Methods Guide

Start here if you have not settled on a method. Choose one by asking what evidence you have and what claim you want to make.

No research question yet?

A method answers a question, so start with your puzzle, your question, and its rival answers.

Go to Puzzles and Questions

Method selection

Match your project to a method

Answer the questions below and use the result as a place to begin.

What is your main material? What is your main analytical task? How comfortable are you with code?

Your starting point

Answer a question to begin

Choose an option above and a recommended section will appear here. Not sure yet? Browse the method cards below.

Browse method cards

Browse Method Cards

Foundations

Puzzles and Questions

Move from a topic to a puzzle, a simple research question, and rival answers.

Foundations

Literature Review

Group sources by the answer they support and write an argument about them.

Foundations

Analytical Framework

Link your question and answers to the evidence and method that can weigh them.

Workflow

Building a Corpus

Start here if the archive, search terms, or inclusion rules are still fuzzy.

Workflow

LLMs and Compute

Decide whether you need OCR, an LLM, or a GPU, and request help with setup.

Qualitative

Comparative Case Study

For two or more cases that share enough context to make their differences count.

Qualitative

Process Tracing

For questions where timing matters and each link in the causal chain needs evidence.

Qualitative / mixed

Framing Analysis

For projects about how an issue is packaged in headlines, speeches, or policy text.

Qualitative

Discourse Analysis

For claims about how language builds identities or turns one view into common sense.

Computational prep

Preprocessing

Decide what counts as a token and what gets dropped. Document each choice.

Computational

Topic Analysis

A first map of a large corpus. The model points to clusters. You still name them.

Computational

Sentiment Analysis

Useful when tone matters and the score can be checked against your own texts.

Computational

Word Embeddings

For similarity, conceptual drift, or retrieval when keyword searches are too brittle.

Combining Methods

Qualitative and computational methods work together when each has a defined role. Common pairings include these.

Framing analysis + topic analysis. A topic model surfaces candidate frames in a large corpus, and you read closely within each cluster.
Discourse analysis + keyword-in-context tooling. The computational side locates passages worth reading, while the interpretive judgment stays with you.
Comparative case study + descriptive corpus statistics. Aggregate measures such as volume or keyword prevalence can support the cross-case interpretation.
Process tracing + digital archives. Computational search helps identify causal-process observations you might otherwise miss.

Other Methods To Consider

This guide does not cover these methods in depth, but some projects need them.

Qualitative interviewing + thematic analysis. Semi-structured interviews coded for repeated themes. Common in all programs, especially fieldwork-based theses.
Content analysis. Rule-based coding of textual material, quantitative or qualitative. Common in IR and media-adjacent topics.
Survey methods and quantitative analysis. Statistical analysis of original or secondary survey data. Common in MAIR especially, and in BAIS theses with a quantitative focus.
Archival research. Close reading of historical documents and official records. Common in Korean Studies and history-focused MAAS theses.

For general planning, see Getting Started, Step 4: Building Your Analytical Framework. For assessment, see Assessment Standards, Application of Knowledge.