The Question Stats Report tells you how a question performed in the assessment you are currently viewing. Performance history tells you how that same question has performed everywhere it has been used.
This is useful when you want to judge whether a question is reliably easy or hard, identify questions that behave inconsistently between cohorts, or review question quality across a whole item bank rather than one exam at a time.
Seeing history alongside current results
In the Question Stats report, switch on Performance history. Each statistic is then shown twice: the value for the assessment you are viewing, and the value across all assessments that use the same question.
The toggle appears only when there is history to show — that is, when the question has recorded attempts in at least one other assessment. If a question has only ever been used once, no toggle is displayed.
The toggle is labelled according to the scope of data you are permitted to see:
| Label | What it means |
|---|---|
| Global data value | Performance across every assessment that uses the question, anywhere in your Atomic Assessments instance |
| Across your courses | Performance across only those assessments that use the question in courses you teach |
See Permissions below for how this is determined.
The detail screen
To open the full history for a question, select the Details icon in that question's row. The Detailed Question Performance History screen has two tables.
Overall Performance
One row per version of the question, giving its aggregate performance across all assessments in scope.
Performance Per Assessment
One row per assessment in which the question has been used, identified by assessment name, course short name and term, and the date the assessment was created. The assessment you opened the report from is labelled Current Assessment and pinned at the top of the list. Other assessments are linked, so you can navigate to them directly.
Metrics shown
| Metric | What it tells you |
|---|---|
| Correct | The proportion of responses that were correct. This is the question's difficulty: a high value means most students answered it correctly, so the question is easy. |
| Not Answered | How often the question was left unanswered. |
| Upper 27% | Performance on this question among the highest-scoring quarter of students on the assessment. |
| Lower 27% | Performance on this question among the lowest-scoring quarter. |
| DI — Discrimination Index | How well the question separates stronger from weaker students. Values run from -1 to +1; higher positive values are better. A negative value usually indicates a problem with the question or its key. |
| PBI — Point-biserial Index | The correlation between performance on this question and performance on the assessment overall. |
Filtering, columns and export
- Version filter — restrict the view to a particular version of the question, or include every version.
- Column toggler — show or hide individual metric columns.
- CSV export — both tables can be downloaded separately for offline analysis.
Versions
A question's identity for reporting purposes combines its unique identifier with its version number. By default, each version is reported separately, so that editing a question does not silently merge results from before and after the change into a single figure.
Selecting all versions widens the view to every version sharing the same underlying question, letting you see the full arc of a question's performance across its revisions.
How performance follows a copied question
When a question is duplicated, or carried into another course by a course copy, Atomic Assessments preserves the identifier that links it back to the original. The copy and the original therefore report as the same question, and their results combine.
Questions copied before the 26.07 release do not merge. The identifier that creates the link was only applied to copies made from that release onwards. A question copied into another course before then is treated as a separate question and reports its own history independently. If your item bank predates 26.07 and relies heavily on course copy, contact Atomic Jolt support to discuss your options.
Permissions
Two levels of scope are available:
- All assessments — administrators see performance across every assessment that uses the question, in any course in the instance.
- Your courses only — instructors see performance across only those assessments that use the question in courses they teach.
Because the two scopes cover different sets of assessments, the same question can show different figures to an administrator and to an instructor. An instructor sees the question's performance in their own courses; an administrator sees it across the institution. Neither figure is wrong — they are answering different questions.
What is included in the figures
- Only completed attempts are counted. Attempts in progress are excluded.
- Canvas Test Student attempts are excluded.
- Questions worth zero points are excluded.
- Discrimination Index, Point-biserial, Upper 27% and Lower 27% are not calculated for any assessment with fewer than three scores. These measures compare students against one another and are not meaningful on very small groups. The Correct and Not Answered figures are still shown.
- Scores are normalised to a maximum of one point per question, so aggregate figures are expressed as percentages and remain comparable across assessments that award different point values.