3rd Annual APSA Workshop on Quantitative Methods

As artificial intelligence becomes increasingly integrated into research and empirical analysis, Large Language Models (LLMs) have rapidly emerged as powerful tools for political scientists. By enabling new approaches to data collection, text analysis, research design, and statistical workflows, LLMs offer researchers unprecedented capacity to address contemporary political questions at scale, while also raising important questions about ethics, transparency, and replication.

This one-day workshop is dedicated to equipping students and early career researchers (ECRs) with the theoretical grounding and practical skills necessary to incorporate LLMs effectively and responsibly into quantitative political research, bridging foundational knowledge with hands-on application.

Applications are invited from students and early career researchers across the social sciences. We particularly welcome applications from historically underrepresented groups, including women, Aboriginal and Torres Strait Islander scholars, first-generation students, ethnic minorities, and members of the LGBTQIA+ community.

Program:

The workshop will feature sessions led by experienced scholars:

• Foundations of LLMs for Quantitative Political Research: What LLMs Do and How They Produce Answers
• Using LLMs to Design Research, Run Simulations, and Explore Models
• Using LLMs for Coding, Data Preparation, and Statistical Analysis
• Ethics, Bias, Transparency, and Replication Standards for LLM Use
• Applications

Session 1. LLMs Unplugged: Building a Language Model from Scratch
Presenter: Dr. Ben Swift (ANU)

This opening session demystifies LLMs by having participants build one themselves using nothing more than pen, paper, and dice. Through a hands-on activity, participants will train a simple bigram model, generate text from it, and discover the core mechanics that underpin contemporary LLMs: tokens, vocabularies, training, and probabilistic text generation. From this foundation, the session scales up to explain embeddings, transformer architectures, and why LLMs sometimes produce confident but incorrect outputs. No coding or technical background is required.

Session 2. Claude Code for Academics: An AI Agent for Empirical Research
Presenter: Alessandro Spina (UTSydney)

This session introduces agentic AI tools, assistants that go beyond chat-based interaction to read project files, execute code, and work alongside researchers across the full arc of an empirical project. The session demonstrates how an AI agent can support quantitative research workflows, including preparing and cleaning data, writing and reviewing analytical code, generating tables and figures, drafting and editing papers, and producing teaching materials. Participants will also see the practical scaffolding that makes these tools reliable for research, including project memory, rules, and verification, alongside a discussion of common failure modes.

Session 3. Introducing quallmer: An AI-Assisted Tool for Text Analysis
Presenter: Seraphine F. Maerz (University of Melbourne)

This session moves from concepts to application, introducing LLM-assisted text analysis as a scalable method for quantitative research. Traditional qualitative coding is in-depth but labour-intensive, while conventional quantitative text methods (dictionaries, topic models, supervised classifiers) are systematic and scalable but often miss context and nuance. LLMs open up a middle path: context-aware classification of large corpora that can be replicated, validated, and audited. Using quallmer, an open-source R package co-developed by the presenter and Kenneth Benoit (SMU), participants will work through workflows for structured coding, replication, validation against human-coded standards, and generating audit trails with executable replication materials.

This event is organised by the ANU School of Politics & International Relations and the Quantitative Methods Research Group of the Australian Political Studies Association (AusPSA).

Coordinators:Dr Thiago Nascimento da Silva (SPIR), Dr Constanza Sanhueza Petrarca (SPIR).

Funding: Limited funding is available to contribute towards domestic travel costs for participants travelling from outside Canberra.

Expression of Interest Deadline: 11 May 2026.

Participants will be notified of selection by 15 May 2026.