Answer first: use single-select for one defensible answer, multi-select for a complete set, short text for unaided recall, ranking for relative order, scales for degree, and polls or word clouds for opinion and idea collection. Add quiz scoring only when correctness or competition is part of the decision. A format is useful only if its validation, timing, feedback, and report behavior match the task.
Key findings, verified 23 September 2026
- Kahoot! documents materially different scoring by type: standard single-select can award up to 1,000 points, multi-select awards up to 500 per correct option with no partial credit, slider scoring weights precision more than speed, and brainstorm scoring depends on votes.
- Kahoot! live sessions can use Accuracy mode for correctness-only scoring; assigned sessions can disable the question timer. A format label alone therefore does not establish how the response will be scored.
- Mentimeter separates polling formats—multiple choice, word clouds, scales, ranking, and open-ended—from quiz presentations built around correct answers, scores, and a shared leaderboard.
- Slido documents multiple choice, word cloud, ranking, rating, open text, quiz, and survey workflows. Those interaction types do not all produce equivalent evidence.
- Socrative and Wayground reports demonstrate why authoring and reporting must be checked together: typed text, correctness, time, score, and manual-evaluation states can appear differently after participation.
What is a quiz-app question type?
A question type is the combination of what a participant can submit and how the system interprets that submission. The visible response control—radio buttons, checkboxes, text box, slider, ranking list, or voting panel—is only the response shape. The scoring rule decides whether a response is correct, partly correct, close enough, popular, or ungraded. The delivery mode decides whether it is answered live, asynchronously, individually, or as part of a shared discussion.
Keeping those layers separate prevents common comparison errors. A multiple-choice poll and a multiple-choice quiz may look identical to the participant while producing different feedback, leaderboards, and reports. Likewise, two multi-select tools can disagree on partial credit, and the same tool can change its point calculation when the host changes the session mode.
Question type decision matrix
| Evidence needed | Best starting format | What it establishes | Main failure | Report check |
|---|---|---|---|---|
| One defensible answer | Single-select or true/false | Recognition among supplied choices | Distractors reveal or cue the answer | Choice, correctness, skipped state, and timing |
| A complete set | Multi-select | Recognition of all required items | Unknown all-or-nothing or partial-credit rule | Each selected option plus the awarded score |
| Unaided recall | Short text | Ability to produce an answer without choices | Spelling, capitalization, or synonym rules distort grading | Original text, normalized value, and manual-review state |
| Numeric estimate | Number or slider | Exact value or proximity to a target | Unclear tolerance and rounding | Submitted value, target, margin, and precision score |
| Relative order | Ranking | Priority or sequence across supplied items | Forced order implies differences participants do not hold | Full order, ties if allowed, and aggregation method |
| Degree or confidence | Scale or rating | Position on a defined continuum | Unlabeled endpoints make responses ambiguous | Scale labels, value, and distribution |
| Ideas or themes | Open text, word cloud, or brainstorm | Participant-generated language and recurring terms | Popularity is mistaken for correctness | Original entries, moderation, voting, and grouping |
| Opinion or choice | Poll | Preference among stated options | A majority result is treated as knowledge evidence | Response counts, anonymity mode, and non-response |
Platform format and scoring matrix
| Platform | Documented formats or workflows | Evaluation behavior established by the source | Boundary |
|---|---|---|---|
| Kahoot! | Single-select, multi-select, slider, brainstorm, live and assigned play | Standard speed formula; no partial credit for multi-select; precision-led slider scoring; vote-led brainstorm scoring; live Accuracy mode | Question type, host setting, and delivery mode can change the result |
| Mentimeter | Multiple choice, word cloud, scale, ranking, open-ended, poll, and quiz presentation | Polls collect and display responses; quizzes add correct answers, scores, and leaderboard behavior | Do not describe all interactive slides as scored quiz questions |
| Slido | Multiple choice, word cloud, ranking, rating, open text, quiz, and survey | Quiz and polling workflows are distinct from Q&A and survey collection | The category page does not establish every plan gate or report field |
| Socrative | Quizzes, Exit Tickets, Space Races, and Quick Question short answer | Reports can include graded responses, scores, question correctness, and full short-answer text | The cited reporting source does not enumerate every authoring control |
| Wayground | Live and assigned sessions with question and participant reports | Reports expose accuracy, score, completion, time, and manual-evaluation states where applicable | The cited reporting source establishes output fields, not a complete format catalog |
Sources: Kahoot! points and scoring, Mentimeter live polling, Mentimeter quiz presentations, Slido interaction types, Socrative reports, and Wayground reports.
Why does Kahoot! scoring change the meaning of a question?
Kahoot! provides the clearest current example of why a format list is insufficient. Its official points guide, updated on 23 September 2026, says a standard correct single-select response can award up to 1,000 points. Multi-select can award up to 500 points for each correct option, but selecting even one incorrect option makes the entire response worth zero. That is not a partial-knowledge measure unless the host deliberately accepts the all-or-nothing rule.
The same guide says slider scoring is based primarily on precision, with a smaller speed component and a configurable answer margin. Brainstorm points depend on peer votes rather than response time. Standard live questions normally use a speed formula, while Accuracy mode awards points for correctness instead; an assigned kahoot can remove the timer. The practical lesson is to record the exact type, point setting, timer, and delivery mode before interpreting a score.
When should a poll stay unscored?
Use an unscored poll when the facilitator needs preferences, confidence, priorities, or discussion material rather than a right answer. Mentimeter documents multiple choice, word cloud, scale, ranking, and open-ended polling with results shown in real time. It separately describes quizzes as interactive competitions with correct-answer and leaderboard behavior. That separation is useful: a confidence scale can guide instruction without turning uncertainty into a low score.
Slido likewise documents multiple interaction types—among them rating, ranking, word cloud, open text, multiple choice, quiz, and survey. The label chosen in the authoring interface does not determine the meeting goal. Write the decision first, then choose whether the result should be graded, discussed, moderated, aggregated, or kept anonymous.
How should short text and manual evaluation be tested?
Short text reduces recognition cues, but automated grading can introduce false negatives. Test case, punctuation, accents, common misspellings, abbreviations, singular and plural forms, and more than one defensible phrase. If manual review is possible, check whether participants see a provisional state and whether a later correction updates the score and export.
Socrative's report documentation establishes that full-text short answers can appear alongside graded responses, scores, correct answers, and question-level results. Wayground documents question reports with accuracy and response time plus manual-evaluation behavior for relevant responses. Those pages support a reporting check; they do not prove that every public plan offers the same authoring or evaluation controls.
Worked example: a safety refresher
A trainer must determine whether participants can identify every required shutdown step, recall the emergency number, and flag an unclear procedure. One question type cannot measure all three needs.
- Use multi-select for the complete shutdown set, but disclose and test whether one wrong choice cancels all credit.
- Use short text for the emergency number so recognition choices do not cue the answer; normalize safe formatting variants.
- Use an unscored open response for the unclear procedure so participants can describe ambiguity without losing points.
- Inspect the report to ensure selected options, original text, correctness, skipped responses, and manual-review states remain distinguishable.
The final score should cover only the defensible knowledge checks. The open-response theme belongs in a follow-up workflow, not in a competitive leaderboard.
Eight-field question-type audit
- Author label: record the exact name the platform gives the type.
- Response shape: capture what the participant can actually submit.
- Allowed responses: record counts, character limits, ordering, ties, ranges, and skip behavior.
- Validation: test capitalization, synonyms, tolerance, duplicates, and conflicting selections.
- Scoring rule: record correctness, partial credit, precision, popularity, point weight, and zero-score conditions.
- Timer: test whether time changes access, score, or both.
- Feedback timing: record when answers, explanations, totals, and rankings become visible.
- Report representation: confirm that original responses, evaluation state, score, and session mode survive into the report or export.
Run the audit with synthetic content before a real class, meeting, event, or training session. Save the host settings with the result because a question label alone is not enough to reproduce the evidence.
Common question-type mistakes
- Using a poll to claim knowledge mastery because one option received the most votes.
- Calling a multi-select item partially scored without testing the platform's zero-score conditions.
- Using speed points when careful reading or accessibility matters more than reaction time.
- Treating a word cloud's visual prominence as a precise frequency table.
- Using a forced ranking when ties or independent ratings would better represent the decision.
- Accepting typed answers automatically without testing safe spelling and formatting variants.
- Publishing a score without preserving the settings that produced it.
Method and limitations
We separated response shape, validation, scoring, timer, feedback, delivery, and report output, then checked current first-party product and support documentation on 23 September 2026. The matrix compares documented behavior and evidence boundaries; it is not an overall product ranking. A cautious cell means the cited public source did not establish a broader claim.
We did not purchase subscriptions, execute authenticated cross-plan tests, run a classroom or workplace study, conduct a psychometric validation, measure learning outcomes, benchmark latency, or perform an accessibility-conformance audit. Formats and settings can vary by plan, account type, workspace policy, activity mode, region, and product release.
Practical next step
Write the evidence you need in one sentence, choose the smallest response shape that can produce it, and run the eight-field audit in the two finalists. For post-session fields, pair this matrix with the reporting and export matrix; for opinion versus correctness, use the quiz-versus-polling guide.
Correction path: report a changed format, scoring rule, timer behavior, feedback state, or report field through the corrections process with the exact claim and current first-party source. Verification date: 23 September 2026.