IMPACT ANALYST · AGGREGATE ONLY
Manage · Measurement · Question sets

Survey and assessment question sets

Two separate libraries serving two different metrics. The self-report survey is voluntary and feeds METRIC-03; the knowledge assessment is the baseline and endline instrument behind METRIC-01. This page is shared with the content admin.

This is not the learning question bank. Quiz questions that follow a video, and Showdown questions, live in the content admin's library. The three libraries do not share questions and cannot import from one another — mixing them turns a measurement instrument into a teaching tool and the measurement stops meaning anything.

Who does what here: the impact analyst is responsible for whether the instrument stays comparable (version, when it was locked, how the scale is built); the content admin reviews wording and readability. Analyst read-only access applies to user data, not to the instruments themselves.

Open period: Q2 2026 Assessment in use: v1.2 Survey in field: v2.0 2 drafts held for next period

1 · Knowledge assessment A-01

Sat twice by each participant: baseline at the end of onboarding and endline two weeks after that baseline. Twenty questions, scored out of 20, no time limit.

The assessment pays no tokens, and it cannot. The assessment has no reward field at all, and the marking service has no connection to the wallet. If a measurement instrument paid out, people would have a reason to sit it repeatedly for tokens and METRIC-01 would be worthless.

Assessment versions

VersionQuestionsPeriod Used forStatus
v1.2
locked 31 Mar 2026
20Q2 2026 64 baseline · 41 endline In use · locked Open locked version
v1.3
draft, opened 12 Jul 2026
20Q3 2026 (planned) Draft · 2 questions reworded Review draft
v1.120Q1 2026 (internal trial) 12 baseline Archived Q1 snapshot
v1.018never released 0 Archived Read-only

Locked means it cannot be edited again, not even to fix a typo. Every assessment record stores the version it was sat under, and METRIC-01 only pairs a baseline with an endline when both used the same version. Editing v1.2 now would make the 41 existing pairs incomparable.

v1.2 · 4 of the 20 questions

Full version record
CodeQuestionWhat it measuresMarks
B-03 Odds of 1.85 correspond to roughly what implied probability? Odds and probability1
B-07 What is the house edge, and where does it sit in the odds you are shown? Recognising the house edge1
B-11 After nine reds in a row, how do the chances of black on the tenth spin change? Gambler's fallacy1
B-14 If you double your stake after every loss, what happens to your long-run expectation? Chasing losses and negative expectation1

Correct answers are not shown on this page and are never sent to the device while someone is sitting the assessment. Participants are not told what they got right after the baseline, so the assessment does not teach its own answers. There is no countdown timer — this is a measurement, not a game.

Draft · not in use

Assessment A-01 · v1.3 review

Read-only for impact analyst

Sam Okafor opened this version for the next cohort. Your role is to check comparability and request a correction; wording remains with the content author.

QuestionChange from v1.2Measurement note
B-07“Bookmaker margin” added beside “house edge”. Meaning is unchanged; readability review requested.
B-14“Every loss” replaced with “each consecutive loss”. May reduce ambiguity; difficulty check still pending.
Locking stays unavailable until authoring and clinical review are complete.

2 · Voluntary self-report survey

Periodic questions about real gambling outside the app. Answers are stored de-identified and only ever read out in aggregate.

Survey sets

Content admin authors new sets
SetQuestionsSent ResponsesStatus
S-02 · Self-reported behaviour v2.0 6month 1 and month 3 38 of 64 · 59% In field View
S-02 · Self-reported behaviour v2.1 7planned for Q3 2026 Draft · question 4 is double-barrelled Review draft
S-01 · Experience of the app v1.0 4once, at end of period 22 of 64 · 34% Closed Read-only

Response counts refresh nightly at 06:00.

Live · locked

Survey S-02 · v2.0 instrument

Read-only for impact analyst

Offered after a run ends · opt-in · locked 14 July 2026 · responses stored without an account identifier. Every question can be skipped and there is no token reward.

  1. In the last month, have you placed a bet with real money?
  2. If yes, roughly how often?
  3. Has that changed since you started using tibbi?
  4. Have you set a limit, taken a break, or blocked yourself anywhere else?
  5. In the last month, have you wanted to cut down or stop betting with real money?
  6. Is there anything here that made things worse for you?
Compare with the v2.1 draft Live wording is immutable; a correction must become a new version.
Draft · next period only

Survey S-02 · v2.1 review

Read-only for impact analyst

The new question asks about both frequency and spend in one answer. That would produce a result that cannot be interpreted, so the draft is held before review.

Requested correction

Split “How often did you bet and how much did you spend?” into two optional questions, each with “Prefer not to answer”. Do not add a token reward or make completion a condition of using tibbi.

Rules that apply to every survey set

  • Entirely voluntary. Every question has a “Skip” option and every set opens with “No thanks”. Declining costs nothing, changes nothing, and the invitation is not shown again.
  • No tokens, the same as the assessment. Paying for survey answers is buying them.
  • Nothing that identifies anyone. No names, no phone numbers, and no free-text question detailed enough to recognise the person answering.
  • No double-barrelled questions. One question measures one thing, or it cannot be analysed.
  • No leading questions. “Has the app helped you bet less?” is not allowed. “In the last 30 days, how many times did you place a real bet?” is.

METRIC-03 is always reported as “38 people reported the following”, never as “the app reduced gambling”.

Why locking a version matters

METRIC-01 compares the same person before and after. That comparison only holds if the ruler does not change between the two measurements.

  • Swap a hard question for an easy one and the endline score rises without any change in knowledge.
  • Add a question and the scale changes, so the two sittings are no longer in the same units.
  • Rewording something to be “clearer” also changes its difficulty, even when the content is untouched.

So the process is: make edits in the draft of the next version, lock it before the new period starts, and state plainly in the report that the two periods used different versions and cannot be pooled. See where the comparison refuses to run →