# when is correlation not causation? **URL:** \ **Category:** Factual Questions\ **Created:** [February 19, 2009, 11:32pm UTC](https://boards.straightdope.com/t/when-is-correlation-not-causation/486633 "2009-02-19T23:32:08Z")\ **Posts on this page:** 20\ **Page:** 2
**Author:** ![Snarky\_Kong](https://avatars.discourse-cdn.com/v4/letter/s/a183cd/32.png) [@Snarky\_Kong](https://boards.straightdope.com/u/Snarky_Kong)\ **Post date:** [February 20, 2009, 5:45pm UTC](https://boards.straightdope.com/t/when-is-correlation-not-causation/486633/21 "2009-02-20T17:45:57Z")
> [@schindler](#): > > Johnny Damon playing for the Red Sox correlated well with the Yankees losing games against them. Since baseball is not basketball, soccer, hockey, or any other symmetrical team sport, there doesn’t seem to have been a causal link between his presence and the Yankees inferior performance as a team-- unless you believe in juju, karma, voodoo, or effecting metaphysical change by having really long hair. Uh… what? If A-rod starts playing peewee baseball there sure as hell is causation in effect when his team wins all the games. Maybe I’m misunderstanding you. ---
**Author:** ![Xema](https://avatars.discourse-cdn.com/v4/letter/x/9de053/32.png) [@Xema](https://boards.straightdope.com/u/Xema)\ **Post date:** [February 20, 2009, 9:55pm UTC](https://boards.straightdope.com/t/when-is-correlation-not-causation/486633/22 "2009-02-20T21:55:09Z")
> [@ultrafilter](#): > > Outside of a properly designed and controlled experiment, correlation never implies causation. This is the essential point. The best you can say is that correlation suggests the possibility of causation. ---
**Author:** ![CC](https://avatars.discourse-cdn.com/v4/letter/c/f1d935/32.png) [@CC](https://boards.straightdope.com/u/CC)\ **Post date:** [February 20, 2009, 10:50pm UTC](https://boards.straightdope.com/t/when-is-correlation-not-causation/486633/23 "2009-02-20T22:50:38Z")
Yes, I’m aware of the distinction. What I had asked for were some simple dramatic examples of that. Thanks, Dopers. xo, C. ---
**Author:** ![YogSothoth](https://avatars.discourse-cdn.com/v4/letter/y/8edcca/32.png) [@YogSothoth](https://boards.straightdope.com/u/YogSothoth)\ **Post date:** [February 20, 2009, 11:34pm UTC](https://boards.straightdope.com/t/when-is-correlation-not-causation/486633/24 "2009-02-20T23:34:13Z")
The ice cream example I learned about was that as ice cream sales go up, drownings go up. But does ice cream cause drownings? Of course not, its because when its hot people buy ice cream and going swimming! And drown each other in tubs of ice cream… \>\_\> ---
**Author:** ![Sublight](https://avatars.discourse-cdn.com/v4/letter/s/439d5e/32.png) [@Sublight](https://boards.straightdope.com/u/Sublight)\ **Post date:** [February 21, 2009, 7:45am UTC](https://boards.straightdope.com/t/when-is-correlation-not-causation/486633/25 "2009-02-21T07:45:20Z")
A good example that turned up in GD recently: Most scientific breakthroughs over the past 500 years were made in Europe and the US. Those areas are predominantly Christian. Therefore, Christianity is required for scientific breakthroughs. ---
**Author:** ![flex727](https://avatars.discourse-cdn.com/v4/letter/f/258eb7/32.png) [@flex727](https://boards.straightdope.com/u/flex727)\ **Post date:** [February 21, 2009, 5:55pm UTC](https://boards.straightdope.com/t/when-is-correlation-not-causation/486633/26 "2009-02-21T17:55:27Z")
> [@Sublight](#): > > A good example that turned up in GD recently: > > Most scientific breakthroughs over the past 500 years were made in Europe and the US. > > Those areas are predominantly Christian. > > Therefore, Christianity is required for scientific breakthroughs. Arguments of the form: If A, then B, even when true, do not mean that A is _required_ for B. ---
**Author:** ![CC](https://avatars.discourse-cdn.com/v4/letter/c/f1d935/32.png) [@CC](https://boards.straightdope.com/u/CC)\ **Post date:** [February 21, 2009, 8:41pm UTC](https://boards.straightdope.com/t/when-is-correlation-not-causation/486633/27 "2009-02-21T20:41:37Z")
I see that as different. _If_ A, then B means that A caused B. Perhaps also C can cause B, but If A means that B’s existence, in this case, would not have occurred without A. ---
**Author:** ![Mosier](https://avatars.discourse-cdn.com/v4/letter/m/7ea924/32.png) [@Mosier](https://boards.straightdope.com/u/Mosier)\ **Post date:** [February 21, 2009, 9:14pm UTC](https://boards.straightdope.com/t/when-is-correlation-not-causation/486633/28 "2009-02-21T21:14:10Z")
A = “eating too much ice cream” B = “getting fat” If you eat too much ice cream, then you will get fat. But you don’t need to eat too much ice cream to get fat. You can also get fat by eating too much pasta, even if you never eat ice cream. A is NOT required for B. ---
**Author:** ![CC](https://avatars.discourse-cdn.com/v4/letter/c/f1d935/32.png) [@CC](https://boards.straightdope.com/u/CC)\ **Post date:** [February 21, 2009, 9:33pm UTC](https://boards.straightdope.com/t/when-is-correlation-not-causation/486633/29 "2009-02-21T21:33:00Z")
A= falling into the water B = getting wet In this case, if A, then B. The obverse is not necessarily true. If B, then A. Also, it cannot be that A does not result in B. If this is so, how can we not say that A caused B? ---
**Author:** ![Hostile\_Dialect](https://avatars.discourse-cdn.com/v4/letter/h/e495f1/32.png) [@Hostile\_Dialect](https://boards.straightdope.com/u/Hostile_Dialect)\ **Post date:** [February 21, 2009, 10:00pm UTC](https://boards.straightdope.com/t/when-is-correlation-not-causation/486633/30 "2009-02-21T22:00:28Z")
> [@Vlad/Igor](#): > > Another example I heard several years ago was that there was an increase in DUIs and in the number of Baptist ministers. Corr=Caus would imply that Baptist ministers were responsible for the DUIs Or, amusingly, that DUIs were responsible for the Baptist ministers. Hostile Dialect, Hostile Dialect, Narcissist ---
**Author:** ![schindler](https://avatars.discourse-cdn.com/v4/letter/s/ba9def/32.png) [@schindler](https://boards.straightdope.com/u/schindler)\ **Post date:** [February 21, 2009, 11:50pm UTC](https://boards.straightdope.com/t/when-is-correlation-not-causation/486633/31 "2009-02-21T23:50:49Z")
> [@Mosier](#): > > A = “eating too much ice cream” > B = “getting fat” > > If you eat too much ice cream, then you will get fat. But you don’t need to eat too much ice cream to get fat. You can also get fat by eating too much pasta, even if you never eat ice cream. A is NOT required for B. It wasn’t clear whether you meant to deny the causal link between eating to much ice cream and getting fat. I assume, since that would be absurd, that you’re just explaining that causation does not imply that "A and only A can ever cause B. " ---
**Author:** ![schindler](https://avatars.discourse-cdn.com/v4/letter/s/ba9def/32.png) [@schindler](https://boards.straightdope.com/u/schindler)\ **Post date:** [February 21, 2009, 11:54pm UTC](https://boards.straightdope.com/t/when-is-correlation-not-causation/486633/32 "2009-02-21T23:54:14Z")
> [@Snarky\_Kong](#): > > Uh… what? > > If A-rod starts playing peewee baseball there sure as hell is causation in effect when his team wins all the games. Maybe I’m misunderstanding you. If we were discussing a ringer in little league, sure, but Damon didn’t play exceptionally well during that world series and the Yankees were no peewee team… he was alright, i guess, but nothing special. It just so happened that more at-bats for Damon correlated with losses for the Yankees. The New Hampshire and Iowa caucuses are also an example, and I think they’re like the Damon example because they share the same confusing feature- yes, iowa and new hampshire are RELEVANT, but it’d be crazy to call them the CAUSE. ---
**Author:** ![Rigamarole](https://avatars.discourse-cdn.com/v4/letter/r/77aa72/32.png) [@Rigamarole](https://boards.straightdope.com/u/Rigamarole)\ **Post date:** [February 22, 2009, 12:09am UTC](https://boards.straightdope.com/t/when-is-correlation-not-causation/486633/33 "2009-02-22T00:09:58Z")
> [@Bill\_Door](#): > > The trees moving does not make the wind blow. I don’t know that this is a good example, because there _is_ a causation relationship here, it’s just the other way around (the wind blowing does cause tees to move). Better to use an example where there is absolutely no causal relationship between two factors at all. ---
**Author:** ![Hostile\_Dialect](https://avatars.discourse-cdn.com/v4/letter/h/e495f1/32.png) [@Hostile\_Dialect](https://boards.straightdope.com/u/Hostile_Dialect)\ **Post date:** [February 22, 2009, 12:53am UTC](https://boards.straightdope.com/t/when-is-correlation-not-causation/486633/34 "2009-02-22T00:53:18Z")
Flags always ruffle when the trees sway, but swaying trees do not cause flag ruffling. Hostile Dialect, Hostile Dialect, Narcissist ---
**Author:** ![elfkin477](https://avatars.discourse-cdn.com/v4/letter/e/a9a28c/32.png) [@elfkin477](https://boards.straightdope.com/u/elfkin477)\ **Post date:** [February 22, 2009, 1:59am UTC](https://boards.straightdope.com/t/when-is-correlation-not-causation/486633/35 "2009-02-22T01:59:42Z")
My nutrition professor told us that there’s a high correlation between the number of TVs and cars a family owns, and the prevelance of heart disease in a population. Yet no one argues that owning TVs and cars (directly) causes heart disease. The only link, of course, has to do with lifestyle choices that are often associated with the owning these things, not just owning these items at all. ---
**Author:** ![KarlGauss](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/karlgauss/32/3713_2.png) [@KarlGauss](https://boards.straightdope.com/u/KarlGauss)\ **Post date:** [February 22, 2009, 2:52am UTC](https://boards.straightdope.com/t/when-is-correlation-not-causation/486633/36 "2009-02-22T02:52:04Z")
I had come here specifically to post about association of coronary heart disease and hormone replacement therapy, with hormone replacement apparently being **protective** in population studies but actually _promoting_ heart disease in the prospective, placebo controlled, randomized, double-blind trial. OK, so here’s another medical-type example (again involving coronary disease): People with the highest levels of vitamin E in their blood have the lowest incidence of coronary heart disease. Moreover, there is a ‘dose-response effect’ present, i.e. mid-range levels of vitamin E are associated with mid-range rates of coronary disease, and low levels of vitamin E are associated with high rates of coronary disease, etc. (As an aside, an extremely compelling mechanism was hypothesized to account for the apparent protective effect of vitamin E.) Of course, and as many of you probably know, when a “proper” randomized, prospective, double-blind, placebo-controlled study was undertaken, there was NO link found (either protective or promoting) between vitamin E and coronary heart disease. It’s because of results like the above (with hormone replacement looking beneficial but actually being deleterious, and vitamin E appearing to protect from heart attacks but actually being neutral in that regard) that the astute modern physician insists on a randomized controlled trial before subscribing to a treatment or test. ---
**Author:** ![AllWalker](https://avatars.discourse-cdn.com/v4/letter/a/db5fbb/32.png) [@AllWalker](https://boards.straightdope.com/u/AllWalker)\ **Post date:** [February 22, 2009, 4:41am UTC](https://boards.straightdope.com/t/when-is-correlation-not-causation/486633/37 "2009-02-22T04:41:16Z")
Cancer causes smoking. Think about it - someone with lung cancer experiences pain and difficulty breathing, and so are more likely to want a cigarette to ease their suffering. The data backs this up, too. 😃 As I get older, time goes forward. Thus when I die, there is a chance time will stop (it may keep continuing, under the philosophy that I am still growing older even though dead.) ---
**Author:** ![johnspartan](https://avatars.discourse-cdn.com/v4/letter/j/a183cd/32.png) [@johnspartan](https://boards.straightdope.com/u/johnspartan)\ **Post date:** [February 23, 2009, 7:15pm UTC](https://boards.straightdope.com/t/when-is-correlation-not-causation/486633/38 "2009-02-23T19:15:52Z")
At the football game, every time the band plays the scoreboard increases. Therefore, the scoreboard must be counting number of times the band plays? The events are correlated, but not cause & effect. There’s an independent causality for both events. ---
**Author:** ![Minnie\_Luna](https://avatars.discourse-cdn.com/v4/letter/m/4bbf92/32.png) [@Minnie\_Luna](https://boards.straightdope.com/u/Minnie_Luna)\ **Post date:** [February 23, 2009, 8:34pm UTC](https://boards.straightdope.com/t/when-is-correlation-not-causation/486633/39 "2009-02-23T20:34:40Z")
The situation I was taught was such: The incidence of breast cancer is higher in populations of people who use a purse or handbag. -or- Purse and/or handbag usage correlates with breast cancer. ---
**Author:** ![CC](https://avatars.discourse-cdn.com/v4/letter/c/f1d935/32.png) [@CC](https://boards.straightdope.com/u/CC)\ **Post date:** [February 23, 2009, 8:41pm UTC](https://boards.straightdope.com/t/when-is-correlation-not-causation/486633/40 "2009-02-23T20:41:20Z")
**Mini** , that’s a very good one. One that continues to arise in public is: A large number of heroin users once smoked marijuana. Ergo, marijuana use causes heroin use. Almost as flawed as: Almost all heroin users drank milk as children. Ergo, drinking milk leads to heroin use. [Previous page](https://boards.straightdope.com/t/when-is-correlation-not-causation/486633.md?page=1)