DAG construction from qual themes

Description

Translates themes and mechanisms from qualitative research into an explicit directed acyclic graph that can guide measurement and causal analysis.

Qualitative research produces rich mechanism stories but rarely commits them to a testable structure. This method converts coded themes into nodes and stated mechanisms into directed edges, producing a DAG that encodes assumptions about what causes what. Each edge must be traceable to evidence (quotes, observations) or labelled as an assumption. The resulting graph identifies which variables must be measured, which are confounders, mediators or colliders, and what quantitative design could test the story. Rigour comes from evidence tracing per edge, deliberately adding plausible unmeasured common causes, checking acyclicity, and testing the graph's implied conditional independencies against data where available.

Attributes

Loss type
Latent
Stage
Definition
Effort
Days

Steps

  1. List candidate variables from the theme codebook, phrasing each as a measurable quantity (e.g. perceived interface trust, frequency of double-checking).
  2. For each causal statement in the data, add a directed edge and log the supporting excerpts; mark edges without direct evidence as assumptions.
  3. Add plausible unmeasured common causes (e.g. financial literacy, market conditions) as latent nodes, even if not mentioned by participants.
  4. Check acyclicity; represent feedback loops by time-indexing variables (trust at t1 to behaviour at t2).
  5. Review the DAG with researchers and domain experts in a workshop, recording disagreements as alternative graphs.
  6. Derive testable implications (conditional independencies) with DAGitty and plan the measures and design needed to test key paths.

Sources

  • DAGs+SEM Playbook (Vivek Gopinath)
  • Causal Inference: What If (Miguel Hernán & James Robins)
  • Statistical Rethinking (Richard McElreath)
  • Causality (Judea Pearl)
  • DAGs+SEM Playbook (Gopinath)
  • Causal Inference (Hernán & Robins)
  • Statistical Rethinking (McElreath)
  • The Inquiry Series, Ch 46 — Directed Acyclic Graphs — Making Assumptions Explicit

Also known as

  • Qualitative-to-causal diagramming
  • Causal model elicitation

Categories

  • Research Operations
  • Research Design & Rigor