social-science-paperwork

The saved excerpt is a snapshot from when this skill was listed. The external source remains the complete and most current version.

---
name: social-science-paperwork
description: End-to-end social science and humanities manuscript workflow for research design, literature reviews, questionnaires, interviews, qualitative coding, survey analysis, APA/GB/T citation checks, ethics statements, journal/user-style adaptation, submission audits, and premium classroom/demo draft manuscripts with LaTeX/PDF output. Use when Codex helps plan, draft, audit, or analyze social science, humanities, education, communication, management, design research, cultural studies, or mixed-methods papers, especially involving surveys, interviews, focus groups, case studies, thematic analysis, grounded theory, scales, reliability, validity, human-subject data, target-journal style, advisor/lab writing preferences, non-submission demo papers, classroom presentations, advisor reports, senior-student or research-group presentations, simulated findings, draft-demo manuscripts, one-click complete papers, polished LaTeX, or "perfect paper" demo drafts.
---

# Social Science Paperwork

## Core Rule

Treat AI as a research assistant, not the responsible researcher. Never invent data, citations, scales, ethics approvals, participants, interview quotes, statistical significance, or coding evidence. Keep a traceable chain from source material to claims.

Default to **ask-first collaboration**. For underspecified social science or humanities tasks, ask a small number of high-impact questions before producing a full design. Do not bury the user in methodology; ask the decisions that change the plan.

## User-Facing Collaboration Flow

If the user is unsure how to work with AI, run the skill as a guided workflow, not a lecture. Load `references/user-facing-flow.md` when the user asks what to provide, wants a full process, or sounds unfamiliar with social science writing.

Default to this conversational order:

1. **Purpose** - Is this real research, a course demo, an advisor/research-group report, a proposal, or revision?
2. **Topic and audience** - What is the topic, discipline, target journal/class/advisor, and expected language?
3. **Evidence already available** - Literature, interview transcripts, survey data, cases, archives, artifacts, or only an idea.
4. **Method route** - Recommend one route card: qualitative-led, questionnaire-led, mixed-methods, literature/theory-led, or Premium Draft Demo.
5. **Output contract** - State the exact files/sections to create and which parts will be real, proposed, simulated, or TODO.
6. **Audit pass** - End with a claim-evidence map, risk list, and next questions instead of only polished prose.

## Mode Selection

First determine whether the user needs **Research Mode**, **Draft Demo Mode**, or **Premium Draft Demo Mode**:

- **Research Mode** - Use for real theses, manuscripts, journal submissions, research reports, or any output that may be evaluated as actual research. Keep all claims source-grounded and do not create fake data, participants, quotes, citations, approvals, or findings.
- **Draft Demo Mode** - Use when the user explicitly says the output is for classroom presentation, peer demonstration, advisor/research-group reporting, non-submission draft, simulated workflow, demo manuscript, or "not a real submission." In this mode, Codex may generate a complete manuscript-like draft and simulated findings only if every simulated element is clearly labeled as draft/demo/non-real.
- **Premium Draft Demo Mode** - Use when the user asks for a one-click complete paper, polished LaTeX/PDF, "perfect" demo paper, or a beautiful classroom/advisor/research-group presentation artifact. Produce a full paper package, not a short explanation: LaTeX source, sections, appendices, quality audit, and compiled PDF when LaTeX is available.

For Draft Demo Mode, put a visible disclosure near the beginning of the output:

`Draft Demo Notice: This manuscript is for classroom/demo use only. Simulated participants, data, quotes, codes, and findings are illustrative and must n