Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

✨ add chapter 8 #6

Merged
merged 1 commit into from
Sep 4, 2024
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
105 changes: 97 additions & 8 deletions 08_causal-paths-and-closing-back-doors.Rmd
Original file line number Diff line number Diff line change
Expand Up @@ -2,23 +2,112 @@

**Learning objectives:**

- THESE ARE NICE TO HAVE BUT NOT ABSOLUTELY NECESSARY
- identify paths in DAGs
- understand the difference between good and bad paths, and front-door vs. back-door paths
- understand how to Close paths
- find colliders
- understand how to use paths to test your DAG

## SLIDE 1 {-}
## Paths

- ADD SLIDES AS SECTIONS (`##`).
- TRY TO KEEP THEM RELATIVELY SLIDE-LIKE; THESE ARE NOTES, NOT THE BOOK ITSELF.
> A path between two variables on a causal diagram is a description of the set of arrows and nodes you visit when “walking” from one variable to another.

## Meeting Videos {-}
- tell us how two variables are related
- how are B and C related?

### Cohort 1 {-}
![](images/ch8_closing-back-doors/causalpaths-example-1.png)

> The paths can tell us the road we want to walk on, and also the road we want to avoid.

## Finding all paths

1. Start at the treatment variable
2. Follow one of the arrows coming in/out of the treatment variable to find another variable
3. Follow one of the arrows coming in or out of *that* variable
4. Keep repeating step 3 until you either come to a variable you’ve already visited (loop!), or find the outcome variable (path)
5. Every time you find a path/loop, back up one and try a different arrow in/out until you’ve tried them all. Then, back up again and try all those arrows

Let's practice!

![](images/ch8_closing-back-doors/causalpaths-basiclook-1.png)

## Solution

![](images/ch8_closing-back-doors/solutions_1.png)


## Let's try another one: Wine and Health

![](images/ch8_closing-back-doors/causalpaths-wine-1.png)


## Solution

![](images/ch8_closing-back-doors/solutions_2.png)

## Path types

- good vs. bad
- front door vs. back door
- beware direct vs. indirect effects

> Good Paths are the reasons why the treatment and outcome variables are related that you think should “count” for your research question. Bad paths are the paths that shouldn’t count, in other words the alternate explanations.

> Paths where all the arrows face away from Treatment are also known as front door paths. The rest would then be “back door paths.”

## Closing paths

> A path is Open if all of the variables along that paths are allowed to vary.

So we close a patch by

- picking a specific data set (e.g., no drug users)
- adding a variable on the path as control variable
- matching

So if we want to get the effect of *Wine* on *Lifespan*, which variables do we have to control for?

![](images/ch8_closing-back-doors/causalpaths-wine-1.png)

## Colliders

> A variable is a collider on a particular path if, on that path, both arrows point at it.

- A path is closed by default if a collider sits on it
- The path opens up if the collider is controlled for

> Because once you control for the collider, the two variables pointing to the collider become related.

We can avoid colliders by

- not controlling for them
- not choosing a sample where the collider is automatically controlled for

## Using paths to test the DAG

1. List all of the paths between A and B.
1. Consider only the open paths.
1. Do what you need to do to make sure they’re all Closed.
1. If A and B are still related to each other (placebo test), that means there must be some other path you didn’t account for.

Interpretation of placebo tests:

- We have a model, so not accounting for all effects is okay.
- small but nonzero relationship: fine
- enormous and super-strong relationship: 😱

## Meeting Videos {.unnumbered}

### Cohort 1 {.unnumbered}

`r knitr::include_url("https://www.youtube.com/embed/URL")`

<details>
<summary> Meeting chat log </summary>

```
<summary>Meeting chat log</summary>

```
LOG
```

</details>
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file added images/ch8_closing-back-doors/solutions_1.PNG
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file added images/ch8_closing-back-doors/solutions_2.PNG
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.