# A public protocol for studying prediction-market users A prospective questionnaire and 90-day research plan for testing what traders understand about alerts, copying costs and settlement. No survey responses have been collected. Canonical URL: https://vultax.com/research/prediction-market-user-research-protocol Author: [Vultax Research](https://vultax.com/editorial-policy) Published: 2026-09-20 Last public revision: 2026-09-20 Published research snapshots and calculations; these exports do not contain current quotes or live account data. Observation windows and populations are stated in the source captions and methodology. Missing metadata is unknown, not zero. Prospective protocol only. Zero respondents, no activated form and no measured user attitudes. Proposed targets and dates are not completed work. Any convenience sample must be labelled and reported by recruitment channel. ## Key figures - Survey responses: 0 — Protocol only; recruitment has not begun. - Initial pilot target: 12 adults — Proposed usability sample, not an achieved recruitment count. - Research horizon: 90 days — Proposed stages and decision gates; no promise of completed data. ## The short answer A useful user study should make it possible to publish an uncomfortable answer. Vultax is releasing a questionnaire and analysis plan before recruitment begins. It asks how people interpret a quoted price, a first-seen wallet timestamp and a market end date, then separates those interpretations from reported trading experience. No survey responses have been collected. The proposed first step is a 12-person adult usability pilot to find confusing wording, followed by a voluntary multi-channel survey if the questions work. That pilot target is a plan, not an achieved sample. Neither stage should be presented as a representative poll of all prediction-market users. Recruiting through a single creator would primarily describe that creator's audience. ## Questions that map to observable misunderstandings The questionnaire begins with consent and adult eligibility. It asks about prediction-market use in broad frequency bands, then presents three short comprehension vignettes: whether a best ask guarantees a 1,000-contract purchase at that price; whether first seen in a dataset means a newly created account; and whether a listing end date proves a payout is late. Each includes a cannot-tell option. Experience questions ask whether the respondent has tried alerts or copy tools, whether an attempted copy was incomplete or changed price, and which settlement milestone was unclear. Recall questions use a 30-day window and include not-applicable and prefer-not-to-answer choices. No wallet address, name, email, exact balance, financial loss amount or political preference is requested. ### Questions that map to observable misunderstandings Planned measures, not observed findings. Do not populate counts until responses are collected. | Construct | Question format | Primary output | | --- | --- | --- | | Price vs capacity | Neutral vignette + cannot tell | Response shares and missing answers | | Dataset history vs account age | Neutral vignette + cannot tell | Response shares by recruitment channel | | End date vs payout | Neutral vignette + cannot tell | Response shares by broad experience band | | Tool experience | 30-day recall + not applicable | Descriptive self-report only | [CSV](https://vultax.com/research/prediction-market-user-research-protocol/data.csv?dataset=questions-table) · [JSON](https://vultax.com/research/prediction-market-user-research-protocol/data.json?dataset=questions-table) ## Freeze the denominator before promoting the link The proposed pilot is 12 adults across at least three experience bands, with an equal invitation target per band. Pilot feedback is for wording and feasibility, not prevalence. Keep the original questionnaire and publish a revision log. Do not pool pilot answers with the later survey if the wording changes. For the main survey, use a fixed 14-day collection window, a target of 100 eligible complete responses and a maximum of 200. End at the fixed date or cap, whichever comes first; do not keep recruiting until a desired result appears. Record recruitment channel without requiring personal identifiers. Report eligible starts, completions, item nonresponse and exclusions by channel. Publish the prespecified comprehension outcomes even if they show little confusion. A comparison by experience band is exploratory unless sample sizes and hypotheses are frozen before collection. Confidence intervals do not repair self-selection bias. If responses are sparse, release a labelled pilot or feasibility report rather than a representative-sounding percentage. ## Consent and data handling before launch The consent text should state the purpose, voluntary nature, approximate five-minute duration, right to skip a question or leave before submitting, intended aggregate publication, and who operates the form. Anonymous responses cannot usually be located for deletion afterward; say this before submission. A research contact should be available for questions. Before recruiting, select a form configuration that does not require account login or collect emails, review its platform metadata behavior, and publish the actual retention period. The proposed policy is deletion of raw response exports after 90 days and retention of aggregate tables and the questionnaire. Do not promise that a third-party form provider collects no technical metadata without checking it. Small cells below five should be combined or suppressed, and optional free text should not be published verbatim without separate consent. ## A 90-day sequence with useful stop conditions Days 1–14: invite methods feedback on the five published audits and agree one narrow joint question. A partner contributes a question, a data definition or an opt-in audience; Vultax contributes the code, data audit and result. No one receives a promised favorable conclusion. Days 15–30: freeze a prospective alert and execution protocol, capture all eligible observations, and verify timing and coverage before calculating outcomes. Days 31–60: run the usability pilot and, only after consent and collection settings are checked, the survey. Days 61–90: publish the complete denominator, counterexamples and a partner edition if the evidence supports one. Stop a collection that cannot preserve identity, timestamps or consent. Keep a null result when the measurement is sound. Measure useful distribution by relevant citations, qualified referral visits, dataset reuse and meaningful replies, with explicit counts and denominators. Impressions and domain scores alone do not demonstrate a return. ## Check the calculation [questionnaire.json](https://vultax.com/datasets/prediction-market-user-research-protocol/questionnaire.json) · [analysis-plan.md](https://vultax.com/datasets/prediction-market-user-research-protocol/analysis-plan.md) · [METHODS.md](https://vultax.com/datasets/prediction-market-user-research-protocol/METHODS.md) · [Reproduction bundle](https://vultax.com/datasets/prediction-market-user-research-protocol/reproduction.zip). The bundle contains the calculation, frozen analysis inputs, results and a SHA-256 manifest. This bundle contains a proposed instrument and analysis plan, not respondent data. No survey form has been activated and no invitations have been sent. Reuse the figures with a link to this study and the stated population and observation window. Vultax welcomes corrections and independent methods reviews. A citation or collaboration does not require a positive conclusion or a reciprocal link. ## Sources and methodology - [Vultax execution audit](https://vultax.com/research/prediction-market-order-size-depth-audit) — Basis for the capacity vignette. - [Vultax wallet-history audit](https://vultax.com/research/polymarket-wallet-first-seen-history-audit) — Basis for the first-seen vignette. - [Vultax closure-clock audit](https://vultax.com/research/polymarket-end-date-closure-clock-audit) — Basis for the settlement vignette.