Causal Cakes

[Website Under Construction]

The causalcakes package creates visual representations of causal mechanisms using "causal cake" diagrams. These diagrams are rarely employed in the social sciences, but can provide an alternative to directed-acyclic graphs (DAGs) and are useful for making explicit a variety of external validity dimensions. They excel at modeling structural mechanisms, such as INUS causation—conditions that are central to external validity inference.

What is a Causal Cake?

🎯

Core Factor

The central causal force (center circle)

🔗

Necessary Conditions

Moderators that must be present (outer ring)

📊

Outcome

The effect produced when conditions align

Think of it like baking: you need flour AND eggs AND heat to make a cake. Each ingredient is necessary (you can't skip any), but insufficient on its own. Together, they're sufficient. Often there are multiple recipes (pathways) that can produce the same outcome.

Installation

# Install from GitHub devtools::install_github("externalvalidity/causalcakes") # Load the package library(causalcakes)

Quick Start

Single Causal Pathway

library(causalcakes) # Example TBA

This creates a diagram with a core causal factor at the center, surrounded by necessary conditions, with an arrow pointing to the outcome.

Multiple Pathways (INUS Structure)

library(causalcakes) # Example TBA

This shows multiple alternative pathways to the same outcome—each sufficient on its own, but neither necessary.

Example Applications

🗳️ Voter Turnout

Two pathways: performance information vs. malfeasance information, each with different necessary conditions.

Coming Soon

🏢 Technology Adoption

Three mechanisms: economic pressure, competitive pressure, and regulatory requirements.

Coming Soon

📋 Policy Compliance

Normative pathway (2 conditions) vs. instrumental pathway (5 conditions) with asymmetric complexity.

Coming Soon

🎓 Academic Success

Four pathways: academic preparation, financial resources, social support, and institutional fit.

Coming Soon

Customization Options

Causal cake diagrams can be customized in several ways:

Integration with EVinference

Causal cakes integrate with the broader EVinference ecosystem. Mechanisms declared using EVinference's declare_mechanism() function can be automatically converted to causal cake visualizations. This allows researchers to move seamlessly between formal mechanism specification and visual representation, making it easier to communicate theoretical models and identify which necessary conditions might differ across target contexts.

Resources

Getting Started Vignette Complete Examples GitHub