mixed-methods-findings-writer（混合方法调查结果作者）

Mixed-Methods Findings Writer

Draft Results/Findings sections for mixed-methods sociology articles using structural patterns discovered in 84 Social Forces and Social Problems articles.

Project Integration

This skill reads from project.yaml when available:

# From project.yaml
type: mixed  # This skill is for mixed methods projects
paths:
  drafts: drafts/sections/
  tables: output/tables/
  figures: output/figures/
  quotes: analysis/outputs/

Project type: This skill is designed for mixed methods projects.

Consumes output from both qualitative analysis (interview-analyst) and quantitative analysis (r-analyst or stata-analyst).

Updates progress.yaml when complete:

status:
  integration_draft: done
artifacts:
  findings_section: drafts/sections/findings-section.md

Connection to Other Skills

File Management

This skill uses git to track progress across phases. Before modifying any output file at a new phase:

1. Stage and commit current state: git add [files] && git commit -m "mixed-methods-findings-writer: Phase N complete"
2. Then proceed with modifications.

Do NOT create version-suffixed copies (e.g., -v2, -final, -working). The git history serves as the version trail.

Workflow

Phase 1: Orient

Gather from the user:

1. Quantitative component: regression output, descriptive statistics, surveys, admin data
2. Qualitative component: interviews, field notes, archival documents, ethnographic observations
3. Integration rationale: Is this *elaboration* (vertical -- both methods address the same question at different depths) or *extension* (horizontal -- each method addresses a different question)? Or triangulation (independent validation)?
4. Theoretical predictions: hypotheses, expectations, or sensitizing concepts
5. Target length: typical is 15-30 paragraphs (3,000-7,000 words; median 4,895)

If the user has already written a draft, read it and assess which cluster it most resembles before suggesting revisions.

Phase 2: Select Cluster

Present the 5 clusters. Recommend 1-2 based on integration goal and analytic strategy:

Selection heuristics:

• Concept-driven analysis + both methods throughout --> Thematic Integrator
• Regression/models as core claims + interviews/fieldwork explaining mechanisms --> Quant-Anchored Elaborator
• Two data sources answering the same question --> Alternating Validator
• Distinct study phases (e.g., survey then interviews) --> Sequential Study Design
• Ethnography/interviews as primary + descriptive stats as context --> Qual-Dominant Quantifier
• Heavy model-building with minimal qualitative --> Quant-Anchored Elaborator (model-progression variant)

Integration rationale and cluster fit:

• *Elaboration* (vertical integration: same question, different depth) → Thematic Integrator or Quant-Anchored Elaborator. Qual deepens quant or vice versa. The two methods produce "more than the sum of parts" because together they address both magnitude and mechanism.
• *Extension* (horizontal integration: different questions, combined for a fuller picture) → Sequential Study Design or Alternating Validator. Each study answers its own question; cross-study transitions must explain how the questions connect.
• *Triangulation* (independent validation) → Alternating Validator. Both methods reach the same conclusion independently, strengthening credibility.
• Mixed rationales are common: a paper may use elaboration within themes and extension across them.

After selection, read the matching guide from clusters/{cluster-name}.md.

Phase 3: Build the Arc

Using the cluster guide, construct a section outline:

1. Map each major finding to a MOVE from the vocabulary below
2. Sequence moves following the cluster's canonical arc
3. Allocate paragraphs using the cluster's paragraph budget
4. Plan method transitions -- where does the evidence type shift?
5. Identify the opening and closing moves

Standardized move vocabulary (16 moves):

Present the arc as a numbered outline with paragraph counts and method labels per move.

Phase 4: Draft

Write each move following corpus norms. Consult techniques/techniques.md for the full catalog.

Opening paragraph (choose one based on cluster):

• *Qualitative vignette* (16%): scene-setting narrative from fieldwork or interview
• *Hypothesis restatement* (29%): "Recall that we expected..."
• *Table reference* (22%): "Table 2 presents results from..."
• *Main finding first* (13%): lead with the most important result
• *Methodological setup* (7%): "To examine this, we draw on..."

Method integration (the core challenge):

• Elaboration (47%): qual evidence explains or deepens quant findings -- *vertical* integration where both methods address the same question at different depths
• Complementarity (25%): each method addresses different facets of the question -- may be *vertical* (same question, different aspects) or *horizontal* (different questions combined)
• Illustration (13%): qual examples make quant patterns concrete -- weakest form; aim for elaboration instead
• Sequential (13%): one method's findings inform the next method's analysis -- *horizontal* integration where later methods address questions raised by earlier ones

The "more than sum of parts" test: at each integration point, ask whether the combined evidence yields an insight that neither method alone could produce. If qual merely restates what quant already showed, the integration is illustrative, not elaborative.

Body paragraphs -- cross-cutting norms:

• Qualitative evidence occupies 60-75% of paragraph space even when quant establishes core claims
• Extended quotations (50+ words) function as analytical anchors, not mere illustration (61%)
• Transition sentences between methods appear in only 17% of articles -- add them deliberately
• Connect to theory at heavy (53%) or moderate (45%) density
• Use subgroup analysis as a bridge device between methods (38%)

Closing paragraph (choose one):

• *Qualitative synthesis* (21%): return to qualitative evidence to frame the takeaway
• *Integration summary* (20%): weave both methods into a unified conclusion
• *Transition to discussion* (16%): bridge paragraph
• *Summary paragraph* (10%): recap all findings
• *Mechanism test* (9%): close with process evidence

Phase 5: Calibrate

After drafting, check against cluster norms:

• Does the arc match the canonical sequence for the selected cluster?
• Is the paragraph budget balanced across moves?
• Are method transitions smooth -- or do quant and qual blocks feel disconnected?
• Is qualitative evidence doing analytical work, not just illustrating?
• Are extended quotations introduced and interpreted, not left to speak for themselves?
• Is theory linking at the right density (heavy for Thematic Integrator/Qual-Dominant; moderate for others)?
• For quant claims: are coefficients translated into substantive terms?
• For qual claims: is there enough evidence breadth (multiple informants/cases)?
• "More than sum of parts" test: at each integration point, does the combined evidence produce an insight that neither method alone could yield? If qual only restates what quant showed, revise toward elaboration.
• Disconfirming evidence: when quant and qual findings diverge, is the divergence reported transparently and interpreted analytically? Suppressing divergence undermines mixed-methods credibility. Treat divergence as an analytic opportunity (e.g., "The regression shows X, but interviews reveal a countervailing process...").
• Computational methods note: if the paper uses NLP, topic models, or automated text analysis alongside interviews or fieldwork, this counts as mixed-methods. Apply the same integration norms: the computational results need qualitative interpretation, not just validation.

Present the draft with a brief calibration note.

Reference Files

• Cluster guides (read the one matching the selected cluster):

- clusters/thematic-integrator.md - clusters/quant-anchored-elaborator.md - clusters/alternating-validator.md - clusters/sequential-study-design.md - clusters/qual-dominant-quantifier.md

• techniques/techniques.md -- 19 writing techniques with descriptions and frequency data
• references/corpus-statistics.md -- summary statistics from the 84-article analysis corpus