Reports and exports¶
Open Reports¶
Open an exam and select Reports. The page is pinned to the displayed Administration version and contains Student reports, a searchable/selectable result table, and Class performance.
Unmatched packets with scanned work appear in the table by packet number. You can export their grades with Export report CSV and preview their reports once grading is complete. The note at the bottom left of Student reports links to Scans, where you can match packets to students. Matching is required before releasing a report or exporting its grade to an LMS.
Packets with an unconfirmed AI name suggestion remain in this unassigned-packet flow. A suggestion does not create an official student record, release or email a report, or enable LMS export. Archived packets and unresolved packet conflicts are excluded. These rules also apply to experimental existing-PDF exams.
Each student row shows question scores/actions, earned total, final percentage, report state, email/release state, and a selection checkbox. A question cell can link directly to free-response review, provisional approval, re-review, auto-graded recognition, or a missing prerequisite.
Adjust exam scores¶
Open the hamburger menu at the bottom right of Student reports and choose Adjust exam scores…. These settings apply to every completed student exam in the displayed administration, including exams completed later. Other administrations and per-question grades stay unchanged.
- Earned points adjustment adds or subtracts points from each completed total.
- Points possible adjustment adds or subtracts points from the exam's denominator.
- Allow scores above points possible starts checked. Uncheck it to cap final scores at 100% of the adjusted denominator.
Both adjustments accept signed numbers with up to two decimal places. Enter 0 in both fields to remove point adjustments; check the above-total option to remove any cap. Scores cannot fall below zero, and adjusted points possible must be greater than zero.
The modal shows an example before saving. For example, a student earning 80 out of 100 with adjustments of +5 and −5 receives 85 out of 95, or 89.47%. Any zero floor or upper cap is applied before calculating the percentage.
Select Save adjustments to update the table, Gradebook, exports, and total-score statistics. The black boxed A symbol beside Total explains the settings on hover, keyboard focus, or tap. Cancel discards unsaved changes. Incomplete grading remains incomplete.
Saving does not release reports or email students. Existing student links retain the previously released scores. The instructor Reports workspace identifies unreleased changes until you release updated reports. Updated reports list each adjustment and any applied floor or cap in separate rows before Total in the questions table.
Free points, bonus questions, and caps¶
Adjustments apply once to the whole exam total, not once per question or only to selected table rows. Filtering or checking student boxes does not limit who receives them. Instructors can manage these settings; grader permissions do not grant access to Reports or score adjustments.
| Goal | Settings and result |
|---|---|
| Give everyone 5 free points | Earned adjustment 5, possible adjustment 0: a completed 80/100 becomes 85/100. |
| Count 10 of the published 100 points as bonus opportunities | Earned adjustment 0, possible adjustment -10: a completed 80/100 becomes 80/90 (88.89%). This changes the denominator; it does not designate or rescore a particular question. |
| Cap adjusted scores at 100% | Uncheck Allow scores above points possible. With a denominator of 90, an adjusted earned total of 95 becomes 90/90. Leaving it checked gives 95/90 (105.56%). |
| Subtract points | Use a negative earned adjustment. A completed 3-point total with adjustment -5 becomes zero, never −2. |
| Restore ordinary totals | Set both adjustments to 0, check Allow scores above points possible, and save. |
The denominator is the sum of published gradeable-question points plus the points-possible adjustment. Gradebird adds the earned adjustment to the student's completed question-score sum, applies the zero minimum and optional cap, then calculates the percentage to two decimal places. Per-question scores, maximums, rubrics, and answer keys remain unchanged. Question scores can therefore sum to a different number from the displayed adjusted Total.
The dialog shows Original points possible, Adjusted points possible, and an example calculation before Save adjustments. A denominator of zero or less cannot be saved; correct the field error and try again. Saving refreshes totals, release status, and statistics while retaining table search and ordering. The boxed A beside the Total column heading appears when either adjustment is nonzero or the cap is enabled, even if no student's score hits the cap.
Where adjusted totals appear¶
Live Reports totals and percentages, course Gradebook, detailed report CSV, LMS earned-point exports, and exam-total statistics use saved adjustments without requiring report release. Class-performance question rows remain on their published scales. The Total average and distribution use each complete student's adjusted score, applying floors and caps before averaging. Scores above 100% remain visible in distributions when allowed.
New experimental student analyses use adjusted exam totals and adjustment settings as evidence; earlier saved analyses remain snapshots. Generate another analysis to reflect updated totals. Already released student reports also remain snapshots until released again. Preview the new report before releasing; rows immediately above Total explain nonzero earned/possible adjustments and any floor or cap actually applied.
Settings belong to one administration version. Publishing a new version starts with zero adjustments and scores-above-total allowed; check settings when changing administration versions. Course copies preserve each source version's settings independently.
Report readiness¶
A report is ready only for a confirmed canonical packet with a current completed result for every gradeable question:
- each free-response question has an approved final against the current canonical rubric, including an explicit missing-work final when appropriate;
- each auto-graded question has a current definitive recognition and a current score against the current key;
- no unresolved packet or late-scan conflict blocks release.
Unsaved, provisional, revoked, and older-rubric approved assessments are not complete. A human-edited provisional score is still pending even when AI candidates originally agreed. Saving as approved against the current rubric completes the approval step; saving as provisional removes any existing final approval and can make live report readiness incomplete again. Revocation retains the saved assessment for review. These changes do not rewrite an already released report snapshot. See grading states and transitions. An unresolved student packet conflict keeps the student in missing-scanned-work status until a canonical packet is chosen.
Report states¶
| State | Meaning |
|---|---|
| Ready | Current grading can be previewed and released; no report has been released for it yet. |
| Released | The released report matches the current grading state. |
| Unreleased changes / Prior version released | The student link still shows the older report, but grading, score adjustments, rubric, key, or scanned work has changed. Preview the new report and release an update. |
| Failed release | The report could not be released; inspect the detail and retry after resolving the cause. |
| Grading incomplete | Scanned work exists but at least one gradeable question lacks a current completed result. |
| Missing scanned work | No confirmed detected-work record exists, or packet conflict remains unresolved. |
Preview and release¶
Choose Preview report to open the current report in a new tab. Preview does not release the report or email the student. If an older report has already been released, View old report and Preview new report allow comparison.
Select ready rows, choose Release reports in Action, and select the release button. Students with distinct nonblank roster emails are automatically sent their report link. A report for a student without email can still be released; use Copy student link to deliver it separately.
The student link remains the same when an updated report is released. A released report contains question and total scores, published questions, student work, the rubric and assessment for free-response questions, and the recorded and correct answers for auto-graded questions. Use browser print to create a PDF copy when needed.
There is no separate manual resend control in the Reports table. Releasing a new or updated report sends the email automatically.
Export detailed report CSV¶
Select rows, choose Export report CSV, and export. This CSV uses current grading and does not require released reports. Columns contain student, sections, question labels, adjusted earned total, and final percentage. A complete question contains its score. An incomplete question instead contains its current action or status, such as Provisional 4.00, Grade, Rubric, Review, or an auto-graded status. Total and final percentage remain blank until the row is complete.
Export to the LMS¶
The Reports and Gradebook table-header checkboxes select or clear eligible rows on the visible page while retaining selections on other pages.
If the course has an LMS format, choose Export to Canvas or Export to D2L Brightspace. LMS export uses current completed canonical-exam grades, including unreleased reports. It is therefore a grade-transfer workflow, not a report-release workflow.
- D2L Brightspace: identity/name/email columns plus exam columns ending in
Points Grade; values are adjusted earned points. This format cannot update the maximum points configured in D2L; update it there when changing the denominator. To update an existing D2L grade item, the exam name in Gradebird must exactly match the grade item name in D2L. Rename the exam in Gradebird before exporting if needed. Gradebird adds thePoints Gradesuffix in the CSV automatically. - Canvas: Canvas Gradebook-style identity columns regardless of accepted Canvas roster source, one column per canonical exam label, and a
Points Possiblerow using the adjusted denominator; values are adjusted earned points.
Students added manually require the course's LMS identity field. Review identifiers, points possible, and a sample row in both the downloaded file and the destination LMS preview before committing an import.
Gradebook¶
Open the course-name menu and select Gradebook to compare active students across canonical exams. Cells link back to the corresponding Reports page with that student search prefilled. Select rows to export the configured LMS format. If no LMS is selected, Gradebird instead exports a general CSV with student, email, LMS student ID, sections, and exam-grade columns; this file is not formatted for direct LMS import. Released and unreleased grades are visually distinguished, but every completed grade is eligible for the gradebook/export regardless of report release.
Class performance¶
The table contains one row per gradeable question, with points, approved coverage, average, distribution, and eligible AI analysis. Context-only multipart parents and provisional/revoked assessments are excluded. The Total row includes only confirmed exams with a completed approved result for every gradeable question.
Open a distribution control to see anonymous score data. Ordinary cohorts use stacked score dots; larger cohorts use a histogram. Accessible descriptions state population, range, and average.
For a free-response question, an authorized instructor can request an AI performance summary from its icon. If the question, rubric, course context, or finalized results change, the prior summary is marked out of date and remains viewable until an update is requested. Auto-graded questions have numeric statistics but no AI analysis action.
Reports and exports¶
The Reports page shows readiness, preview, release, delivery, CSV/LMS export, and class-performance information for the displayed administration version. Instructors can filter records, preview a student report, release eligible reports, generate secure access, and export documented data. Preview and selection alone do not release anything.
Student report preview¶
This instructor-only preview renders the report that current grading state would produce for one student record. Use it to inspect question results and evidence before release; it does not create a public link, send email, or mark the report released.
Released student report¶
This instructor view renders the currently released report snapshot for one student record. It may identify unreleased changes made after that snapshot. Viewing it does not publish an update or notify the student.
Course analysis¶
The course Analysis tab and course-menu entry are available to instructors and organization administrators. Exam performance and Student surveys do not require experimental-feature access. The Student reports card remains experimental and appears only when experimental features are enabled on your account.
The Student reports card contains saved reports and a form to generate a new one. Reports are listed newest first by student, included exams, and creation date, with 20 reports per page. Click Next reports or Previous reports below the list to browse history without clearing your new-report selections. Expand View report to load and read a completed report. If loading fails, close View report and open it again to retry. Generating or reusing a report displays the history page containing that report.
Under Generate a new report, start typing a name in Student, choose a matching roster entry, select one or more exams, and select Generate report. There is no separate apply-selection step. The new report appears in the same card as Queued or Generating, and updates automatically when ready. Use Retry report after a failed attempt. Submitting the same selection with unchanged grading data and model settings reuses the existing report.
Gradebird uses the course's ordered AI models to analyze finalized scores and feedback. Only exams for which that student has a complete final score are included. At least one fully graded, confirmed exam and an active course AI model are required. An exam with unfinished grading, provisional assessments, pending rubric or answer-key re-review, or unresolved scan conflicts is excluded entirely; its partial feedback is not sent to the AI.
Each saved report reflects the grading data available when it was generated. To use updated grades or settings, generate another report using the form; earlier reports remain available. Reports created under the earlier partial-exam behavior are labeled accordingly. A note identifies use of a backup model. These experimental reports describe evidenced strengths, difficulties, and possible instructional next steps for instructor review. They do not change grades, release student feedback, or send email. Saved experimental student analyses are excluded from course copies; generate them again in the copied course when needed.
The separate Exam performance card compares point totals, fully graded coverage, class averages, and score distributions across published exams' current administration versions. Select a distribution thumbnail to enlarge it. Class averages include only confirmed students with approved work for every assessable question; missing or incomplete grades are not counted as zero.
Anonymous student surveys¶
On Analysis, the Student surveys card lets you create multiple copies of one standardized survey. Each copy has its own private instructor title and student-facing introduction. The title is for your reference only and is not shown to students, including in emails. Students see the Gradebird logo, the course name as the main heading, the anonymity explanation directly below it, and numbered questions. The introduction appears as a highlighted note from you in invitation and reminder emails and above the survey questions. Use the introduction to identify an exam, suggest questions to think about, or invite participation. All questions are optional; students can skip any question. Each survey retains its exact questions and answer choices, so later changes to the standard survey do not change surveys already created. Results show answered and skipped counts, with percentages based on answers to each question. The three numbered questions ask how helpful the comments and explanations were, how clearly grading explained the score, and how the overall grading experience compared with typical grading. An unnumbered free-response prompt follows: “Please share any thoughts or feedback about your experience with Gradebird.” View standard questions shows the complete questionnaire.
The survey table lists titles, dates sent, and status. Select the blue + New survey action below the table, enter a required Private survey title (up to 150 characters) and optional Introduction for students (up to 3,000 characters), and choose Create draft survey. Select a survey title to expand its editing controls or results. View standard questions shows the numbered questions and available answer choices. Select Save introduction before sending. Send survey now shows a confirmation and sends one personal link to each distinct email address among active roster students. Students without an email address are skipped. Email sending runs in the background; large surveys use personalized batches, and each student receives only their own private link. Sending freezes the title and introduction; create another survey for another exam or revised wording.
Each link accepts only one response and requires no student account. Students may select a rating for any of the three questions, enter an optional comment of up to 4,000 characters, and select Submit anonymous feedback. Even an entirely blank response can be submitted. The confirmation says Thank you for your feedback.; the submission cannot be edited or repeated through the same invitation. Answers are stored separately from invitation delivery records, without names, emails, or a link back to the recipient. The free-response field asks students to avoid identifying details in comments. You cannot see which student submitted a response.
New surveys offer these rating choices, ordered from most to least favorable:
- Helpfulness: Extremely helpful, Very helpful, Moderately helpful, Slightly helpful, Not at all helpful.
- Clarity: Very clearly, Clearly, Somewhat clearly, Not very clearly, Not at all clearly.
- Overall experience: Much better, Better, About the same, Worse, Much worse.
Surveys are listed by their send date. Expand one to see invitation delivery totals, response counts, a bar chart for each rating question with full answer labels and counts (hover for percentages), and anonymous comments. Sentiment totals show Positive, Neutral, Negative, and Not classified. The organization's ordered clerical models classify comments; blank comments are skipped, and failed classifications are not counted as neutral. Configure a clerical model and select Retry comment classification if necessary. Review the comments themselves: AI labels may be incorrect.
Retry invitation delivery sends only invitations whose delivery was not confirmed. An interrupted email delivery can cause the same link to arrive again; it still accepts only one response. Reopen the survey details to fetch new responses. Delivery and classification results update while jobs are active. Close survey stops further submissions and retains the existing results. Surveys, invitations, and responses are not carried into course copies.
A closed survey can be opened again with Reopen survey. Its existing responses are preserved, and its original invitation links work again for students who have not responded. Reopening does not send email.
For an open survey, Send reminder shows how many unanswered invitations can receive email and asks for confirmation. Reminders go only to nonrespondents whose original invitation email still matches an active roster email. Original and reminder links share the same one-response limit. Students who respond while reminders are queued are skipped before delivery. Each student can receive at most three emails for a survey (the original invitation plus up to two reminders), at least 24 hours apart. Delivery attempts, including uncertain failures, count toward the limit. The delivery card shows the original email date and most recent reminder date. You can send another reminder when eligible; Retry invitation delivery retries unconfirmed deliveries after a failure.