Question Bank
The Question Bank (/question-bank in the authenticated app) is Skilladder’s shared library of assessment items: vetted, tagged questions that can be reused when building live assessments. It sits next to Skills and Assessments in the product philosophy: skills content (what we measure) stays consistent, while assessments (how you assemble and deliver a test) stay flexible.
Why a question bank exists
| Rationale | Plain language |
|---|---|
| Quality & consistency | The same skill should be tested with questions that meet the same bar, use the same taxonomy labels, and align to job families and levels—instead of each author inventing wording from scratch every time. |
| Governance | Items move through draft → review → live (and can be flagged or retired) so Product, psychometrics, or content teams can control what is “in circulation.” |
| Reuse & speed | Authors can pull from the bank into an assessment instead of retyping; usage is tracked so you can see what is actually deployed. |
| Analytics | The Usage Dashboard helps answer “where are unvalidated or flagged items concentrated?” (e.g. by job family or difficulty level) so remediation is targeted. |
:::info Not the same as “writing a one-off question inside an assessment”
You can still add or edit questions inside a specific assessment where the product allows—but the bank is the catalogue intended for shared, maintained content. Items taken from the bank keep a link back to the catalogue for traceability (question bank id in engineering terms).
:::
Who can see what (internal summary)
There are two different ideas: opening the Question Bank page vs operating it like a content admin.
| Persona | Typical access |
|---|---|
| Platform admin | Full access; can create / edit / delete / import / review / bulk status / retire items and use Review Queue and Usage Dashboard as operators. |
| Directory admin (“Content admin”) | Same operator abilities as Platform for the bank (curate global content). |
| Multi‑company admin (MCA) | Can open the bank and browse the Question List only. No Review Queue, Usage Dashboard, or Item Analysis. |
| Company primary admin | Can open the bank and browse the Question List; can pull questions into assessments. No operator tabs. |
| Company staff | Sees the Question Bank menu only if the Question bank (browse) toggle is enabled in Account Settings → Team. Browse and use only — no editing, reviewing, or operator tabs. |
| Staff viewer | Can open the bank and browse the Question List in a read-only way. No operator tabs. |
| Talent / candidate | Never see the Question Bank—participants only take assessments, they do not curate them. |
:::tip Customer conversations
If someone asks “why can’t every HR coordinator edit the bank?” — the answer is governance: publishing and lifecycle are meant to be central so reporting, fairness, and versioning stay under control. Delegation for company staff is deliberately browse (assembly) rather than publish unless they are in a Directory/Platform role.
:::
What you see when you open the Question Bank
This depends on your role:
Operators (Platform and Directory admins) see four tabs:
| Tab | Purpose |
|---|---|
| Question List | Search, filter, browse all questions; create, edit, bulk-update, import via CSV. |
| Review Queue | Questions waiting for a human approve/reject decision before going live. |
| Usage Dashboard | Counts of live unvalidated and flagged questions by job family and level — helps schedule content clean-up. |
| Item Analysis | Psychometric quality metrics after real candidate responses: difficulty (p-value), discrimination (D-index), and automated verdicts. See Item Analysis how-to. |
All other roles with Question Bank access (MCA, company admins, staff with browse permission, Staff viewers) see only the Question List — no Review Queue, Usage Dashboard, or Item Analysis tabs.
Exact filter names and columns can evolve; trust the live UI for demos.
Item Analysis (psychometrics at a glance)
The Item Analysis tab answers: ”How did this question behave when candidates actually took it?”
- Metrics are derived from completed assessment responses (data rail / disassembly pipeline) — not from editorial review alone. Both Build and Match platform assessments feed the same pipeline.
- Each analysed row gets a verdict: Bank It, Monitor, Review, Review Distractors, Mis-keyed, or Remove.
- p-value summarises difficulty; D-index summarises whether the item separates high and low scorers.
- Click a row for a detail view showing the full question text, upper/lower group sizes, and per-option distractor effectiveness for MCQs.
- An Assessment dropdown (always visible) lists every assessment that has response data, identified by name. Selecting one filters the table and reveals an Assessment Reliability Card showing Cronbach's Alpha, KR-20, mean score, standard deviation, and a breakdown of all six verdict counts for that assessment.
- Operators can trigger or refresh analysis directly with the Run Analysis / Re-run Analysis button — no engineering intervention needed once response data is imported.
- The four summary cards at the top always show global totals across all assessments regardless of any filter applied to the table.
If the tab shows no rows, select an assessment and click Run Analysis. If an assessment is missing from the dropdown, its response data has not been imported yet — coordinate with engineering.
Status labels (what they mean for teammates)
These values drive whether an item is safe to expose in high-stakes contexts:
| Status | Plain language |
|---|---|
| Draft | Work in progress—not ready for review or live use in the formal sense. |
| Review | Submitted for scrutiny; expect it in the Review Queue for operators. |
| Live (unvalidated) | In use or promotable for pilot contexts but not yet through full validation—Usage Dashboard highlights concentration risk. |
| Live (validated) | Passed the review bar you have defined operationally—strongest default for “gold standard” content. |
| Flagged | Marked for attention (automation or workflow may set this); should be triaged rather than ignored. |
| Retired | No longer recommended for new assemblies; historical assessments may still reference old copies depending on product behaviour. |
“Flagged” in the edit modal is normally not a status people set casually by hand—it reflects operational signals in the system.
Formats & tagging (high level)
Questions are classified in ways that tie to skills, roles, job families, levels (e.g. difficulty tiers), format, and source lineage (where the content came from—internal, employer engagement programmes, job-description corpora, etc.). This is what makes filters and usage rollups meaningful.
Format on the bank may include classic labels (Multiple choice, Scenario, Short answer, Open-ended) and extended labels such as Multiple correct, Fill in the blank, Ordering, Image MCQ, and Likert. When those items are inserted into an assessment, they become the corresponding runtime question type. Some formats rely on structured config in the catalogue row—see Assessment question types (extended) for what participants can actually answer today.
You do not need to memorise enums for customer calls—use the Filters panel language on screen.
Link to assessments
Inside Assessment authoring, authors with the right permissions can open a question bank drawer / flow to insert catalogue items into that assessment. Usage may be recorded for the bank so Reporting and the Usage Dashboard stay honest about where items travelled.
For a map of the rest of the sidebar, see Navigation map (sidebar).