# Fictional market-research studies

Five newly written studies provide complete ODIN and Qube Script instruments,
independent questionnaire specifications and synthetic respondent scenarios.
They are designed for regression testing, author practice and reviewed demos.
No study has a real client, fieldwork, respondents or findings. Brands, places
and response text are invented; private questionnaire wording is not included.

| Study | Research question | Specification |
|---|---|---|
| [Breakfast grains](consumer-grocery/README.md) | Where does a brand funnel lose buyers, and how do selected brand users rate their experience? | [28 measures](consumer-grocery/SPEC.md) |
| [Connectivity service](connectivity-service/README.md) | How do fixed/mobile context, reliability, support experience and switching differ? | [30 measures](connectivity-service/SPEC.md) |
| [Household cleaning concept](household-concept/README.md) | Can buyers understand a hypothetical concentrate concept, and what affects appeal, concerns and priced intention? | [30 measures](household-concept/SPEC.md) |
| [Business software](b2b-software/brief.md) | How do purchase-influencing decision makers evaluate adoption and continuation of work tools? | [35 item definitions](b2b-software/questionnaire.md) |
| [Leisure travel](leisure-travel/brief.md) | How do recent domestic day/overnight visitors plan and evaluate their most recent visit? | [33 item definitions](leisure-travel/questionnaire.md) |

Each directory contains `study.qube` and `study.odin`. Read its brief, frozen
requirements, scenarios and limits before changing source. Keep substantive
measurement, routing, units, recall windows, missingness and denominators aligned
between implementations. Different engine seeds need not produce identical
random orders; stable codes identify the answers.

Run the focused existing-engine checks from the repository root:

```sh
node tools/test_fictional_consumer_studies.mjs
node tools/test_fictional_professional_studies.mjs
```

The harnesses need the built local ODIN and QSS WASM bindings. They make no
network/provider calls and do not rebuild bindings. Their expected paths and
answer contracts come from scenarios, not from comparing the engines with each
other. Passing results qualify those executed local behaviors. They do not
establish vendor-DLL parity, a trusted hosted ODIN final/export path, production
fieldwork, browser accessibility or statistical validity.

The briefs state sampling, denominator and pretest assumptions. Estimated
10–15-minute durations still need timing. New authors should inspect variable
paths and repeat counts rather than equating item definitions with screens.
Host configuration owns recruitment, quota authority, permissions, retention,
contact handling and collection. No payments or incentives are implemented.

Human wording, methodology and readability review precede public training or
marketing use. This collection does not promote any pattern implementation or
prove that Qube Script is shorter or better than ODIN. Concrete scripting gaps
and tested workarounds belong in the individual packages and qualification record.