages for each student. If scores are rising without a corresponding change in meeting quality, recalibrate.
Likability bias	Students with engaging personalities or who seem to enjoy the meetings score higher than quieter students of equal demonstrated understanding.	After each meeting, ask yourself: "Would I score this the same if a different personality delivered these exact answers?" Score the content, not the presentation style.
Fatigue scoring	Late-meeting subjects (last course in a long meeting) receive lower scores because teacher energy is lower.	Note which subject your student's meetings typically end on. If it consistently scores lower than the others, reorder occasionally or flag for attention.
Fear of zeros	Reluctance to award 0 for Teacher Conviction even when a significant discontinuity is clearly present.	The discontinuity flag exists precisely for this situation. Awarding a 1 when a 0 is warranted protects no one — and undermines the model's primary AI-integrity mechanism.
Notes neglect	Discretionary scores are entered but not documented in col H.	Build a habit: any time you depart from the rubric anchor in any direction, type a brief note in col H immediately. It takes 15 seconds and protects your professional judgment.

End-of-semester calibration check

The most important calibration check: at the end of each semester, pull your LV scores and curriculum grades side by side for each student. If they are consistently and significantly misaligned in either direction across multiple students, that is a signal worth reflecting on — either the digital textbook is not measuring what you think it is, or the meetings need a different structure.

Quick scoring decision tree — Learning Verification

At the end of each section's presentation, ask:

1. Can they explain the core concepts in their own words?
Yes, with depth and connection → C: 8–10 | Yes, basically → C: 5–7 | Partially → C: 3–4 | Barely → C: 1–2 | No → C: 0

2. When I pivoted to something unprepared, did they engage?
Yes, fluently → D: 4 | Yes, with some prompting → D: 3 | Partially → D: 2 | Barely → D: 1 | No → D: 0

3. Do they know where they stand and have a real plan?
Specific gap + specific plan → E: 4 | Vague awareness + vague plan → E: 3 | Minimal awareness → E: 2 | None → E: 0–1

4. Does the overall picture cohere with the curriculum grade?
Yes, fully → F: 2 | Somewhat → F: 1 | Significant gap — flag → F: 0 (document in col H)