Before choosing colors or building a form, write the answer-to-result mapping. A useful personality quiz maker should let you explain why a result appeared and what happens when several results tie. For this example, the question is deliberately modest: Which project format would you like to try? The result suggests a next activity from the choices made in this round. It does not establish a personality type, skill level, or suitability for a job.
Use the original question bank, mapping ledger, three result messages and explicit tie rule to inspect the local HTML example. The fictional answer sets below are worked examples. Actual participant responses remain blank.
Choose an outcome you can explain
Use three activity formats:
| Outcome ID | Suggested format | Small next step |
|---|---|---|
| W | Short written explanation | Write one paragraph explaining an idea to a reader. |
| V | Annotated diagram | Draw one diagram and label the important relationships. |
| P | Paper prototype | Make one tabletop paper model and try its intended action. |
These labels describe possible work products. A participant can choose another format afterward. “Try a written explanation” is a useful invitation; “You are a verbal thinker” introduces a claim the question bank does not establish.
The mapping distinction matters. Typeform's official comparison describes outcomes linked to answers separately from endings selected through numerical score ranges. Our example uses answer routes to suggest an activity. There is no correct-answer key and no accuracy score. Typeform's outcome and score quiz comparison.
Write the bank before the screen
Call this authored bank version 1. Give every question and answer a stable ID. The label is for the reader; the ID is for the mapping. Rearranging the visible choices must not change their destinations.
| Question ID and wording | Answer ID | Visible choice | Route |
|---|---|---|---|
| Q1 — How would you like to start? | Q1-W | Outline ideas in words | W |
| Q1 | Q1-V | Sketch relations | V |
| Q1 | Q1-P | Try a paper mechanism | P |
| Q2 — What would you like to produce? | Q2-W | One paragraph | W |
| Q2 | Q2-V | One labeled diagram | V |
| Q2 | Q2-P | One tabletop prototype | P |
| Q3 — How would you like to review it? | Q3-W | Read and revise wording | W |
| Q3 | Q3-V | Compare label placement | V |
| Q3 | Q3-P | Try the action and adjust | P |
For each question, select exactly one of its three declared answers. Each selection contributes one count to its mapped outcome. No question has extra weight. The counts start at zero when computing a complete, valid submission; a missing answer does not become a zero-value response.
Keep the bank and mapping ledger together in the native authoring document. Record an author version, the tie policy, and the result copy beside them. The local HTML reading copy follows that record. Editing a label may change meaning even when its ID stays the same, so review wording changes as well as mapping changes.
Trace a result from three selections
Consider this fictional complete submission:
| Question | Selected answer ID | Added count |
|---|---|---|
| Q1 | Q1-W | W +1 |
| Q2 | Q2-W | W +1 |
| Q3 | Q3-V | V +1 |
The resulting ledger is W = 2, V = 1, P = 0. W has the unique highest count. Show this message:
Try a short written explanation. Two of your three choices in this round mapped to writing. Start with one paragraph that explains your idea to a reader. You can still choose a diagram or paper prototype if either would help your project.
The explanation identifies the evidence and its scope. It does not say “67% writer,” translate two selections into confidence, or claim that someone will perform better in that format. Two is a route count under this authored rule.
A second fictional submission, Q1-W / Q2-W / Q3-W, produces W = 3, V = 0, P = 0. The same activity suggestion can say that all three choices mapped to writing. Even unanimous selections describe this round's choices, not a permanent trait.
For a unique V result, use: “Try an annotated diagram. [Count] of your three choices in this round mapped to a diagram. Draw one diagram and label the important relationships. You may choose another format.” For a unique P result, use: “Try a paper prototype. [Count] of your three choices in this round mapped to a prototype. Make one tabletop paper model and try its intended action. You may choose another format.” Replace [Count] with the validated count, either two or three for a unique winner in this bank. Keep the same evidence-based structure for W.
Make a tie visible
Now use Q1-W / Q2-V / Q3-P. The ledger is W = 1, V = 1, P = 1. All three outcomes share the highest count.
Show “Your choices are evenly split. Choose a format to try.” Then present the written explanation, annotated diagram, and paper prototype as three coequal options, each with its small next step. Do not call W the winner because it appears first in the data or on the page.
This is an authored policy, not a universal rule for quiz software. Typeform's Match quiz documentation describes adding a point to a related ending and displaying the highest-scoring ending. It also documents a first-ending default for ties. Our bank deliberately uses a different tie policy: every outcome tied for the highest count appears without a winner badge. Typeform's Match quiz mapping and tie behavior.
Change the visual order to P, W, V as a review case. The tied set must still contain P, W, and V, with equal treatment. A display order is necessary for a list; it is not a ranking.
Hold incomplete or invalid submissions
Q1-W / Q2-V / unanswered is incomplete. Its two available selections are not a finished result. Show which question needs an answer and keep the outcome area pending.
An unknown ID, such as Q2-X, is invalid. So is Q1-W supplied as the answer to Q2. Validate membership in the specific question's bank before counting. Do not silently drop an unknown answer, guess its route from its label, or replace it with the first choice.
The local HTML example has three labeled radio groups, a submission control, and a Reset control. Submit would validate the full current selection set before producing counts and result copy. Any answer edit would clear the accepted result and mark it pending until the person submits again. Reset would clear all selections, counts, messages, and the accepted result; it would not preselect a new “personality.”
A bank or mapping revision also invalidates an accepted result. Keep the earlier authored version in your native history, but do not reuse its result as evidence for the revised bank. The local example illustrates this mapping; actual participant responses and native authoring history remain separate.
Keep this actual-observation record separate from the fictional examples:
| Actual record field | Value |
|---|---|
| Author version used | |
| Q1 / Q2 / Q3 selected answer IDs | |
| Validated counts and result | |
| Interface and host tested | |
| Keyboard, edit, tie, and Reset observations |
The empty cells are intentional. A worked answer set does not fill an actual response or establish a passed interface test.
Review the prepared HTML against the bank
NOS can save and render prepared HTML for review. Put the question bank, the W = 2 / V = 1 / P = 0 worked result, and the three-way tie expectation near one another so a reviewer can compare the explanation with the authored mapping. Keep the native question bank authoritative.
That review has a concrete purpose: check whether the visible result says what the ledger supports. The writing result should mention two selections, and the tie copy should preserve all three options. Saving and rendering a page does not establish that radio controls, validation, counting, focus behavior, or Reset work in the host. Check those functions in the intended host before using the quiz with participants.
The first screen should state the activity question, explain that the result is a suggestion, and show that all three questions are required. On a narrow screen, stack the answer groups and result options with readable labels. During later testing, use keyboard navigation, verify visible focus, and ensure missing-answer feedback identifies the affected question. Actual production article width and visual reference comparison remain to be checked.
Copy this prompt to prepare the authoring package
This prompt preserves the original question bank and design contract. It does not request an executed quiz, collected responses, or a published page.
Prepare an original English authoring package for a playful quiz titled
“Which project format would you like to try?”
Purpose:
Suggest an activity format from declared choices in this round. Do not
measure personality, skill, psychological validity, career suitability,
classroom ability, confidence, or a correct-answer score.
Author version: 1.
Outcomes:
W = short written explanation; next step: write one paragraph explaining
an idea to a reader.
V = annotated diagram; next step: draw one diagram and label relationships.
P = paper prototype; next step: make one tabletop paper model and try
its intended action.
Use this exact question bank and mapping:
Q1: How would you like to start?
Q1-W: Outline ideas in words -> W
Q1-V: Sketch relations -> V
Q1-P: Try a paper mechanism -> P
Q2: What would you like to produce?
Q2-W: One paragraph -> W
Q2-V: One labeled diagram -> V
Q2-P: One tabletop prototype -> P
Q3: How would you like to review it?
Q3-W: Read and revise wording -> W
Q3-V: Compare label placement -> V
Q3-P: Try the action and adjust -> P
Contract:
Require exactly one valid answer belonging to each question. Only a
complete valid submission produces a result. Each answer adds one count
to its declared outcome. Do not weight questions or infer destinations
from visible labels. Preserve stable question, answer, and outcome IDs.
Unknown IDs, answers belonging to another question, missing answers, or
multiple answers for one question must produce no accepted outcome.
Never fill missing responses with zero or a default choice.
Result rule:
Show the outcome with the unique highest count. Explain the count as
selected answer routes, not a probability, percentile, trait, or grade.
If several outcomes share the highest count, show every tied outcome
as a coequal option with its next step. Never use array order or visual
order as a tiebreaker. Keep result copy invitational.
Fictional expected cases:
Q1-W / Q2-W / Q3-V -> W = 2, V = 1, P = 0; suggest W.
Q1-W / Q2-V / Q3-P -> W = 1, V = 1, P = 1; show all three equally.
Q1-W / Q2-W / Q3-W -> W = 3, V = 0, P = 0; suggest W.
Q1-W / Q2-V / unanswered -> pending, no outcome.
Q1-W / Q2-X / Q3-P -> invalid, no outcome.
Q1-W / Q1-W supplied for Q2 / Q3-P -> invalid, no outcome.
Display order P, W, V must preserve the same three-way tied set.
Proposed interface, not an implemented artifact:
Three labeled radio groups, Submit, Reset, a count ledger, and result copy.
Validate before counting. An answer edit invalidates the accepted result
until another valid Submit. Reset clears selections, counts, messages,
and the accepted result to pending. A bank or mapping revision requires
a new author version and invalidates the accepted result.
Plan keyboard, focus, missing-choice, tie, display-order, narrow-screen,
and host tests. Mark actual implementation and test results pending.
Deliver:
1. The native question bank and answer-to-outcome ledger.
2. A concise explanation of the count and tie rules.
3. Result copy for W, V, P, and the tied set.
4. The fictional worked traces and expected-case matrix.
5. Blank fields for actual responses and actual test observations.
6. A review plan for a prepared HTML reading copy in NOS. Limit the
product statement to saving and rendering prepared HTML. Keep the
native bank authoritative; do not promise host execution of quiz controls.
Do not collect names or responses, create accounts, call services or
models, add autosave, build HTML now, capture screens, or publish.
Keep fictional traces separate from actual participant evidence.
Finish with a choice the reader can make
Before approving the copy, read the result without its decorative heading. Can the participant tell which selections produced it? Can someone with an even split see all three options? Can an unfinished submission remain unfinished?
The deliverable is a transparent activity suggestion: a person can inspect the three counts and choose a manageable next step. Keep actual responses blank until they exist, and record host observations only after testing the interface.
Use NOS to save and render prepared HTML beside this mapping review. End with an invitation to try a format, not a label the participant has to accept.