Causal Models

Goals

1. Describe causal models
2. Apply causal models
3. Build and improve them
4. Get the rewards from Causal Models

What?

A model is a simplified view of reality.

A causal model is a type of model that specifies a relationship because input variables and desired result variables.

Why?

A causal model relates (input variables) to (desired results). Since great-results are generally better than so-so-results, we usually describe causal models as follows

So the reason why Causal Models are worthwhile is that they equip actors (people and groups) to achieve great results (success). Causal models are a “Recipe for Success”. They put you in the driver’s seat so that you can reliably create success.

Even better, CMs equip actors with a tool for trouble shooting. When you want great results and you are not getting them, then look at the drivers of success (input variables) and figure out which one to alter.

Examples of Causal Models

Apply Pie

1. Input Variables: Ingredients + Cooking Instructions + Oven
2. Desired Result Variables: Tastes Great + Pleasing Aroma + Pleasing Appearanc
3. Arrow Notation: (Ingredients) + (Instructions) + (Oven) → (Great Apply Pie)

If the apple pie is not yet great, then tweak one or more of the drivers (input variables) until the results are at the level of quality you desire.

Fire

1. Input Variables: Heat + 0xygen + Fuel
2. Desired Result Variable: Fire
3. Arrow Notation: Heat + 0xygen + Fuel → Fire

In this example, the goal state (fire) is simply an outcome, not a “great result.” Causal models can be used for examining any results and is not restricted to desirable ones.

Notes to be processed

- Essence
    - A causal model is a recipe for success
- What
    - A causal model explains what produces what in a system.
    - It specifies elements and the cause → effect relationships between them.
- Examples
    - heat + O2 + fuel → fire
    - balance + right motions → skilled swimming
- Why
    - Intentionally increase the probability of producing desired outcomes.
    - To explain why outcomes occur rather than merely describing patterns.
    - To identify where intervention will change results.
    - To support prediction, diagnosis, and improvement.
- Who
    - Used by anyone trying to understand or change outcomes:
        - scientists
        - engineers
        - teachers
        - decision-makers
        - learners
- When
    - Used when outcomes matter and need to be explained, predicted, or improved.
    - Especially useful after observing unexpected or undesirable results.
- Where
    - Applied in any domain with cause-and-effect structure:
        - learning
        - behavior
        - health
        - economics
        - engineering
        - social systems
- How
    - Identify key elements in the system.
    - Specify directional cause → effect links.
    - Describe the mechanism by which causes produce effects.
    - Use the model to reason about changes and interventions.
- How much
    - Varies in precision and scope:
        - simple models explain core dynamics
        - detailed models capture fine-grained effects
    - The best model is as simple as possible while still useful.
- Summary
    - A causal model explains how changes propagate through a system.
    - It enables prediction, diagnosis, and deliberate improvement of outcomes.