Our approach is minimalist – we do not code the strength of a link

💐 Causal Mapping: A garden of ideas

(⚒️ Work in progress!)

Glossary
- Glossary
Causal mapping – overview
Task 1 – Gathering causal mapping data
Task 2 – Causal coding – minimalist style
Task 2 & 3 Key ideas and conventions
Task 3 – Answering questions – General
Task 3 – Answering questions – Individual questions
Causal mapping in evaluation
Causal Mapping as QDA
Causal Map app and alternatives
- Causal mapping with kumu
Deductive coding with AI
- Intro – deductive auto-coding
- The deductive autocoding prompt
Inductive coding with AI
Improving rigour in the use of AI in social science
Qualia
Case studies
Getting philosophical
AI and the wider world
Finally
Causal Map App

© Causal Map Ltd 2025 ·

Chapters▼

Chapters

Home Glossary Causal mapping – overview Task 1 – Gathering causal mapping data Task 2 – Causal coding – minimalist style Task 2 & 3 Key ideas and conventions Task 3 – Answering questions – General Task 3 – Answering questions – Individual questions Causal mapping in evaluation Causal Mapping as QDA Causal Map app and alternatives Deductive coding with AI Inductive coding with AI Improving rigour in the use of AI in social science Qualia Case studies Getting philosophical AI and the wider world Finally Causal Map App

/

Pages▼

Pages in this chapter

Task 2 – Introduction Minimalist coding does not help with blockers and enablers Our approach is minimalist – we code only bare causation Our approach clearly distinguishes evidence from facts and does not automatically warrant causal inferences Our approach is minimalist – factors are not variables 1a A minimalist approach to coding helps capture what people actually say 1b A minimalist approach to coding makes aggregation easier 1c A minimalist approach to coding does not code absences Our approach is minimalist – we do not code the strength of a link In a causal mapping dataset there is no need for a special table of factors Factor labels – a creative challenge Factor label tags – coding factor metadata within its label Factor labels – semi-quantitative formulations can help Factor labels – do not over-generalise Coding with and using link metadata Link metadata – Sentiment Link metadata – Time reference Link metadata – quality of evidence Research on LLMs' ability to detect causal claims A formalisation of causal mapping

🌻 Our approach is minimalist – we do not code the strength of a link

Backlinks

Coding with and using link metadata