# Illustrative custom study codebook

**All data and quotations are invented. No human or model study was run. This is a format example, not a forecast, validation result or client finding.**

## Files

- custom-study-sample.pdf: seven-section report for reading and sharing.
- custom-study-sample.html: accessible HTML version of the complete report.
- custom-study-data.json: 80 invented coded records, computed summaries and eight independently invented text examples.
- custom-study-codebook.md: this dictionary and calculation guide.

## Data structure

The JSON contains title, version, data_status, usage, design, scales, responses, summaries and illustrative_notes. Every record is invented. There are four concepts with 20 records each. Each concept has 10 records in scenario_1 and 10 in scenario_2. Scenario groups are arbitrary, and their metric distributions are deliberately equal within each concept. Records are constructed examples; they do not represent people or generated personas.

## Response fields

| Field | Type and allowed values | Meaning |
| --- | --- | --- |
| record_id | Text, EX001 through EX080 | Local example identifier; no external identity. |
| concept | concept_a, concept_b, concept_c, concept_d | Neutral concept label; no creative or real product is associated with it. |
| scenario | scenario_1, scenario_2 | Arbitrary example subgroup. |
| clarity | Integer 1-5 | Example response to: How easy is this concept to understand? |
| fit | Integer 1-5 | Example response to: How well does this concept fit the task described? |
| explore | Integer 1-5 | Example response to: How interested would you be in exploring this concept further? |
| data_status | invented | Required status marker on every record. |

For clarity and fit, the common scale is 1 = not at all, 2 = slightly, 3 = moderately, 4 = very, 5 = extremely. For explore, use 1 = not at all interested through 5 = extremely interested. These are invented example questions, not a reproduced client questionnaire. No stimulus or task description was administered. No missing values are included.

## Calculations

Mean = sum of the coded responses / number of included records. Top-two count = count of values 4 or 5. Top-two percent = 100 x top-two count / number of included records. Distribution is a five-element array of counts in the order 1, 2, 3, 4, 5. Results are unweighted. Each report concept base is 20; each concept-by-scenario base is 10. Means display two decimal places and percentages display whole numbers.

No calibration, confidence intervals, significance tests or population inference are applied. Intent scores are not observed behavior or conversion rates. The small constructed dataset is intended only to make the tables easy to verify.

## Summary fields

Each summaries item includes concept, label, records, and metrics. Each metric includes mean, top_two_count, top_two_percent and distribution. These fields can be recomputed from responses.

## Illustrative notes

Each note contains example_id, concept, theme, text, data_status and linked_to_record. The eight notes are separately written examples, not respondent quotations. linked_to_record is false for every note. Theme names are editorial examples; there are no theme frequencies or prevalence estimates. Do not join these notes to response records.

## Safe reuse

Use this package to review the presentation and handoff format. Replace all invented values and text before using it for an actual project. This sample does not establish pricing, delivery scope, access rights, or commercial terms. An actual study's deliverables are agreed in scope.