Claude Code for Social Scientists
A bilingual, open-source guide for social scientists who want to use Anthropic's Claude Code in their research, teaching, and academic writing. Written by a working clinical psychologist and PhD candidate — for researchers outside the English-speaking world as much as inside it — drawing on real academic production, not demonstration projects.
Status: v3.0.0 release. Twenty-one booklets in Turkish and English, human-reviewed and citation-audited, at least one in every one of the twelve categories, plus twenty companion Claude Code project skills that turn the booklets into repeatable workflows. Aggregate verified citations stand at 354, with zero fabricated. v3.0.0 refreshes the Journal of Open Source Education paper against this finished surface and adds the submission preparation package. The skills install with pip (social-cc-plugin) or as a Claude Code plugin.
TR readers: A Turkish overview lives at the bottom of this file. The full Turkish version is in README.tr.md. Every booklet exists as tr.md and en.md side by side.
What this guide is
A practical, evidence-led handbook for using Claude Code in academic work outside computer science. The audience is researchers in psychology, sociology, education, public health, communication, political science, anthropology, and adjacent fields. Every booklet is delivered in Turkish and English in full parallel.
The guide covers, across twelve thematic categories, the questions a social scientist actually faces:
- Foundations. What Claude Code is, how it differs from a chat window, where it earns its keep in scholarly work.
- Academic access. PubMed and Semantic Scholar MCPs, EZproxy and institutional VPN realities, ORCID, Zotero, DergiPark, ULAKBIM TR Dizin, HEAL-Link.
- Memory systems. Long-running vaults, persistent context, retrieval over a decade of notes, the Memory-as-Vault engineering pattern.
- Vault architecture. Folder discipline, MOCs (Maps of Content), Markdown conventions that survive software changes.
- Hooks and automation. Session lifecycle events, ritual hooks, lightweight CI for personal knowledge bases.
- MCP and plugins. Authoring, auditing, and curating Model Context Protocol servers for academic workflows.
- Academic writing. IMRAD scaffolding in Turkish and English, APA 7 with DOI discipline, journal fit, manuscript revision.
- Data analysis. Reproducible workflows, statistical test selection, qualitative coding, mixed-methods discipline.
- Ethics and IRB. Helsinki Declaration, COPE, WAME, ICMJE, STM 2025, EU AI Act 2024/1689 Art. 50, ENAI, KVKK, GDPR.
- Peer review. Rebuttal letters with traceability matrices, reviewer-response discipline, anti-AI-trace writing.
- Conference presentation. Slides, posters, lightning talks, networking sequences.
- Troubleshooting. When tools fail, when papers disagree, when reviewers ask the wrong question.
Each booklet is short, opinionated, and tested against the author's own academic practice.
Why bilingual
Turkish and English are presented in full parallel. There are roughly ninety million Turkish speakers globally and a large diaspora in Western Europe; Turkish-language academic AI resources are scarce relative to that demand. The gap is not incidental. A large 2026 survey of coding agents in the social sciences samples researchers in the United States and Canada and finds adoption skewed by career stage, gender, and institutional prestige (Anthropic, 2026). A guide written in full parallel from outside that frame is one concrete way to push against the skew. The English version exists so the work is reviewable by international colleagues, citable in English-language journals, and reachable through global academic search engines. Each booklet lives in a folder with tr.md and en.md as siblings. A continuous-integration check refuses any commit that breaks this pairing.
Who this is for, who it is not
It is for the assistant professor running a survey study, the PhD student writing a systematic review, the postdoc preparing an R&R, the lecturer designing a syllabus, the clinical researcher navigating IRB. It is for people who can read code but who do not want to spend a week learning a new toolchain to write one paragraph.
