Employer guide
How to Mix Question Types in a Structured Screen on the Live Job
One LeetCode-style round rarely answers whether someone can do this job. A structured screen combines multiple signal types — coding depth, knowledge checks, situational judgment, and written reasoning — on the same role context. FeedbackAI supports six employer-facing question types on assessments attached to live postings: Coding, Multiple Choice, True/False, Psychometric (weighted choice), Text, and Essay. Match-ranked applicants and assessment scores inform your pipeline; they do not auto-shortlist without recruiter review.
Last updated 22 October 2026
One coding round is not a structured screen
The familiar failure mode: resume keyword match → single algorithm puzzle → hiring manager still asks whether the candidate can ship in your stack, on-call rotation, and team norms.
Structured screening means multiple evidence types on the same job description — so technical depth, knowledge recall, judgment under pressure, and communication show up in one flow you can defend in a debrief.
Six question types in the employer builder
Labels match what recruiters see when they build a round. Psychometric maps to weighted-choice items in the schema; True/False normalizes boolean prompts to fixed True/False options.
| Type | Employer label | Signal | Grading |
|---|---|---|---|
| Coding | Coding | Problem-solving depth | AI rubric + test cases |
| Multiple Choice | Multiple Choice | Knowledge recall | Auto-graded |
| True / False | True / False | Quick policy or compliance checks | Auto-graded |
| Psychometric | Psychometric | Situational judgment | Weighted partial credit |
| Text | Text | Short written answers | AI rubric |
| Essay | Essay | Long-form reasoning | AI rubric |
When to use each type (employer playbook)
Use Coding for core SWE, AI/ML, and platform roles — keep one or two problems per screen so time stays fair. Stack MCQ and True/False for familiarity checks (security basics, datastore patterns, API conventions) that grade instantly.
Use Psychometric for prioritization and trade-off scenarios — weighted options, not trick questions. Use Text and Essay for incident write-ups, design narratives, or “how would you roll this out?” answers that need rubric factors, not a single correct letter.
Example 45-minute mix (mid-level backend)
Post the role, attach the assessment to an interview round, invite candidates, and review AI evaluations alongside match scores in the same pipeline — no CSV export as the primary workflow.
- True/False × 3 (~5 min) — API security and auth basics.
- Multiple Choice × 5 (~10 min) — Postgres / Redis / queue patterns.
- Coding × 1 (~20 min) — service endpoint with tests in-browser.
- Psychometric × 2 (~10 min) — on-call prioritization scenarios.
How Psychometric scoring works (honest explainer)
Each option carries a weight; candidates earn partial credit instead of a single binary mark. Full credit on a weighted item aligns with the best-weighted option (internally scored at the 99.5% band for “best answer”).
This is employer-built situational content with weighted-choice scoring — not a marketed third-party psychometric inventory. Review AI-generated Psychometric items before publish, same as other generated questions.
Puzzle library vs structured screen on the job
Generic libraries optimise attempts on detached templates. FeedbackAI optimises evaluation on the live role — mixed types in one assessment, applicants ranked by match score, and reports in your hiring pipeline.
For a full compare narrative, see the HackerRank vs AI hiring page and the automated assessment buyer guide.
| Dimension | Generic test library | FeedbackAI structured screen |
|---|---|---|
| Job context | Detached template | Attached to live posting |
| Question mix | Often coding-only | Six types in one flow |
| Psychometric | Rare or add-on | Built-in weighted choice |
| After score | Export or handoff | Pipeline + match-ranked intake |