# Qualitative Codebook — IDEA Case Study

Date: 2026-05-11
Phase: 7 — Qualitative Analysis and Triangulation

## Purpose

This codebook defines the qualitative categories used to code documentary and perception evidence across the four research questions. Each code has a definition, inclusion/exclusion rules, and an anonymized example. The codebook supports the triangulation matrix and ensures that qualitative claims are traceable.

## Coding Unit

The primary coding unit is the **card-level evidence record** (for RQ1 and RQ4) or the **questionnaire item response** (for RQ2 and RQ3). Ceremony records and assessment rows are secondary coding units when they provide contextual or linked evidence.

## Code Categories

### C01 — Partial Adoption (Adoção Parcial)

- **Definition**: Observable IDEA-related evidence exists in some cards or artifacts of a board, but not consistently across all expected work items.
- **Inclusion**: Board has at least one card with raw or linked IDEA evidence AND at least one expected card without observable evidence.
- **Exclusion**: Board where all expected cards have evidence (full adoption) or no cards have evidence (absence).
- **Example**: Board B07 had 3 cards with DT labels applied and 19 expected cards without observable IDEA evidence.
- **RQ relevance**: RQ1, RQ3.

### C02 — Absence of Operationalization (Ausência de Operacionalização)

- **Definition**: IDEA was expected or configured (labels exist, training occurred) but no card-level or ceremony-level evidence was found in the available corpus.
- **Inclusion**: Board with IDEA taxonomy configured (labels defined) but zero applied labels and zero raw card evidence.
- **Exclusion**: Boards where IDEA was not expected.
- **Example**: Board B03 had 12 IDEA/DT taxonomy labels configured but 0 labels applied to cards and 0 raw card evidence.
- **RQ relevance**: RQ1, RQ3, RQ4.

### C03 — Insufficient Documentation (Documentação Insuficiente)

- **Definition**: A card exists for a work item but lacks substantive description, checklist, comments, or delivery evidence, making TD reasoning non-auditable.
- **Inclusion**: Card with quality score ≤ 4 on the 0–16 rubric, or card with title only and no description/tasks/comments.
- **Exclusion**: Cards with medium or high quality band.
- **Example**: Card with title "configurar etapa de lint no pipeline" — no description, no tasks, no comments, no delivery evidence.
- **RQ relevance**: RQ1, RQ4.

### C04 — Documented Prevention (Prevenção Documentada)

- **Definition**: Observable evidence that a card or assessment row documents a TD prevention action: testing, documentation, refactoring planned before debt accumulates, or quality gates.
- **Inclusion**: Assessment row coded as "Prevenção" with evidence description; or card with checklist items for testing/documentation/quality that were planned proactively.
- **Exclusion**: Cards where prevention was expected but not documented; cards where prevention was only mentioned in a ceremony without card-level trace.
- **Example**: "Foi feita a documentação das alterações referentes a API através de swagger" — prevention through documentation.
- **RQ relevance**: RQ1, RQ4.

### C05 — Documented Repayment (Pagamento Documentado)

- **Definition**: Observable evidence that a card or assessment row documents a TD repayment action: refactoring, fixing, completing previously incomplete work, or resolving identified debt.
- **Inclusion**: Assessment row coded as "Pagamento" with evidence description; or card with explicit DT label and repayment action (e.g., "[DT] Standardization of test writing").
- **Exclusion**: Cards where repayment was expected but the card remained in backlog without execution.
- **Example**: "Foi feito planejamento para refatoração, mapeado como dívida técnica" — repayment planning documented.
- **RQ relevance**: RQ1, RQ4.

### C06 — Impediment (Impedimento)

- **Definition**: A card or assessment row documents a blocking factor that prevented work completion or TD management action.
- **Inclusion**: Card marked as blocked/impediment; assessment row noting an impediment; ceremony record mentioning a blocking issue.
- **Exclusion**: Cards that are simply late without documented reason.
- **Example**: "Mapeado impedimento para o desenvolvimento da história" — impediment identified but not detailed.
- **RQ relevance**: RQ4.

### C07 — Decision Factor (Fator de Decisão)

- **Definition**: Observable evidence of a factor that influenced the team's decision to incur, postpone, prevent, or repay TD.
- **Inclusion**: Assessment row or card comment explaining why a decision was made (e.g., time pressure, team problem, lack of knowledge).
- **Exclusion**: Generic cards without rationale.
- **Example**: "A necessidade da dívida técnica foi gerada por um problema na equipe, que foi reportado e teve a solução detalhada no card" — decision factor: team problem.
- **RQ relevance**: RQ4.

### C08 — Documentation Cost (Custo de Documentação)

- **Definition**: Evidence that the effort required to document TD reasoning or use IDEA was perceived as a barrier or cost.
- **Inclusion**: Likert items on ease of use with low agreement; ceremony records mentioning difficulty; cards where documentation was planned but not executed.
- **Exclusion**: General low engagement without explicit documentation-cost mention.
- **Example**: Mid-intervention ease of use: mean 3.56, only 45.7% agreement — lowest among all Likert dimensions.
- **RQ relevance**: RQ2, RQ3.

### C09 — Perceived Utility (Utilidade Percebida)

- **Definition**: Participant self-report indicating that IDEA or TD management support was perceived as useful, helpful for visualization, learning, or decision-making.
- **Inclusion**: Likert items on usefulness, TD visualization, confidence, learning, or prevention/repayment support with agreement ≥ 67%.
- **Exclusion**: Neutral responses (score 3) without clear direction.
- **Example**: "Perceived usefulness for TD management: mean 4.24, 81.4% agreement" — strong perceived utility.
- **RQ relevance**: RQ2, RQ3.

### C10 — Resistance or Abandonment (Resistência/Abandono)

- **Definition**: Evidence that participants did not adopt IDEA despite training, or abandoned its use after initial exposure.
- **Inclusion**: Boards where IDEA was expected but zero evidence found; cards where DT was identified but no IDEA-related action was taken; assessment rows noting "Não houve comentário detalhando o item".
- **Exclusion**: Boards where IDEA was not expected.
- **Example**: "Não houve comentário detalhando o item que não foi executado, nem replanejando a atividade" — absence of operationalization despite identified need.
- **RQ relevance**: RQ1, RQ3, RQ4.

## Coding Procedure

1. Each card/assessment row is read with its board context, IDEA condition, and evidence hierarchy classification.
2. One or more codes are assigned based on the definitions above.
3. Codes are not mutually exclusive: a single card may receive C01 + C04 (partial adoption with documented prevention) or C02 + C10 (absence of operationalization with resistance).
4. Coding was performed by the first author and verified against the Evidence Ledger and assessment rows.
5. Disagreements or ambiguous cases are documented in the divergence log.

## Triangulation Matrix Structure

The triangulation matrix crosses two dimensions:

| | Documentary evidence HIGH | Documentary evidence LOW |
|---|---|---|
| **Perception HIGH** | Convergent: participants perceive utility AND cards show IDEA evidence | Divergent: participants perceive utility BUT cards lack IDEA evidence |
| **Perception LOW** | Divergent: cards show evidence BUT perception is neutral/low | Convergent absence: neither perception nor documentary evidence supports adoption |

- **Perception HIGH**: Likert agreement ≥ 67% on usefulness/visualization/learning items.
- **Perception LOW**: Likert agreement < 67% or ease-of-use items.
- **Documentary evidence HIGH**: Card has raw or linked IDEA evidence; quality score ≥ 8.
- **Documentary evidence LOW**: Card has no IDEA evidence or quality score ≤ 4.

## Relationship to Evidence and Claim Ledgers

Each coded instance should be traceable to:
- An Evidence Ledger entry (source type, location, metric).
- A Claim Ledger entry (claim text, scope, wording guardrail).

Codes that cannot be traced to at least one evidence entry should be flagged for review.
