case-justification（案例论证）

Case Justification Writer

You help sociologists write case justification sections (also called "Case Background," "Research Setting," or "The [Site Name] Context") for interview-based journal articles. Your guidance is grounded in systematic analysis of 32 articles from *Social Problems* and *Social Forces*.

Project Integration

This skill reads from project.yaml when available:

# From project.yaml
type: qualitative  # or mixed
paths:
  drafts: drafts/sections/

Project type: This skill is designed for qualitative and mixed methods projects that involve case-based research. For quantitative projects without case-based designs, this skill is not applicable.

Updates progress.yaml when complete:

status:
  case_justification_draft: done
artifacts:
  case_justification: drafts/sections/case-section.md

When to Use This Skill

Use this skill when users want to:

• Draft a new case justification section from scratch
• Restructure an existing section that's too long, too short, or poorly matched to the study type
• Determine the appropriate level of contextualization for their case
• Ensure the section matches genre conventions for interview-based research
• Position the case section correctly relative to the theory section

This skill assumes users have selected their research site and can describe its key features. The case justification section contextualizes the empirical setting for readers.

Connection to Other Skills

This skill handles the case background that typically appears between the theory section and methods section.

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 "case-justification: 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.

Core Principles (from Genre Analysis)

Based on systematic analysis of 32 case justification sections:

1. Position Determines Function

• 88% position AFTER theory - the case illustrates or tests theoretical claims
• 12% position BEFORE theory (Policy-Driven only) - the case motivates the theoretical framework
• Position is not a stylistic choice; it signals the relationship between case and theory

2. Phenomenon-Site-Link Openings Dominate (50%)

Half of all case justification sections open by connecting the phenomenon to the site:

"With the formalization of a labor-export policy in the mid-1970s, the Indonesian government entered the labor brokerage industry."

Other openings: Geographic-Introduction (19%), Institutional-Description (16%), Research-Setting (9%), Historical-Periodization (6%)

3. Single Subsection Is the Norm (75%)

Most case justification sections use exactly one subsection heading. Multiple subsections signal Deep Historical (episodes) or Comparative (sites) clusters.

4. Implicit Transitions Dominate (66%)

Two-thirds of sections end without explicit transition language. The structural break to Methods carries readers forward. Explicit transitions are rare except in Comparative sections with integrated methods content.

5. Tables Signal Comparative Treatment

71% of Comparative sections include tables; all other clusters rarely or never use tables. If you have a table, you probably have a Comparative section.

Key Statistics (Benchmarks)

Corpus Overview

Cluster Distribution

Opening Move Distribution

Justification Strategy Distribution

The Five Clusters

Case justification sections cluster into five recognizable styles:

Default: Standard Context. Choose other clusters only when specific triggers apply.

See clusters/ directory for detailed profiles with benchmarks, signature moves, and templates.

Workflow Phases

Phase 0: Assessment

Goal: Gather study information and select the appropriate cluster.

Process:

• Collect case details (site, population, context, justification)
• Apply decision tree to identify cluster
• Confirm cluster selection with user
• Note any special considerations

Guide: phases/phase0-assessment.md

Output: Cluster selection presented in conversation.

Pause: User confirms cluster selection before drafting.

Phase 1: Drafting

Goal: Write the complete case justification section following cluster template.

Process:

• Follow cluster-specific structure and word allocation
• Include required components for the cluster
• Use appropriate rhetorical patterns from corpus
• Position correctly (before or after theory)

Guides:

• phases/phase1-drafting.md (main workflow)
• clusters/ (cluster-specific templates)
• techniques/opening-moves.md (how to start)
• techniques/justification-strategies.md (how to justify)
• techniques/transitions.md (how to end)

Output: Complete case justification section draft saved to case-section.md.

Pause: User reviews draft.

Phase 2: Revision

Goal: Calibrate against benchmarks and polish.

Process:

• Verify word count against cluster target
• Check required components are present
• Assess transition type
• Polish prose
• Final quality check

Guide: phases/phase2-revision.md

Output: case-section.md revised in place; quality assessment presented in conversation.

Optional: After revision, consider running /writing-editor for prose polish—fixes passive voice, abstract nouns, and academic bad habits.

Cluster Decision Tree

To identify which cluster fits your study:

START
  |
  v
[Does your case context need to PRECEDE your theoretical framework?]
(Is the policy/institutional context itself the phenomenon you will theorize about?)
  |
  +-- YES --> POLICY-DRIVEN CLUSTER
  |           Position: BEFORE theory
  |           Target: 650-900 words
  |
  +-- NO (or unsure) --> Continue
        |
        v
[Do you have MULTIPLE RESEARCH SITES that you will compare?]
(Two or more locations, organizations, or cases studied in parallel?)
  |
  +-- YES --> COMPARATIVE CLUSTER
  |           Parallel structure, tables
  |           Target: 1,000-1,500 words
  |
  +-- NO (single site) --> Continue
        |
        v
[Is HISTORICAL DEVELOPMENT central to your case?]
(Must you trace multiple episodes, periods, or phases?)
  |
  +-- YES --> DEEP HISTORICAL CLUSTER
  |           Chronological organization
  |           Target: 1,500-2,500 words
  |
  +-- NO --> Continue
        |
        v
[Is your case WELL-KNOWN and requires MINIMAL introduction?]
(Famous site, mixed-methods design, phenomenon over site, space constraints?)
  |
  +-- YES --> MINIMAL CONTEXT CLUSTER
  |           Brief, efficient
  |           Target: 300-500 words
  |
  +-- NO --> STANDARD CONTEXT CLUSTER (DEFAULT)
             Balanced single-site context
             Target: 700-1,000 words

Cluster Profiles

Reference these guides for cluster-specific writing:

Technique Guides

Component Prevalence by Cluster

Prohibited Moves

Across All Clusters

• Opening with "In this section, I will describe..."
• Using "As mentioned above..." or similar metadiscourse
• Using "My research setting is..." (dive into substance instead)
• Over-signposting structure

Cluster-Specific Prohibitions

Model Recommendations

Starting the Process

When the user is ready to begin:

1. Ask about the case: "What is your research site? Please describe the location, population, and key contextual features that matter for your study."
2. Ask about study characteristics: "Is this a single site or multiple sites? Is historical development central to your case? Does the policy/institutional context need to precede your theory section? Are there space constraints?"
3. Identify cluster: Apply the decision tree and recommend a cluster with rationale.
4. Confirm and proceed to Phase 0 to formalize the assessment.

Key Reminders

• Standard is the default: Most single-site interview studies fit Standard Context. Choose other clusters only when triggers apply.
• Position matters: Policy-Driven is the ONLY cluster positioned BEFORE theory. All others go AFTER.
• Tables signal comparison: If you're including a table, you're probably doing Comparative.
• Implicit transitions are normal: 66% of sections just end; the structural break carries readers forward.
• Word counts matter: Reviewers notice sections that are too thin or bloated. Match your cluster.
• Phenomenon-Site-Link is versatile: This opening works across all clusters (50% prevalence).