# Phase 1: Data Familiarization You are a sub-agent executing Phase 1 of a qualitative research synthesis. Your goal is deep familiarity with the interview data — knowing what participants said, how they said it, and what stands out — before any systematic coding begins. ## Context You will receive: - A batch of ~15 interview files (transcripts, notes, or summaries) - `analysis/config.md` — the analysis configuration - The research overview, objectives, and interview guide - Your batch number (for file naming) ## Your Outputs Write all outputs to `analysis/phase1-familiarization/`. ### 1. Batch Memos File: `batch-{n}-memos.md` For each interview in your batch, write a structured memo: ```markdown ## Interview Memo: [Participant ID or Name] **Participant Context**: [Role, demographics, background — whatever is available] **Overview**: [2-3 sentences: what is this person's story in relation to the research questions?] **Key Topics & Stances**: - [Topic 1]: [Their experience, perspective, or behavior] - [Topic 2]: [Their experience, perspective, or behavior] - [Continue for all major topics discussed] **Notable Quotes** (3-5 excerpts that capture something important): > "[Verbatim quote]" > — Context: [what they were discussing when they said this] **Behavioral Observations**: [What do they DO vs what they SAY? Note workarounds, actual practices, revealed preferences] **Emotional Tenor**: [How did they feel about what they discussed? Note intensity — frustration, resignation, excitement, pride, anxiety] **What Stood Out**: [Anything surprising, puzzling, contradictory, or particularly striking about this interview] **Potential Connections**: [How might this interview relate to others you've read? Early groupings?] ``` ### 2. Batch Observations (appended to memo file) After all memos in your batch, add a section: ```markdown ## Batch {n} Cross-Interview Observations **Emerging Patterns**: [What do multiple participants in this batch talk about similarly?] **Notable Variation**: [Where do participants differ? What might explain the differences?] **Surprises**: [What didn't you expect? What challenges assumptions?] **Puzzles**: [What doesn't make sense yet? What needs further investigation?] **Potential Groupings**: [Are there types or clusters forming? Which participants seem similar?] **Strongest Quotes in This Batch**: [List 3-5 quotes that are particularly vivid, analytical, or representative — these are early candidates for the evidence bank] ``` ## Reading Stance Approach each interview asking: - What is this person trying to tell me? - What matters to them about this? - What would I need to know to understand their experience from the inside? - What do they DO vs. what do they SAY they do? (Behavioral > stated preferences) - What surprises me? What confirms expectations? - What is NOT said — silences, deflections, avoided topics? - What language and metaphors do they use? (In-vivo candidates) ### Distinguishing Input Types **Raw transcripts**: Rich in language, pauses, hedging, emphasis. Pay close attention to: - What they say at length vs. briefly - Where they become animated or emotional - Self-corrections and contradictions within the same interview - Unprompted topics (what they bring up without being asked) **Interview notes**: More condensed, filtered through the researcher's lens. Pay attention to: - What the note-taker chose to record (signals importance) - Bracketed or italicized researcher observations - Gaps between questions asked and topics noted **Interview summaries**: Most processed — treat as an interpretation layer. Pay attention to: - Themes the summarizer highlighted - Quotes they chose to preserve - Whether the summary captures behavior or just attitudes ## Quality Gate Before writing your outputs, verify: - [ ] **Cognitive Empathy**: For each memo, can a reader understand this participant's perspective from the inside? If any memo reads like a clinical summary rather than an empathic account, revise. - [ ] **Heterogeneity**: Do your memos capture how participants DIFFER, not just what they share? If all memos sound the same, you are flattening variation — go back and sharpen distinctions. - [ ] **Specificity**: Do quotes include enough context to understand them later? If any quote is orphaned from context, add it. - [ ] **Behavioral vs. Attitudinal**: Have you noted what participants DO (behaviors, workarounds, practices) separately from what they SAY they want? If behavioral observations are thin, re-read for them. ## For the Consolidation Agent If you are the **consolidation agent** (running after all batch agents complete): Read ALL `batch-{n}-memos.md` files and produce: ### `consolidated-observations.md` ```markdown # Consolidated Observations Across All Interviews **Dataset Overview**: [Total interviews, participant characteristics, data types] ## Strong Patterns (appeared across multiple batches) - [Pattern 1]: [Description + which batches/participants] - [Pattern 2]: ... ## Notable Variation - [Dimension of variation]: [How participants split + tentative explanation] ## Contradictions & Tensions - [Contradiction]: [What participants disagree about or where accounts conflict] ## Surprises - [Surprise]: [What challenges prior assumptions] ## Emerging Groupings - [Group A]: [Characteristics + which participants] - [Group B]: ... ## Strongest Quotes (Top 10-15 across all batches) > "[Quote]" — [Participant ID] (Context: ...) ## Puzzles for Phase 2 - [Question that coding should help answer] ``` ### `phase1-summary.md` A concise handoff (aim for ~500 words) containing: 1. Dataset description (count, participant types, data quality) 2. Top 5 patterns observed 3. Top 3 dimensions of variation 4. Key surprises and contradictions 5. Recommended coding focus areas for Phase 2 6. Any data quality concerns (e.g., thin interviews, inconsistent depth) --- ## Parallel Extensibility Slot _The `parallel/` directory is reserved for future analysis signals embedded in this phase. Currently empty. Examples of what could be added here:_ - _`sentiment-extraction.md` — Per-interview emotional valence ratings and high-intensity moments_ - _`journey-raw-data.md` — Temporal touchpoints and emotional states extracted per interview_ - _`language-patterns.md` — Distinctive vocabulary, metaphors, and framing devices_