# Goddammit, learn the difference between correlation and causality!! **URL:** \ **Category:** The BBQ Pit\ **Created:** [July 25, 2002, 1:41pm UTC](https://boards.straightdope.com/t/goddammit-learn-the-difference-between-correlation-and-causality/120759 "2002-07-25T13:41:44Z")\ **Posts on this page:** 20\ **Page:** 1
**Author:** ![Spoonbender](https://avatars.discourse-cdn.com/v4/letter/s/a8b319/32.png) [@Spoonbender](https://boards.straightdope.com/u/Spoonbender)\ **Post date:** [July 25, 2002, 1:41pm UTC](https://boards.straightdope.com/t/goddammit-learn-the-difference-between-correlation-and-causality/120759/1 "2002-07-25T13:41:44Z")
I fucking hate it when the news media does this, and they do it every single day. Some statistical study shows a correlation between X and Y, and they run the headline “Study Shows X Causes Y.” Case in point: [Study: Living together may lead to breakups](http://www.cnn.com/2002/US/07/24/cdc.marriagereport/index.html) And of course, if you read deep enough into the article, the researchers themselves state in plain freakin’ English that the study does NOT imply a causality, yet as usual the reporter (or in this case, at least, the author of the headline) misses the point. Someday we’re going to see the headline, “Lung Cancer Leads to Smoking.” :rolleyes: ---
**Author:** ![gruven](https://avatars.discourse-cdn.com/v4/letter/g/a9a28c/32.png) [@gruven](https://boards.straightdope.com/u/gruven)\ **Post date:** [July 25, 2002, 2:08pm UTC](https://boards.straightdope.com/t/goddammit-learn-the-difference-between-correlation-and-causality/120759/2 "2002-07-25T14:08:00Z")
You took the words right out of my mouth. ---
**Author:** ![THespos](https://avatars.discourse-cdn.com/v4/letter/t/8c91f0/32.png) [@THespos](https://boards.straightdope.com/u/THespos)\ **Post date:** [July 25, 2002, 2:09pm UTC](https://boards.straightdope.com/t/goddammit-learn-the-difference-between-correlation-and-causality/120759/3 "2002-07-25T14:09:30Z")
Oh, God. This is one of my pet peeves. I absolutely HATE it when news media try to establish cause-effect relationships where there are none. Before news media report on a cause-effect relationship, they need to consider quite a few things: 1. That there might not be a causality relationship at all, 2. That the reverse of the statement they’re trying to make might be true. (e.g. - Unca Cece’s example of “pot smoking causes apathy” vs. “apathy causes pot smoking”). 3. The apparent cause-effect relationship might be due to an unseen third factor. (i.e. - If a certain town has a higher rate of breast cancer than most other towns, maybe it’s not something in the water. Maybe it’s due to a higher percentage of women in the population, who are more at risk for breast cancer than men.) 4. Changes in behavior warrant a second look at trends. In the case of Spoonbender’s example, of course more people who live together before getting married are breaking up. That’s not necessarily due to the living together situation - it’s probably due to the fact that more couples are living together before they get married. Higher population of couples living together = higher volume of breakups, but what is this as a percentage of the total number of people living together before getting married? And even if that percentage is growing, does this necessarily imply a cause-effect relationship? Are there external factors that play a role? 5. Jumping the gun is bad. If someone notices a correlation between kids who play console video games and carpal tunnel, it merits a second look, not a report on how “video games cause carpal tunnel.” It requires a deeper look at the habits of kids who play console video games. Maybe they’re using computer keyboards more. Maybe they’re doing something else that requires repetitive stress on their fingers and wrists. One doesn’t know until one does a study looking at all potential contributors. Maybe journalists who don’t understand statistics and how to intepret research methodology should keep a statistician on staff to help out when they’re thinking of running stories like this. -TFH ---
**Author:** ![Revtim](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/revtim/32/1042_2.png) [@Revtim](https://boards.straightdope.com/u/Revtim)\ **Post date:** [July 25, 2002, 2:23pm UTC](https://boards.straightdope.com/t/goddammit-learn-the-difference-between-correlation-and-causality/120759/4 "2002-07-25T14:23:21Z")
Some idiot on the radio this morning claimed that subscriptions to women’s magazines causes divorce. Same bullshit. ---
**Author:** ![CrankyAsAnOldMan](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/crankyasanoldman/32/11670_2.png) [@CrankyAsAnOldMan](https://boards.straightdope.com/u/CrankyAsAnOldMan)\ **Post date:** [July 25, 2002, 2:24pm UTC](https://boards.straightdope.com/t/goddammit-learn-the-difference-between-correlation-and-causality/120759/5 "2002-07-25T14:24:07Z")
Well, you know, Marriage causes divorce, too. ---
**Author:** ![BlinkingDuck](https://avatars.discourse-cdn.com/v4/letter/b/eada6e/32.png) [@BlinkingDuck](https://boards.straightdope.com/u/BlinkingDuck)\ **Post date:** [July 25, 2002, 2:26pm UTC](https://boards.straightdope.com/t/goddammit-learn-the-difference-between-correlation-and-causality/120759/6 "2002-07-25T14:26:53Z")
I am a statistician… Your idea of keeping a stat man on staff won’t work. They’ll just be bypassed because they would spoil the story. Even where I work, I am sometimes bypassed because they know I’ll find their wonderful conclusion to not be true. Fortunately, the owner of the company greatly frowns on this since he likes conclusions to actually have a high probability of actually being true. ---
**Author:** ![mouthbreather](https://avatars.discourse-cdn.com/v4/letter/m/a698b9/32.png) [@mouthbreather](https://boards.straightdope.com/u/mouthbreather)\ **Post date:** [July 25, 2002, 2:27pm UTC](https://boards.straightdope.com/t/goddammit-learn-the-difference-between-correlation-and-causality/120759/7 "2002-07-25T14:27:21Z")
Being alive may lead to dying. The research continues. ---
**Author:** ![Shalmanese](https://avatars.discourse-cdn.com/v4/letter/s/45deac/32.png) [@Shalmanese](https://boards.straightdope.com/u/Shalmanese)\ **Post date:** [July 25, 2002, 2:32pm UTC](https://boards.straightdope.com/t/goddammit-learn-the-difference-between-correlation-and-causality/120759/8 "2002-07-25T14:32:33Z")
> [@](#): > > \*Originally posted by mouthbreather \* > \*\*Being alive may lead to dying. > > The research continues. \*\* I am currently on a controversial project to debunk this myth 🙂 ---
**Author:** ![AHunter3](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/ahunter3/32/368_2.png) [@AHunter3](https://boards.straightdope.com/u/AHunter3)\ **Post date:** [July 25, 2002, 2:51pm UTC](https://boards.straightdope.com/t/goddammit-learn-the-difference-between-correlation-and-causality/120759/9 "2002-07-25T14:51:09Z")
One of my favorite examples is left-handedness and longevity. Let’s say you did a study in the middle 1990s on age at time of death and “handedness”, using an American 20th century data table as the data source. Your study shows a significant positive correlation between right-handedness and greater lifespan. What’s going on here? (Does anyone doubt that the typical interpretation would say that right-handed people live longer?) FACTOR C?: In the early part of the 20th century in America, there was a bit of social pressure to “be” right-handed whether you were or not; teachers would take the pencil out of your left hand and try to make you learn to write with your right hand. In the later 20th century, this lessened somewhat. Assuming, therefore, that people who were “switched” in this fashion would identify themselves as right-handed, a random individual born in the early part of the 20th century in America was more likely to be identified as right-handed than a random individual born in the latter part. AN OBSERVATION: At any given time (including, for example, the time of their death), people who were born **more** than 50 years before that time are substantially more likely to have reached the age of 50 than people who were born **less** than 50 years before that time. Replace “50” with “70” or “85” or any other number and this statement remains valid. ---
**Author:** ![Koxinga](https://avatars.discourse-cdn.com/v4/letter/k/4af34b/32.png) [@Koxinga](https://boards.straightdope.com/u/Koxinga)\ **Post date:** [July 25, 2002, 2:56pm UTC](https://boards.straightdope.com/t/goddammit-learn-the-difference-between-correlation-and-causality/120759/10 "2002-07-25T14:56:55Z")
> [@](#): > > \*Originally posted by Shalmanese \* > \*\* > > I am currently on a controversial project to debunk this myth 🙂 \*\* This [Onion article](http://www.theonion.com/onion3102/deathrate.html) suggests that you’ve got an uphill battle there, bub. ---
**Author:** ![Necros](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/necros/32/11086_2.png) [@Necros](https://boards.straightdope.com/u/Necros)\ **Post date:** [July 25, 2002, 3:07pm UTC](https://boards.straightdope.com/t/goddammit-learn-the-difference-between-correlation-and-causality/120759/11 "2002-07-25T15:07:35Z")
I dunno, **Doghouse Reilly**. I’ll bet you that every one of those people consumed some sort of liquid before they died. I think that the “drinking liquids causes death” idea is an important avenue of investigation that the Onion ignored. ---
**Author:** ![Coldfire](https://avatars.discourse-cdn.com/v4/letter/c/a587f6/32.png) [@Coldfire](https://boards.straightdope.com/u/Coldfire)\ **Post date:** [July 25, 2002, 3:23pm UTC](https://boards.straightdope.com/t/goddammit-learn-the-difference-between-correlation-and-causality/120759/12 "2002-07-25T15:23:36Z")
Somewhat related. 🙂 [Dutch newspaper comic “Fokke en Sukke”](http://www.bulloki.nl/Plaatjes/foksukger.gif) The TV displays a warning box in the same style as it is printed on cigarette packs in Europe these days. A football match is playing in the background. The headline reads _“Fokke and Sukke are glued to the TV”_. One says to the other, _“Football can make Germany world champions”_. Says the other, _“Good initiative, those warnings.”_. 😃 ---
**Author:** ![SuaSponte](https://avatars.discourse-cdn.com/v4/letter/s/67e7ee/32.png) [@SuaSponte](https://boards.straightdope.com/u/SuaSponte)\ **Post date:** [July 25, 2002, 3:25pm UTC](https://boards.straightdope.com/t/goddammit-learn-the-difference-between-correlation-and-causality/120759/13 "2002-07-25T15:25:04Z")
[scene: on the beach in Delaware. Several high school students sunning themselves. A wind kicks up.] Friend: Where’s the wind coming from? Sua: What? Friend: I mean, there aren’t any trees here. Sua: Huh? Friend: (patiently) You see, trees cause wind when they wave back and forth. Sua: Oh. Thanks. Sua ---
**Author:** ![IzzyR](https://avatars.discourse-cdn.com/v4/letter/i/2acd7d/32.png) [@IzzyR](https://boards.straightdope.com/u/IzzyR)\ **Post date:** [July 25, 2002, 3:35pm UTC](https://boards.straightdope.com/t/goddammit-learn-the-difference-between-correlation-and-causality/120759/14 "2002-07-25T15:35:44Z")
> [@](#): > > \*Originally posted by THespos \* > 4) Changes in behavior warrant a second look at trends. In the case of Spoonbender’s example, of course more people who live together before getting married are breaking up. That’s not necessarily due to the living together situation - it’s probably due to the fact that more couples are living together before they get married. Higher population of couples living together = higher volume of breakups, but what is this as a percentage of the total number of people living together before getting married? And even if that percentage is growing, does this necessarily imply a cause-effect relationship? Are there external factors that play a role? It does not appear that you’ve read the article. The study does NOT claim - as you seem to assume - that the fact that there has been an increase in breakups at the same time as an increase in living together is itself suggestive of causality. What it DOES say is that couples who lived together were compared to couples who did not, and had a higher rate of breakups. So your specific point is moot. The larger point is still true, as was noted by the study authors, as mentioned in the OP. What they noted is that it is possible that the two populations differ - IOW that people who live together before marriage are the type of people who are (on average) more predisposed to breakup anyway. So that the living together is a sign, but not a cause, of an increased likelihood of breakup. In general, I agree with the OP that this issue is a common shortcoming of numerous studies and media reports. (Does paying violent video games or watching violent media cause kids to be violent, or are kids who are already predisposed violence more likely to participate in such activity?) I would say this particular instance is a less egregious example - there was a qualifier in the title, and the disclaimer was noted in the article. ---
**Author:** ![stofsky](https://avatars.discourse-cdn.com/v4/letter/s/dfb087/32.png) [@stofsky](https://boards.straightdope.com/u/stofsky)\ **Post date:** [July 25, 2002, 4:11pm UTC](https://boards.straightdope.com/t/goddammit-learn-the-difference-between-correlation-and-causality/120759/15 "2002-07-25T16:11:43Z")
> [@](#): > > \*Originally posted by Coldfire \* > \*\*\*"Fokke and Sukke \* \*\* Just tell me that these names don’t mean what they look like in English! :eek: ---
**Author:** ![THespos](https://avatars.discourse-cdn.com/v4/letter/t/8c91f0/32.png) [@THespos](https://boards.straightdope.com/u/THespos)\ **Post date:** [July 25, 2002, 4:35pm UTC](https://boards.straightdope.com/t/goddammit-learn-the-difference-between-correlation-and-causality/120759/16 "2002-07-25T16:35:45Z")
_ **IzzyR** wrote…_ > [@](#): > > It does not appear that you’ve read the article. The study does NOT claim - as you seem to assume - that the fact that there has been an increase in breakups at the same time as an increase in living together is itself suggestive of causality. What it DOES say is that couples who lived together were compared to couples who did not, and had a higher rate of breakups. So your specific point is moot. Yep. Ya got me there. I didn’t read the article and made some assumptions that turned out to not be true. My bad. However, while this in itself may _suggest_ causality, I think it’s irresponsible for journalists to imply causality when that might not be the case. In general, journalists tend to ignore or fail to report external circumstances that might contribute to the positive correlation. You’re right, though. I was trying to wind down that post and actually had another example in there, but it was too confusing and I took it out. So I made an incorrect assumption. 30 lashes with a wet noodle. ---
**Author:** ![Coldfire](https://avatars.discourse-cdn.com/v4/letter/c/a587f6/32.png) [@Coldfire](https://boards.straightdope.com/u/Coldfire)\ **Post date:** [July 25, 2002, 8:39pm UTC](https://boards.straightdope.com/t/goddammit-learn-the-difference-between-correlation-and-causality/120759/17 "2002-07-25T20:39:29Z")
> [@](#): > > \*Originally posted by stofsky \* > \*\*Just tell me that these names don’t mean what they look like in English! :eek: \*\* Heh. Actually, they _are_ genuine Dutch names, but they’re very archaic. Combine that with the general knowledge of the English language, and your average Dutchman will have the same chuckle you had. 🙂 ---
**Author:** ![The\_Ryan](https://avatars.discourse-cdn.com/v4/letter/t/7feea3/32.png) [@The\_Ryan](https://boards.straightdope.com/u/The_Ryan)\ **Post date:** [July 25, 2002, 9:26pm UTC](https://boards.straightdope.com/t/goddammit-learn-the-difference-between-correlation-and-causality/120759/18 "2002-07-25T21:26:29Z")
> [@](#): > > \*Originally posted by Spoonbender \* > I fucking hate it when the news media does this, and they do it every single day. Some statistical study shows a correlation between X and Y, and they run the headline “Study Shows X Causes Y.” Are you sure that it is not in fact the headline which causes the statitiscal study showing correlation, not vice versa? 😛 ---
**Author:** ![sailor](https://avatars.discourse-cdn.com/v4/letter/s/a587f6/32.png) [@sailor](https://boards.straightdope.com/u/sailor)\ **Post date:** [July 25, 2002, 9:43pm UTC](https://boards.straightdope.com/t/goddammit-learn-the-difference-between-correlation-and-causality/120759/19 "2002-07-25T21:43:00Z")
- Couples who lived together before marrying had a greater chance of getting divorced than those who didn’t - Correlation does not imply causality - Nope, but the correlation remains and for some decisions the correlation is all that is relevant. ---
**Author:** ![Dr.Rieux](https://avatars.discourse-cdn.com/v4/letter/d/aeb1de/32.png) [@Dr.Rieux](https://boards.straightdope.com/u/Dr.Rieux)\ **Post date:** [July 26, 2002, 12:01am UTC](https://boards.straightdope.com/t/goddammit-learn-the-difference-between-correlation-and-causality/120759/20 "2002-07-26T00:01:15Z")
> [@](#): > > \*Originally posted by CrankyAsAnOldMan \* > \*\*Well, you know, Marriage causes divorce, too. \*\* That’s how I’ve managed to avoid it so far…😉 [Next page](https://boards.straightdope.com/t/goddammit-learn-the-difference-between-correlation-and-causality/120759.md?page=2)