Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Seven basic rules for causal inference

pedermisager.org

Post reply on HN
Print view View original

51–60 of 71 posts

Re: Seven basic rules for causal inference

#51

Post by kyllo »

levocardia wrote: post #47

>Controlling for a collider leads to correlation This is a big one that most people are not aware of. Quite often, in economics, medicine, and epidemiology, you'll see researchers adjust for everything in their regression model: income, physical activity, education, alcohol consumption, BMI, ... without realizing that they could easily be inducing collider bias. A much better, but rare, approach is to sit down with s…

Collider bias or "Berkson's Paradox" is a fun one, there lots of examples of it in everyday life: https://en.wikipedia.org/wiki/Berkson%27s_paradox

Re: Seven basic rules for causal inference

#52

Post by kyllo »

islewis wrote: post #37
jdhwosnhw wrote: post #36

Earlier quoted context omitted.

This isnt a correction to your post, but a clarification for other readers: correlation implies dependence, but dependence does not imply correlation. Conversely, two variables share non-zero mutual information if and only if they are dependent.

Could you give some examples of dependence without correlation?

> A sailor is sailing her boat across the lake on a windy day. As the wind blows, she counters by turning the rudder in such a way so as to exactly offset the force of the wind. Back and forth she moves the rudder, yet the boat follows a straight line across the lake. A kindhearted yet naive person with no knowledge of wind or boats might look at this woman and say, “Someone get this sailor a new rudder! Hers is broken!” He thinks this because he cannot see any relationship between the movement of the rudder and the direction of the boat.

https://mixtape.scunning.com/01-introduction#do-not-confuse-...

Re: Seven basic rules for causal inference

#53

Post by 082349872349872 »

Vecr wrote: post #7
shiandow wrote: post #6

This is missing my favourite rule. 0. The directions of all arrows not part of a collider are statistically meaningless.

What's not part of a collider? Good luck with your memory in that case.

I'm guessing they mean that given a bunch of correlated nodes but no collider (in which case the casual graph must be a tree of some sort) you not only don't know if the tree be bushy or linear, you don't even know which node may be the root.

(bushy trees, of which there are very many compared with linear ones, would be an instance of Gwern's model* of confounds being [much] more common than causality?)

* https://news.ycombinator.com/item?id=41291636

Re: Seven basic rules for causal inference

#54

Post by mister_jervis »

Vecr wrote: post #5
lordnacho wrote: post #2

This is brilliant. The whole causal inference thing is something I only came across after university, either I missed it or it is a hole in the curriculum, because it seems incredibly fundamental to our understanding of the world. The thing that made be read into it was a quite interesting sentence from lesswrong, saying that actually the common idea that correlation does not imply causation is wrong. Now it's not wr…

When did you go to university? The terminology here came from Pearl 2000, and it probably took years and years after that to diffuse out.

[flagged]

Re: Seven basic rules for causal inference

#56

Post by curiousgal »

nomilk wrote: post #55

Humble reminder of how easy R is to use. Download and install R for your operating system: https://cran.r-project.org/bin/ Start it in the terminal by typing: R Copy/paste the code from the article to see it run!

Can't use R without RStudio. It so much better than the terminal.

Re: Seven basic rules for causal inference

#57

Post by nomilk »

curiousgal wrote: post #56
nomilk wrote: post #55

Humble reminder of how easy R is to use. Download and install R for your operating system: https://cran.r-project.org/bin/ Start it in the terminal by typing: R Copy/paste the code from the article to see it run!

Can't use R without RStudio. It so much better than the terminal.

Agree RStudio makes R a dream, but isn't necessary for someone to run the code in the article =)

Re: Seven basic rules for causal inference

#58

Post by Vecr »

082349872349872 wrote: post #53
Vecr wrote: post #7

Earlier quoted context omitted.

What's not part of a collider? Good luck with your memory in that case.

I'm guessing they mean that given a bunch of correlated nodes but no collider (in which case the casual graph must be a tree of some sort) you not only don't know if the tree be bushy or linear, you don't even know which node may be the root. (bushy trees, of which there are very many compared with linear ones, would be an instance of Gwern's model* of confounds being [much] more common than causality?) * https://new…

Right, but your memory functions as a collider, if there are literally no colliders anywhere you by definition won't be able to remember anything.

Re: Seven basic rules for causal inference

#59

Post by bubblyworld »

chrsig wrote: post #50

> Rule 8: Controlling for a causal descendant (partially) controls for the ancestor perhaps this is a quaint or wildly off base question, but an honest one, please forgive any ignorance: Isn't this essentiallydefining the partial derivative? Should one arrive at the calculus definition of a partial derivative by following this?

You probably could if you interpret that sentence very creatively. But I think it's useful to remember that this is mathematics, and words like "control", "descendant" and "ancestor" have specific technical meanings (all defined in the article, I believe).

The technical meaning of that sentence has to do with probability theory (probability distributions, correlation, conditionals), and not so much calculus (differentiable functions, limits, continuity).

Re: Seven basic rules for causal inference

#60

Post by gweinberg »

islewis wrote: post #37
jdhwosnhw wrote: post #36

Earlier quoted context omitted.

This isnt a correction to your post, but a clarification for other readers: correlation implies dependence, but dependence does not imply correlation. Conversely, two variables share non-zero mutual information if and only if they are dependent.

Could you give some examples of dependence without correlation?

Imagine your data points look like a U. There's no (lineral) correlation between x and y, you are equally likely to have a high value of y when x is high or low. But low values of y are associated with medium values of x, and a high value of y means x will be very high or very low.
Post reply on HN