Computational & AI-Assisted Methods for Social Sciences
Companion website to the research design primer by Ji Ma
This site carries everything the book produced and everything that must stay current: computing setup, primers on the techniques the exercises assume, living tool guides, the studies database, and worked notebooks with their data. All of it is free, needs no account, and lives at one address.
If you are reading the book and hit a step that assumes knowledge you do not have yet, this is where to look first.
Jump to your chapter
The printed book points here at exact moments. If you arrived from one of those pointers, or you know which chapter you are working through, go straight in:
Each hub states what the chapter asks you to produce, what it assumes, and carries its notebooks, data, tool guide, and the studies filtered to its lens. Chapters 1, 11, and 12 are read rather than worked through, so they have no hubs; Chapter materials explains how the pairs fit together.
About the book
Computational & AI-Assisted Methods for Social Sciences: A Research Design Primer (Ji Ma, SAGE, forthcoming) teaches computational and AI-assisted methods as research design rather than as a catalogue of tools. Four stages organize the whole book: data management, concept representation, data analysis, and scientific communication. Theory chapters build the reasoning; exercise chapters apply it to a literature dataset you build from your own field, so the exercises cannot be completed by copying; and every AI-assisted technique arrives paired with the validation it requires.
It is written for graduate students and researchers in the social sciences, and assumes no programming background.
The book teaches through three complementary pieces, and it helps to know which one you are supposed to be in at any moment.
| What it does | Where it lives | |
|---|---|---|
| Theory chapters (1, 2, 3, 5, 7, 9, 11, 12) | Build the concepts and the reasoning: what a measure commits you to, what a method can and cannot claim, what makes a result trustworthy | The book |
| Exercise chapters (4, 6, 8, 10) | Put the same concepts to work on your own data, one deliverable at a time | The book |
| This site | Environment setup, technique primers, current tool lists, downloadable templates and datasets: the material that would be stale in print within a year | Here |
The book stays deliberately language-agnostic, so its exercises work in Python, R, or Stata. This site is more concrete: it shows one worked path (usually Python) so that you have something runnable in front of you while you adapt the idea to your own tools.
The book’s argument in compressed, citable form is on the For researchers page; course fit and a fifteen-week shape are on For instructors.
Start where you are
New to computational work?
Run your first analysis in the browser with nothing installed, then take the three primers the exercises assume: text data, supervised learning, and prompting.
Teaching with the book?
Course fit, a fifteen-week shape, assessment that works when students use AI, and the pages that absorb the first-week support load.
Engaging it as research?
The book’s eight claims with chapter pointers, the frameworks as reusable instruments, and the materials with their limits stated.
The library
- CSS Empirical Studies Database: 122 published computational studies, annotated by design stage, searchable and downloadable.
- Worked notebooks and data: three notebooks with committed output, plus the sample network the Chapter 8 exercises run on.
- Tool guides: data management, automated coding, analysis, and communication, revised as software changes.
- Resource list: where to go beyond this book.
Every AI-assisted technique in this book comes paired with a way to check it. That pairing is the point, not a hedge: an unverified model output is not evidence, however fluent it sounds. When a page here shows you how to have a model code text, draft code, or propose a schema, it will also show you how to audit what came back. Please do not skip that half.