back to list

Quackcast 679 - Correlation doesn't equal causation

Ozoneocean at 12:00AM, March 19, 2024
likes!

LISTEN on our player!
Or TuneinRadio - https://tunein.com/podcasts/Books–Literature/Drunkduck-Quackcast-p1150194/

The phrase “Correlation doesn't equal causation” is something associated with science and statistics but it really applies to EVERYTHING and that's important to understand. But was does it mean? If a bunch of things happened at the same time, those things aren't necessarily related or causal. An example Tantz gives is that statistics show in the summer there are more drownings and that people eat more ice cream. That means that those two things are correlated. We know they aren't causal though: ice ream doesn't drown people and people drowning don't cause people to eat more ice cream… the third hidden variable is that it's summer: it's the rise in temperature that causes people to want more ice cream and to swim more, which increases the chances of drowning.

I was thinking of the correlation causation fallacy when I was musing on the topic of history. There's this idea that if you know a lot about history it will give you a lot of information on current events, but this is heavily flawed by our tendency to create artificial connections between events, we come up with stories that sound good and plausible and make us feel better for why things are connected. Think of all the pop-science and pop-history books that come out and easily explain world events and complicated things in history. They're all pretty much bullshit because they fall for the causation fallacy: this happened which caused this, that and this, rather than things all happening at the same time for other reasons. This is also related to the hindsight fallacy, where we look back at events and incorrectly think a conclusion should have been obvious because we can see how things ended.

Because of this, while knowledge of history is very useful, that use is more limited than we think so it helps to know current events too, especially from an outside perspective so that we're not as fooled by false connections and mistaken causal relationships.
But why do I say this affects everything? Any story you hear or come up with to explain something is a victim of this. My Quackcast topics on the evolution of elves or goblins in fantasy are a victim of this, when people talk about the motivating factors of a serial killer they do this, when we talk about WW2, the current Israeli/Palestinian conflict, or the reason you drink coffee in the morning we fall afoul of these fallacies. It's what drives conspiracy thinking, folk histories and urban myths.

It even affects things like self image: are we influenced by the images we see or are the images we see influenced by how we want to look or are there other factors?

A key example we mention in the Quackcast was a pop-science story about how the use of lead in fuel made people dumber and lead to more violence. A truly moronic conclusion, very easily debunkable and yet people as esteemed as pop-sceince communicator Veritasium were fooled by it. I realise I fall afoul of it every time I come up with a story for what influenced me to do comics, photography, or cosplay- I have at least 5 different stories that explain anything I do, all of them make sense and seem perfectly true to me at the time, but in reality they're a product of the hindsight bias, the causation fallacy and being selective with data and variables.

Can you think of a time you've fallen afoul of this? If you can't you're probably not thinking hard enough ;)

This week Gunwallace did not have time for a new theme but he suggested that we put up the theme to PleaseRewind again because it's a great comic that is currently being reposted! PleaseRewind - Quiet threat, creepy, seeping, strumming, thrumming, coming CLOSER, inside, peering around, waiting to begin. This is a quiet track filled with an undertone of urgency suggested by the constant quick rhythm and lonely guitar.

Topics and shownotes

Links

Featured comic:
Sandra's Day - https://www.theduckwebcomics.com/news/2024/mar/12/featured-comic-sandras-day/

Featured music:
PleaseRewind - https://www.theduckwebcomics.com/Classic_Please_Rewind/ - by Paneltastic, rated T.

Special thanks to:
Gunwallace - http://www.virtuallycomics.com
Ozoneocean - https://www.theduckwebcomics.com/user/ozoneocean
Tantz Aerine - https://www.theduckwebcomics.com/user/Tantz_Aerine/
Banes - https://www.theduckwebcomics.com/user/Banes/


VIDEO exclusive!
Become a subscriber on the $5 level and up to see our weekly Patreon video and get our advertising perks!
- https://www.patreon.com/DrunkDuck
Even at $1 you get your name with a link on the front page and a mention in the weekend newsposts!

Join us on Discord - https://discordapp.com/invite/7NpJ8GS

next Quackcast: Intelligence in fiction

comment

anonymous?

Ozoneocean at 8:46PM, March 20, 2024

The world is out to get you... in a totally non-caring, impersonal way LOL

plymayer at 4:50AM, March 20, 2024

Of course it also doesn't mean the world isn't out to get you either.

plymayer at 4:50AM, March 20, 2024

Just because you are paranoid it doesn't mean the world is out to get you.

Ozoneocean at 9:36PM, March 19, 2024

@PaulEberhardt - that sounds like it could hve been good if it wasn't for the evil AI crap :(

Ozoneocean at 9:35PM, March 19, 2024

@Gunwallace - that graph is 100% true and real and not something I drew in 2 seconds because I was too scared to make a joke about breast size increase over time!

Ozoneocean at 9:34PM, March 19, 2024

@PaulEberhardt - yup, they indicate a relationship, it's just that a simplistic way of looking at it, as you know, is a causal one... when in reality it's most likely some very indirect factor like time.

Ozoneocean at 9:32PM, March 19, 2024

@Marcorossi - Yes, it doesn't equal causation but it can indicate there's an indirect relationship in there that results in both factors seeming to correlate :) Shooting deaths is a good example of that: More guns are not the direct cause of more gun deaths but they can show a relationship... i.e. a more gun obsessed community, lax gun laws, etc.

PaulEberhardt at 12:38PM, March 19, 2024

I wanted to recommend Tyler Vigen's Spurious Correlations website earlier, as I remembered it to be a fun thing that makes the same point. It still is, at the core. However, I saw he has taken to using AI to generate cartoon illustrations for it and come up with funny fake explanations for the mock correlations. So I don't recommend it now. Shame on him!

Gunwallace at 12:01PM, March 19, 2024

There's something funky about that graph. How can the number of Quackcasts go down over time? I think someone has fudged the data on this one ;-)

Tantz_Aerine at 9:46AM, March 19, 2024

@marcorossi, you're so right. But the dudes concentrating on outliers to make a generalization forget that in statistics, outliers are eliminated from the sample altogether. Tell them that next time.

PaulEberhardt at 8:16AM, March 19, 2024

Ah, one of my favourite topics. I think you did an excellent job on it. Of course everyone has fallen afoul of this at some point, including me, it's just how we humans are wired. Even more: I'll still say, trust your instincts even if this is one way in which they might lead you astray. As long as you know that and question them every now and then, it's not much of a problem. The thing is, correlations often occur for a reason even if we can't spot it right away, as long as they're not made from totally random input.

marcorossi at 6:00AM, March 19, 2024

On the one hand it is true that correlation is not causation, but on the other, the only way we can spot causation is through correlation. For example, some years ago I was discussing with a USA guy about weapon control, and he was very much a pro-weapons guy. At some point I found a site that did show the (unsurprising) statistic that in USA states with more weapons, there a re more people killed by weapons, and his answer was "correlation is not causation". Which is true in the abstract, but if one uses this too much it becomes impossible to have a fact-based discussion. Another common thing is when there is a correlation, but there are some outliers, and people concentrate to the outlier to negate the causation (also common when speaking of weapon control).


Forgot Password
©2011 WOWIO, Inc. All Rights Reserved Mastodon