Reviewing Grade Submission Exception Patterns

Protocol status

Grade submission can be reviewed as a section-level control: the useful evidence is whether each section has a named submitter, a reconciled roster, a documented exception, and a posting confirmation rather than a dashboard status.

This page describes a review design and its proposed controls. It does not report an executed sample, measured outcomes, or evidence that the controls improved performance.

Review protocol

Planned review sequence

1. Define the population from the registrar's official section list for a declared term. Include every section required to submit grades, such as cross-listed, independent study, practicum, and adjunct-taught sections. Record the term, the extract date, the source system, and the count. Exclude sections that were cancelled before the grading window only under a written rule.
2. Record the official deadline and its source. If a school or program sets an earlier deadline, record that date and the authority. The review should not compare sections to a deadline remembered by the office. Keep the calendar version with the review file so a later change can be distinguished from a late submission.
3. Identify the accountable submitter and the person with system access for each section. These may be the same person or different people. Record the instructor of record, the department, and any approved alternate. A section with no named submitter is an exception before the deadline, not only after it.
4. Reconcile the grade roster to the registrar enrollment extract. Record students who appear in one system and not the other, including late adds, drops, audits, withdrawals, and pass/fail conversions. Send the comparison to the registrar for resolution. The review should not treat a roster mismatch as an instructor error until the enrollment record is confirmed.
5. Define submission states before reading the data
6. Identify incomplete and deferred grade records separately. Record the approval, the expected completion date, and the owner of the remaining work. An incomplete is a documented agreement, not a blank grade. A section with several incompletes should still have a submission record for every other student.
7. Measure the time from the deadline to the first complete submission and to the posting confirmation. Use business days and record the source of each timestamp. A section that was entered on time but posted late has a different cause from one that was never entered. Do not collapse the two into a single late label.
8. Stratify the sample by department, section size, delivery mode, and instructor type. Include at least one large lecture, one small seminar, one online section, and one section taught by an adjunct or a substitute. A sample of only large departments may miss the access and ownership problems that affect small programs.
9. Review escalations and responses. Record who was contacted, when, through which channel, and whether the contact produced a status change. Compare the escalation date to the deadline. An escalation that occurs only after the deadline cannot prevent a late posting, even if it eventually resolves the case.
10. Check the posting feed for a sample of students in each completed section. Record the academic history view, the transcript view, and the degree-audit feed where available. If a student's grade is missing from one view but present in another, record the discrepancy and the ticket. The review does not evaluate whether the grade itself was correct.
11. Examine late and partial sections for cause categories
12. Test the joins between the section list, the gradebook, the approval workflow, and the posting feed. Record duplicate section identifiers, missing instructor-of-record keys, and cross-listed sections that could be counted twice. Report the match rate before sampling. A section that cannot be joined is a data exception and should be listed, not silently dropped from the denominator.
13. Pilot the state definitions on one college or department. Record the cases where entered, approved, and posted were ambiguous, and where an incomplete grade looked like a missing value. Adopt the state model once the pilot is complete and do not add a category mid-review. Keep the pilot cases separate so the final counts use one state model.
14. Define decision rights for a late or corrected submission. Name who can reopen a posting batch, approve a retroactive grade, or extend a deadline for a whole section. Record the authority and the date. The assistant can prepare the evidence and the section list, but should not grant an extension because a review found a late section.
15. Review the student-facing grade display for a sample of completed and incomplete cases. Record whether the student sees the correct term, course, and status, and whether an incomplete is explained rather than shown as a blank. A posting that is technically correct but confusing to the student still generates support work and can obscure a real error.
16. Set the reporting audience and cadence. The registrar needs section counts and control gaps, departments need their own exceptions, and student services need to know which messages are going out. Report aggregates without ranking individual instructors. Set a follow-up owner and a review date so the report does not end as a one-time exercise.
17. Check the deadline change log for the term. If a deadline moved, record the old date, the new date, the authority, and the sections notified. A late section measured against a revised deadline may be on time. Cases decided under different deadline versions should be reported separately rather than pooled.
18. Document the sample selection record. State the frame, whether the sample was random or a census, and how replacements were handled. Keep the selection file. A list of sections that happened to be discussed in meetings is not a sample of the population.
19. Test the correction trail. For any grade that was changed after posting, record the original value, the new value, the approver, the reason category, and the date. Confirm that the academic history and the student view changed together. A correction that is visible only in the gradebook is not complete.
20. Do not let the review drift into judging an individual grade or an instructor's intent. The review does not decide whether a grade was correct, whether an instructor was late without cause, or whether a student deserved an incomplete. Those are academic judgments. The review reports whether the submission and posting controls can be reconstructed.
21. Report the population, the sample, the state counts, the time-to-post measures, the exception causes, and the limits. Retain the section list, deadline source, roster comparison, escalation log, and posting confirmations. The protocol measures submission and posting controls; it does not measure grading quality or the appropriateness of an individual grade.

Operational implications

The section, not the instructor's inbox, is the useful unit for a submission review.

The deadline should be traced to an official calendar version before lateness is counted.

Accountable ownership and system access are separate fields and should be recorded separately.

A roster mismatch must be resolved by the registrar before it is treated as an instructor error.

Entered, approved, posted, and corrected are distinct states with different evidence.

An incomplete is a documented agreement, not a missing grade for the whole section.

Time to entry and time to posting should be measured separately.

Escalations that begin after the deadline cannot prevent a late posting.

A sample should include small, online, and adjunct-taught sections, not only large lectures.

Unknown causes should be reported honestly rather than replaced with a plausible guess.

The review measures submission controls, not the quality or fairness of the grades themselves.

A revised deadline must be versioned, or a late section will be measured against the wrong date.

Unmatched sections should be reported as data exceptions rather than dropped from the denominator.

The count should remain reproducible from the retained files.