CLASS-WIDE STATISTICS
What do the cohort's drafted scores show?
Beside the submission drafts, Lectora shows the cohort's score distribution, average and spread for that assignment. The overview comes from the assessment work already being done. These are internal performance scores, not final grades set by an examiner.
Internal score distribution
Illustrative · one assignment · 12 completed drafts
- 0–10: 0
- 10–20: 0
- 20–30: 0
- 30–40: 1
- 40–50: 1
- 50–60: 1
- 60–70: 1
- 70–80: 3
- 80–90: 3
- 90–100: 2
Scores summarize completed drafts before examiner review. They are not final grades.
What is included in the cohort overview?
The panel belongs to one assignment. It shows the total number of submissions, how many have completed AI grading, and how many of those completed results have an internal performance score.
Beside those counts are the average, median and standard deviation, plus a distribution across score bands. These summaries are computed from the available internal 0–100 scores. They can include results that have been reviewed or published, but the measure remains an internal performance score.
What does the score distribution tell you?
The distribution shows how the available scores are spread across the cohort. The average gives one summary; the median and spread help you read it in context.
It is a picture of the assessment drafts, not a grade distribution approved by examiners. A score is also not a confidence estimate. Neither a high score nor a low score tells you whether the model interpreted a particular answer correctly.
Why do the completion counts matter?
A cohort with unfinished results is not yet represented in full. The number of results with performance scores shows which portion of the completed work contributes to the statistics.
Read those counts alongside the chart. The overview lets you see what is available while the submission and its feedback remain the evidence for each individual review.
How does this inform the next step?
Use the distribution to guide your questions, then return to the student work and feedback to investigate them. You decide which findings call for individual follow-up or a topic to revisit with the cohort.
The panel does not produce a category radar, configurable weakness thresholds or overlays of previous cohorts. For a different question, which declared learning goals an assessment does not test, use the approved-question coverage view described under Curriculum.