Reply
 
Thread Tools Display Modes
  #51  
Old 01-06-2013, 08:00 PM
Kimmy_Gibbler Kimmy_Gibbler is offline
Guest
 
Join Date: Mar 2009
Posts: 3,712
Quote:
Originally Posted by SandyHook View Post
My findings were that 100% of them started out life by drinking milk.
This exemplifies another problem with the webcomic level of understanding. You can indeed run a correlation between two binary variables (X=drank milk as a child, yes or no; Y=became a violent criminal in adulthood, yes or no). But my guess is that since nearly everyone drank milk as child, the correlation coefficient would be close to zero. That is, if your X-series is nearly entirely 1s and your Y-series is a random sequence of 0s and 1s, no correlation would be demonstrated. Still further, it would show P(became a violent criminal in adulthood|drank milk as a child)= P(became a violent criminal in adulthood).

Last edited by Kimmy_Gibbler; 01-06-2013 at 08:03 PM.
  #52  
Old 01-06-2013, 08:01 PM
Ambivalid Ambivalid is offline
Guest
 
Join Date: Jun 2011
Location: In my head
Posts: 12,078
A good one is the claim that diet sodas cause weight gain, because many people who drank diet soda gained weight.
  #53  
Old 01-06-2013, 08:02 PM
Leo Bloom Leo Bloom is offline
Member
 
Join Date: Jun 2009
Location: Here
Posts: 11,579
People often claim that obesity is rife among poor people, so they can't be that poor.
  #54  
Old 01-06-2013, 08:06 PM
Leo Bloom Leo Bloom is offline
Member
 
Join Date: Jun 2009
Location: Here
Posts: 11,579
Quote:
Originally Posted by Kimmy_Gibbler View Post
This exemplifies another problem with the webcomic level of understanding. You can indeed run a correlation between two binary variables (X=drank milk as a child, yes or no; Y=became a violent criminal in adulthood, yes or no). But my guess is that since nearly everyone drunk milk as child, the correlation coefficient would be close to zero. That is, if you X-series is nearly entirely 1s and your Y-series is a random sequence of 0s and 1s, no correlation would be demonstrated.
I'm confused. You are fleshing out the milk/heroin false syllogism, or saying that there might be situations in which a true-quick assumption of the falseness is actually incorrect ("a problem with [this] web comic-type example")?

Sorry to be dense.

Last edited by Leo Bloom; 01-06-2013 at 08:07 PM.
  #55  
Old 01-06-2013, 08:19 PM
Blake Blake is offline
Member
 
Join Date: Mar 2001
Posts: 11,049
Quote:
Originally Posted by Chronos View Post
But obesity does cause consumption of diet drinks.
Not really. Many people drink them because they are overweight, no obese. Many people drink them because they fear becoming overweight/obese. Many people drink them because they are diabetic. Many people drink them because sugar is bad for their teeth.

IOW, while it's true that many people who are obese consume diet drinks for that reason, it's not by any means the only reason.
  #56  
Old 01-06-2013, 08:20 PM
TriPolar TriPolar is offline
Guest
 
Join Date: Oct 2007
Location: rhode island
Posts: 37,388
Quote:
Originally Posted by zoid View Post
Do you have a cite for this? I'm not saying you're incorrect but this really is something I've never heard before and I'd be facinated to read more but my Google-fu is really weak.
Quote:
Originally Posted by Leo Bloom View Post
Ditto. Not that one cite is a summer.
I'm not rushing to provide cites because it opens a can of worms. The internet is full of competing stories by advocates on either side making claims that are difficult to verify.

Here for example is a cite from an MD who claims to have analyzed WHO/CDC data. This same paper is also used by right-wing political sites that claim scientists are always lying to the public. So without further information about this doctor I have no idea whether he knows what he's talking about.

Here is an example of site that claims that smoking does cause cancer. Despite that claim, the experiment cited within it shows that exposing lung cells to cigarette smoke, and hydrogen peroxide on top of that, failed to cause cancer.

If I have the time and patience later on, I'll try to dig through the material again.

But the lack of a causal relationship between smoking and lung cancer is not new. I was surprised to hear about it myself, over 20 years ago.


Quote:
Originally Posted by njtt View Post
[Heavy sigh] And that is called being a cause of lung cancer. For just about any cause and effect pair in existence, you can find further events in the causal connection between them that may legitimately be viewed as intervening causes. Indeed, I am sure that, on your virus story (which,, incidentally, you have still not done anything to substantiate) there are events in the cell that intervene between the virus being there in a smoke damaged cell and the cell actually beginning to divide cancerously. They are all causes of the cancer too.
I never claimed that lung cancer was caused by a virus. The cause is unknown, and a virus is a hypothetical cause. Radon exposure is another.

There is some specific cause of lung cancer (assuming it's not magic), and plenty of research has shown cigarette smoking is not it. Only 1/3 of smokers ever develop lung cancer, and around 30% of people who do develop lung cancer are not smokers*. If you want to extend the chain of causation then you may as well call breathing a cause of a lung cancer.

*This is a tough are to straighten out. Until recently many people had smoked some cigarettes in their lifetime. There are some making the claim that even smoking one cigarette in a lifetime will lead to lung cancer. The percentage of people who have non-smoking related lung cancer is probably lower than 30%.

Before the point is lost, the cause of lung cancer is unknown. Cigarette smoking may one day turn out to be the cause. But that is not determined by the strength of the correlation. That will be determined by proof of a causative mechanism.
  #57  
Old 01-06-2013, 08:43 PM
Superhal Superhal is offline
Guest
 
Join Date: Sep 2009
Location: The Internet. Since '84.
Posts: 6,460
My favorite example was used in two different psych 100 classes I've taken:

In this community, there live 10,000 people and 200 storks. After 10 years, now there's 20,000 people and 400 storks. Why did the stork population increase?

After some spurious discussion about storks bringing babies and some garden-path thinking that that somehow the storks caused the increase in people (e.g. "More people moved in because the storks were a tourist attraction?") the usual answer is that storks eat trash, and thus as the population increased, there was enough trash to support a larger colony of birds.

What I like about it is that it flips the original idea (more storks caused more humans) and there is a real causative relationship, but in the opposite direction (more humans caused more storks.)

Last edited by Superhal; 01-06-2013 at 08:44 PM.
  #58  
Old 01-06-2013, 08:45 PM
Shmendrik Shmendrik is offline
Guest
 
Join Date: Nov 2007
Posts: 1,550
Quote:
Originally Posted by TriPolar View Post
There is some specific cause of lung cancer (assuming it's not magic), and plenty of research has shown cigarette smoking is not it. Only 1/3 of smokers ever develop lung cancer, and around 30% of people who do develop lung cancer are not smokers*. If you want to extend the chain of causation then you may as well call breathing a cause of a lung cancer.

*This is a tough are to straighten out. Until recently many people had smoked some cigarettes in their lifetime. There are some making the claim that even smoking one cigarette in a lifetime will lead to lung cancer. The percentage of people who have non-smoking related lung cancer is probably lower than 30%.
This is very confusing. Your statement that "there is some specific cause of lung cancer" is strange. There are many types of lung cancer and even more causes of lung cancer. That doesn't mean there is no causal relationship between tobacco smoke and some common types of lung cancer.

To extend your analogy, many people who are hit by cars do not suffer broken bones, and many people with broken bones have never been hit by a car, so therefore being hit by a car does not cause broken bones.
  #59  
Old 01-06-2013, 08:50 PM
TriPolar TriPolar is offline
Guest
 
Join Date: Oct 2007
Location: rhode island
Posts: 37,388
Quote:
Originally Posted by Shmendrik View Post
This is very confusing. Your statement that "there is some specific cause of lung cancer" is strange. There are many types of lung cancer and even more causes of lung cancer. That doesn't mean there is no causal relationship between tobacco smoke and some common types of lung cancer.
There is one type of lung cancer most strongly associated with cigarette smoking. The research I've seen separates that from the other forms.

Quote:

To extend your analogy, many people who are hit by cars do not suffer broken bones, and many people with broken bones have never been hit by a car, so therefore being hit by a car does not cause broken bones.
You are missing the point. We can say getting hit by a car causes broken bones because we can prove the causative relationship, not because of the high correlation between the two.
  #60  
Old 01-06-2013, 09:03 PM
Chronos Chronos is offline
Charter Member
Moderator
 
Join Date: Jan 2000
Location: The Land of Cleves
Posts: 73,079
Quote:
Not really. Many people drink them because they are overweight, no obese. Many people drink them because they fear becoming overweight/obese. Many people drink them because they are diabetic. Many people drink them because sugar is bad for their teeth.

IOW, while it's true that many people who are obese consume diet drinks for that reason, it's not by any means the only reason.
I never claimed that it was the only reason, nor an absolute reason: There are obese people who drink full-Calorie drinks, and there are people of healthy weight who drink diet drinks. But to the extent that there's a correlation between obesity and consumption of diet drinks, it's mostly because of obesity causing the consumption.
  #61  
Old 01-06-2013, 09:08 PM
Blake Blake is offline
Member
 
Join Date: Mar 2001
Posts: 11,049
Quote:
Originally Posted by TriPolar View Post
You are missing the point. We can say getting hit by a car causes broken bones because we can prove the causative relationship, not because of the high correlation between the two.
Well, no. Science can't "prove" anything. All it can do look at evidence for something, and then disprove the alternative explanations for the evidence. We believe that car crashes cause broken bones because the evidence suggests that, and the alternatives such as evil spirits don't stack up.

Nobody has ever proved a causal relationship between car crashes and broken bones. I doubt if anyone has ever even tried to investigate. All we know is that when people get involved in car crashes they often get broken bones, car crashes involve large impact forces, and bones can be broken by large impact forces. We combine those three observations and come up with the conclusion: car crashes cause broken bones.

And exactly the same applies to cancer and smoking. We know that smokers often get cancer. We know that smoke contains a range of carcinogens, and when know that when cells are exposed to carcinogens, they often become cancerous. We combine those three observations and come up with the conclusion: smoking causes cancer.

Neither conclusion has a "proven" causative relationship. Science doesn't work that way.
  #62  
Old 01-06-2013, 09:09 PM
Shmendrik Shmendrik is offline
Guest
 
Join Date: Nov 2007
Posts: 1,550
Quote:
Originally Posted by TriPolar View Post
You are missing the point. We can say getting hit by a car causes broken bones because we can prove the causative relationship, not because of the high correlation between the two.
Definitely not. If we take a bunch of pregnant mice, feed half of them alcohol, and observe increased birth defects in the alcohol group compared to the control group, we can infer causality even if we don't understand the mechanism.
  #63  
Old 01-06-2013, 11:12 PM
TriPolar TriPolar is offline
Guest
 
Join Date: Oct 2007
Location: rhode island
Posts: 37,388
Quote:
Originally Posted by Shmendrik View Post
Definitely not. If we take a bunch of pregnant mice, feed half of them alcohol, and observe increased birth defects in the alcohol group compared to the control group, we can infer causality even if we don't understand the mechanism.
You can take credit for indirectly causing a win with this. I was getting a headache looking for peered review papers on the topic, veered off on another direction, ended up here: Probabilisitic Causation. I can't argue against using that definition in this case.
  #64  
Old 01-07-2013, 05:17 AM
njtt njtt is offline
Guest
 
Join Date: Jul 2004
Location: U.K.
Posts: 12,068
Quote:
Originally Posted by TriPolar View Post
I never claimed that lung cancer was caused by a virus. The cause is unknown, and a virus is a hypothetical cause. Radon exposure is another.

There is some specific cause of lung cancer (assuming it's not magic), and plenty of research has shown cigarette smoking is not it. Only 1/3 of smokers ever develop lung cancer, and around 30% of people who do develop lung cancer are not smokers*. If you want to extend the chain of causation then you may as well call breathing a cause of a lung cancer.

*This is a tough are to straighten out. Until recently many people had smoked some cigarettes in their lifetime. There are some making the claim that even smoking one cigarette in a lifetime will lead to lung cancer. The percentage of people who have non-smoking related lung cancer is probably lower than 30%.

Before the point is lost, the cause of lung cancer is unknown. Cigarette smoking may one day turn out to be the cause. But that is not determined by the strength of the correlation. That will be determined by proof of a causative mechanism.
No. You have very fundamental misconceptions about the nature and logic of causality. Here is a primer (I was going to send you to the Stanford Encyclopedia of Philosophy rather than Wikipedia, as it would be more reliable, but everything they have seems to be far more advanced, and to take that basic level of understanding of causation for granted).

As I previously stated, there is no single cause of anything. Any particular event had indefinitely many causes, and many types of events have many types of causes. THE cause of lung cancer will never be known because there is no such, unique thing. On the other hand, it is among the most firmly established of scientific facts that smoking is a cause of lung cancer. Many other things may be causes of it too, some together with smoking and some quite independently of it.

Last edited by njtt; 01-07-2013 at 05:19 AM.
  #65  
Old 01-07-2013, 09:20 AM
TriPolar TriPolar is offline
Guest
 
Join Date: Oct 2007
Location: rhode island
Posts: 37,388
Quote:
Originally Posted by njtt View Post
No. You have very fundamental misconceptions about the nature and logic of causality. Here is a primer (I was going to send you to the Stanford Encyclopedia of Philosophy rather than Wikipedia, as it would be more reliable, but everything they have seems to be far more advanced, and to take that basic level of understanding of causation for granted).

As I previously stated, there is no single cause of anything. Any particular event had indefinitely many causes, and many types of events have many types of causes. THE cause of lung cancer will never be known because there is no such, unique thing. On the other hand, it is among the most firmly established of scientific facts that smoking is a cause of lung cancer. Many other things may be causes of it too, some together with smoking and some quite independently of it.
No such facts are established. And if you choose to call the Big Bang the cause of all things then there's no point in discussing such matters with you.
  #66  
Old 01-07-2013, 11:59 AM
Buck Godot Buck Godot is offline
Guest
 
Join Date: Mar 2010
Location: MD outside DC
Posts: 4,348
My favorite example (similar to max the Immortal)

Those Americans who regularly eat McDonalds Happy meals have a lower incidence of cardiovascular disease than those that don't.

Another interesting finding that came up in an analysis I was doing a while ago found that smokers had a better chance of surviving a heart attack than non-smokers. This was probably due to the fact that smoking caused heart attacks in a healthier group of people, while non-smokers would be less healthy when they had the heart attack.
  #67  
Old 01-07-2013, 12:08 PM
Andy L Andy L is offline
Member
 
Join Date: Oct 2000
Posts: 4,744
Quote:
Originally Posted by Buck Godot View Post
Those Americans who regularly eat McDonalds Happy meals have a lower incidence of cardiovascular disease than those that don't.
Nice. How about this one: Americans whose first name ends with a vowel are signficantly shorter than Americans whose first name does not end with a vowel
  #68  
Old 01-07-2013, 12:32 PM
UltraVires UltraVires is offline
Guest
 
Join Date: Jul 2007
Location: Bridgeport, WV, US
Posts: 12,541
Quote:
Originally Posted by TriPolar View Post
No such facts are established. And if you choose to call the Big Bang the cause of all things then there's no point in discussing such matters with you.
Not to speak for njtt, but that misstates his position. You are correct that if we take such a loose definition of "cause" that the Big Bang did cause everything.

As was said earlier, most disciplines whether that be law, science, or otherwise use "proximate cause." Generally speaking that means a cause which, without a new intervening or efficient cause, is linked to the harm.

So let's say that we hypothesize that cars cause lung cancer because most smokers drive to the store in cars, buy cigarettes, smoke them, cause the virus that you spoke of, and get lung cancer. That breaks down because the step between cars and buying cigarettes creates a new cause that breaks the previous string: buying cigarettes. Driving the car to the store doesn't necessitate or directly lead to buying cigarettes.

However, smoking cigarettes directly leads to the virus which directly causes the lung cancer. It's an unbroken string with no intervening cause.
  #69  
Old 01-07-2013, 12:50 PM
Leo Bloom Leo Bloom is offline
Member
 
Join Date: Jun 2009
Location: Here
Posts: 11,579
I haven't read it in a while, but The Chain of Chance, the detective/science-philosophy novel by Stanislaw Lem, explores causation quite entertainingly.
  #70  
Old 01-07-2013, 01:12 PM
Evil Economist Evil Economist is offline
Guest
 
Join Date: Jan 2009
Posts: 2,869
I used to read a magazine called "Iron Mind"* that covered a bunch of different strength sports. I ended my subscription after reading an article in which an entirely serious author wrote about how he noticed that shortly after he started hearing birds singing in the morning, plants started to grow. His conclusion was that birdsong made plants grow. If I remember correctly, he then concluded that it was the vibrations in the birdsong that made the plants grow, therefore vibrations were good for strength, therefore you could make your muscles grow through being exposed to certain types of vibrations, i.e. birdsong. It was my favorite article of all time, just for the way the guy started out so absurdly wrong and just kept digging.

I think it was the same author who noticed that his athletes who walked with their toes pointed out were worse athletes that those who walked with their toes pointing forward. The guy was apparently a high school strength coach, because if I remember correctly he had his players practice walking with their toes pointed forward, in order that they could become better athletes. (The more likely chain of causation is that excess fat causes you to walk with your feet pointing outward, which makes you slower on the field. Practing walking with your toes pointing forward won't help you at all.)

I used to read old weight training books. In one of them the author proudly recalled how his secretary had asked him for advice on how to lose weight. This guy went around studying all the secretaries and noticed that the thin ones sat more upright compared to the fat ones. He then recommended to his secretary that she sit more upright, and if she did this she would lose weight. What I loved most about that whole story was that the guy was so proud of his analysis and conclusion that he made it an entire chapter of his book, and that no one in the entire publishing chain from the writer to the editor to the proofreader to the librarian who put the book on the shelves noticed any problems with any of it.

*Edit: the magazine was called Milo.

Last edited by Evil Economist; 01-07-2013 at 01:17 PM.
  #71  
Old 01-07-2013, 03:09 PM
Hershele Ostropoler Hershele Ostropoler is offline
Guest
 
Join Date: Jun 2010
Location: BK.NY.US
Posts: 1,678
Quote:
Originally Posted by Blake View Post
And exactly the same applies to cancer and smoking. We know that smokers often get cancer. We know that smoke contains a range of carcinogens, and when know that when cells are exposed to carcinogens, they often become cancerous.
I'm thinking this is circular. The claim (of which I'm skeptical) is "smoking is not part of a causal chain that disproportionately leads to lung cancer." To rebut this with "cigarette smoke contains carcinogens" seems to me like assuming the conclusion.
  #72  
Old 01-07-2013, 10:00 PM
CookingWithGas CookingWithGas is offline
Charter Member
 
Join Date: Mar 1999
Location: Tysons Corner, VA, USA
Posts: 11,495
Quote:
Originally Posted by Hershele Ostropoler View Post
I'm thinking this is circular. The claim (of which I'm skeptical) is "smoking is not part of a causal chain that disproportionately leads to lung cancer." To rebut this with "cigarette smoke contains carcinogens" seems to me like assuming the conclusion.
Not at all. There are substances that are known a priori to be carcinogens, and they are found to be in cigarette smoke. The conclusion that they are carcinogens does not follow from an assumption that cigarette smoke causes cancer.
  #73  
Old 01-08-2013, 03:24 AM
UltraVires UltraVires is offline
Guest
 
Join Date: Jul 2007
Location: Bridgeport, WV, US
Posts: 12,541
Quote:
Originally Posted by Hershele Ostropoler View Post
I'm thinking this is circular. The claim (of which I'm skeptical) is "smoking is not part of a causal chain that disproportionately leads to lung cancer." To rebut this with "cigarette smoke contains carcinogens" seems to me like assuming the conclusion.
It would be like saying that drunk driving does not cause death by pointing out the fact that your Uncle Fred drives drunk 4 times a week and hasn't killed anyone. The exceptions don't always prove the rule.
  #74  
Old 01-08-2013, 03:50 AM
eschereal eschereal is offline
Guest
 
Join Date: Aug 2012
Location: Frogstar World B
Posts: 11,864
My favorite, twenty years ago, was the expert who said "Since Canada implemented UHC, 90% of Canadians have moved to within 100 miles [160km] of the US border." Not exactly in the actual realm of the original question, but breathtakingly deceitful.

Last edited by eschereal; 01-08-2013 at 03:51 AM.
  #75  
Old 01-08-2013, 06:04 AM
TokyoBayer TokyoBayer is online now
Guest
 
Join Date: Oct 2003
Location: Taiwan
Posts: 9,097
Quote:
Originally Posted by TriPolar View Post
*This is a tough are to straighten out. Until recently many people had smoked some cigarettes in their lifetime. There are some making the claim that even smoking one cigarette in a lifetime will lead to lung cancer. The percentage of people who have non-smoking related lung cancer is probably lower than 30%.
Actually, no need to guess. I'll provide a cite from the National Cancer Institute at the National Institute of Heath. If you have anything different, bring it on. and let's skip silly websites, OK?
Quote:
Lung cancer is the leading cause of cancer death among both men and women in the United States, and 90 percent of lung cancer deaths among men and approximately 80 percent of lung cancer deaths among women are due to smoking.
  #76  
Old 01-08-2013, 02:03 PM
CC CC is offline
Charter Member
 
Join Date: Apr 2000
Location: not elsewhere
Posts: 4,291
I'm loving the course on logic and causation unfolding here. However, as to the OP, a classic is this: Roosters crow before sunrise. Therefore, roosters cause the sun to rise.

Or is this a different but related case? - the post hoc fallacy.
  #77  
Old 01-08-2013, 02:32 PM
suranyi suranyi is offline
Member
 
Join Date: Sep 2000
Posts: 7,445
One of my favorites is this one: "A study discovered there was a strong correlation between foot size and mathematical problem solving ability."

The study included babies and children.
  #78  
Old 01-09-2013, 03:10 AM
drewtwo99 drewtwo99 is offline
BANNED
 
Join Date: Aug 2011
Location: Houston, Texas
Posts: 8,757
Quote:
Originally Posted by CC View Post
I'm loving the course on logic and causation unfolding here. However, as to the OP, a classic is this: Roosters crow before sunrise. Therefore, roosters cause the sun to rise.

Or is this a different but related case? - the post hoc fallacy.
It's similar. If you correlated the times that roosters crowed with the time the sun came up after taking lots of data, you could correlate the two, but not prove that one causes the other.
  #79  
Old 01-09-2013, 04:26 AM
Puzzler Puzzler is offline
Guest
 
Join Date: Aug 2006
Location: Israel
Posts: 476
Quote:
Originally Posted by Superhal View Post
My favorite example was used in two different psych 100 classes I've taken:

In this community, there live 10,000 people and 200 storks. After 10 years, now there's 20,000 people and 400 storks. Why did the stork population increase?

After some spurious discussion about storks bringing babies and some garden-path thinking that that somehow the storks caused the increase in people (e.g. "More people moved in because the storks were a tourist attraction?") the usual answer is that storks eat trash, and thus as the population increased, there was enough trash to support a larger colony of birds.

What I like about it is that it flips the original idea (more storks caused more humans) and there is a real causative relationship, but in the opposite direction (more humans caused more storks.)
But... but... Storks [do] Deliver Babies (p= 0.008) ...
  #80  
Old 01-09-2013, 06:00 PM
Senegoid Senegoid is offline
Guest
 
Join Date: Sep 2011
Location: Sunny California
Posts: 14,594
Quote:
Originally Posted by Andy L View Post
Nice. How about this one: Americans whose first name ends with a vowel are signficantly shorter than Americans whose first name does not end with a vowel
In this thread so far, have we made an explicit distinction between correlations that are just happenstance (that is, you measured two seemingly-unrelated variables from 1980 to 1990 and found a high correlation, but it appears to be totally coincidental and not repeatable), versus systematic correlations that are related by a common cause and are thus reliably repeatable (like the ice cream / snakebite thing)?
  #81  
Old 01-09-2013, 08:15 PM
Dr. Strangelove Dr. Strangelove is offline
Guest
 
Join Date: Dec 2010
Posts: 6,078
Quote:
Originally Posted by Max the Immortal View Post
People who eat Count Chocula for breakfast have a lower cancer rate than people who eat oatmeal for breakfast.
Also, people undergoing chemotherapy are more likely to die of cancer than people who aren't.
  #82  
Old 01-09-2013, 10:11 PM
Napier Napier is offline
Charter Member
 
Join Date: Jan 2001
Location: Mid Atlantic, USA
Posts: 8,857
Tweedledee: "I used to think correlation proved causation. Then I took a statistics class. Now I don't think that way anymore."

Tweedledum: "So, the class helped?"

Tweedledee: "Well, maybe."

Last edited by Napier; 01-09-2013 at 10:12 PM.
  #83  
Old 01-09-2013, 10:22 PM
CC CC is offline
Charter Member
 
Join Date: Apr 2000
Location: not elsewhere
Posts: 4,291
Quote:
Originally Posted by Napier View Post
Tweedledee: "I used to think correlation proved causation. Then I took a statistics class. Now I don't think that way anymore."

Tweedledum: "So, the class helped?"

Tweedledee: "Well, maybe."
psst. Look at post #6
  #84  
Old 01-10-2013, 10:02 AM
Andy L Andy L is offline
Member
 
Join Date: Oct 2000
Posts: 4,744
Quote:
Originally Posted by Senegoid View Post
In this thread so far, have we made an explicit distinction between correlations that are just happenstance (that is, you measured two seemingly-unrelated variables from 1980 to 1990 and found a high correlation, but it appears to be totally coincidental and not repeatable), versus systematic correlations that are related by a common cause and are thus reliably repeatable (like the ice cream / snakebite thing)?
I'm not sure if this thread has consistently distinguished between those cases, but just to be clear the correlation between height and first name is one of the cases which involves a common cause.
  #85  
Old 01-11-2013, 12:05 AM
Napier Napier is offline
Charter Member
 
Join Date: Jan 2001
Location: Mid Atlantic, USA
Posts: 8,857
Quote:
Originally Posted by CC View Post
psst. Look at post #6
Oops. How embarrassing. Though, not necessarily because you pointed it out to me.
  #86  
Old 01-11-2013, 09:57 AM
panache45 panache45 is offline
Member
 
Join Date: Oct 2000
Location: NE Ohio (the 'burbs)
Posts: 39,366
Dog owners prefer Ragu; cat owners prefer Prego.
  #87  
Old 01-11-2013, 10:06 AM
Leo Bloom Leo Bloom is offline
Member
 
Join Date: Jun 2009
Location: Here
Posts: 11,579
Quote:
Originally Posted by Andy L View Post
I'm not sure if this thread has consistently distinguished between those cases, but just to be clear the correlation between height and first name is one of the cases which involves a common cause.
The obviousness of this fallacy isn't obvious in this discussion for me at the moment. Because they're all humans and humans have names?
  #88  
Old 01-11-2013, 11:10 AM
Shakester Shakester is offline
Guest
 
Join Date: Dec 2008
Location: Melbourne, Australia
Posts: 2,911
To return, for a moment, to the OP:

A: What are you doing?
B: I'm applying tiger repellent.
A: But there are no tigers around here.
B: See? It works!
  #89  
Old 01-11-2013, 11:43 AM
Andy L Andy L is offline
Member
 
Join Date: Oct 2000
Posts: 4,744
Quote:
Originally Posted by Leo Bloom View Post
The obviousness of this fallacy isn't obvious in this discussion for me at the moment. Because they're all humans and humans have names?
"Americans whose first name ends with a vowel are signficantly shorter than Americans whose first name does not end with a vowel"

This is because women tend to be significantly shorter than men, and American first names for women tend to end in a vowel, while American first names for men rarely end in a vowel - thus "having a first name that ends with a vowel" is a good proxy for being shorter than average, even though the name didn't cause the shortness, and the shortness didn't directly cause the name.
  #90  
Old 01-11-2013, 12:22 PM
Baracus Baracus is offline
Guest
 
Join Date: Dec 2006
Location: Cary, NC
Posts: 1,530
Quote:
Originally Posted by Shakester View Post
To return, for a moment, to the OP:

A: What are you doing?
B: I'm applying tiger repellent.
A: But there are no tigers around here.
B: See? It works!
This, along with the crow causing the sun to rise and criminals drink milk as kids examples, is something more like confirmation bias than correlation not implying causation. There isn't actually a correlation between the variables. The observer is merely ignoring that people who don't use repellent aren't getting attacked (or that the sun always rises or that every kid drinks milk). In order for there to be a correlation, those things would have to not be true to some significant degree.

Last edited by Baracus; 01-11-2013 at 12:23 PM.
  #91  
Old 01-11-2013, 12:36 PM
njtt njtt is offline
Guest
 
Join Date: Jul 2004
Location: U.K.
Posts: 12,068
Quote:
Originally Posted by Andy L View Post
"Americans whose first name ends with a vowel are signficantly shorter than Americans whose first name does not end with a vowel"

This is because women tend to be significantly shorter than men, and American first names for women tend to end in a vowel, while American first names for men rarely end in a vowel - thus "having a first name that ends with a vowel" is a good proxy for being shorter than average, even though the name didn't cause the shortness, and the shortness didn't directly cause the name.
Well, OK, it is probably true that being female causes one to tend to be shorter than average, but does being female tend to cause one to have a name ending in a vowel? It seems to me that that might be just a mere coincidental correlation. Female names (in our culture) just happen to often end in vowels, but there is nothing about lacking a Y chromosome that makes that so.

Last edited by njtt; 01-11-2013 at 12:37 PM.
  #92  
Old 01-11-2013, 03:50 PM
Fuzzy Dunlop Fuzzy Dunlop is offline
Guest
 
Join Date: Apr 2007
Posts: 3,457
A person's foot size and their ability to do mathematics are highly correlated. It's a heteroskedastic relationship though - there's a much higher variance in mathematic ability among large foot people than small foot people, who are fairly uniformly limited in their ability to do math.
  #93  
Old 01-11-2013, 04:57 PM
Chronos Chronos is offline
Charter Member
Moderator
 
Join Date: Jan 2000
Location: The Land of Cleves
Posts: 73,079
Quote:
Well, OK, it is probably true that being female causes one to tend to be shorter than average, but does being female tend to cause one to have a name ending in a vowel? It seems to me that that might be just a mere coincidental correlation. Female names (in our culture) just happen to often end in vowels, but there is nothing about lacking a Y chromosome that makes that so.
Depends on how you want to slice it. If you start by assume that a culture typically has different names for boys and girls, then you could say that a child being a girl causes that child to have a culturally-female name. If you further assume from the start that we're working with the set of names that modern American culture considers female, then you could say that a child being a girl causes her to have one of the names from that set.
  #94  
Old 01-11-2013, 05:18 PM
bup bup is offline
Guest
 
Join Date: Sep 1999
Location: glenview,il,usa
Posts: 11,905
Quote:
Originally Posted by Max the Immortal View Post
People who eat Count Chocula for breakfast have a lower cancer rate than people who eat oatmeal for breakfast.
Is that true? Where did you hear that?
  #95  
Old 01-11-2013, 05:35 PM
zoid zoid is offline
Charter Member
 
Join Date: Sep 2001
Location: Chicago Il
Posts: 9,451
Quote:
Originally Posted by bup View Post
Is that true? Where did you hear that?
I'm guessing that people who eat Count Chocula tend to be kids and people who eat oatmeal tend to be much older.
  #96  
Old 01-11-2013, 06:27 PM
njtt njtt is offline
Guest
 
Join Date: Jul 2004
Location: U.K.
Posts: 12,068
Quote:
Originally Posted by Chronos View Post
Depends on how you want to slice it. If you start by assume that a culture typically has different names for boys and girls, then you could say that a child being a girl causes that child to have a culturally-female name. If you further assume from the start that we're working with the set of names that modern American culture considers female, then you could say that a child being a girl causes her to have one of the names from that set.
So are you saying that something that is a mere fortuitous correlation can turn something into a cause?
  #97  
Old 01-11-2013, 06:31 PM
Eurograff Eurograff is offline
Guest
 
Join Date: Mar 2003
Location: Inland Finland
Posts: 240
Quote:
Originally Posted by Leo Bloom View Post
I'm confused. You are fleshing out the milk/heroin false syllogism, or saying that there might be situations in which a true-quick assumption of the falseness is actually incorrect ("a problem with [this] web comic-type example")?

Sorry to be dense.
No one has answered to you, so I'll try. As Kimmy_Gibbler explained using mathematical notation, the point is simply that this kind of example for correlation without causation is not a suitable example, since there is no correlation at all. As Baracus explains in post #90, these "examples" are based more on a confirmation bias. Since every kid drinks milk, both the criminals and non-criminals will be separate subsets of the ex-milk drinker population, and no meaningful correlations can be established. So using this kind of examples to show how correlation doesn't prove causation doesn't work because they lack even the correlation part already.
  #98  
Old 01-11-2013, 07:51 PM
Andy L Andy L is offline
Member
 
Join Date: Oct 2000
Posts: 4,744
Quote:
Originally Posted by njtt View Post
Well, OK, it is probably true that being female causes one to tend to be shorter than average, but does being female tend to cause one to have a name ending in a vowel? It seems to me that that might be just a mere coincidental correlation. Female names (in our culture) just happen to often end in vowels, but there is nothing about lacking a Y chromosome that makes that so.
I was thinking that being born female in the context of American culture causes a person to be likely to be given a name that ends with a vowel sound (much as being born in France causes a person to grow up speaking French). I wouldn't call either fact coincidental, but I agree they are both arbitrary facts, and are not caused in the same way that lacking a Y chromosome causes to female features, or having two blue-eyed parents causes one to have blue eyes.
  #99  
Old 01-11-2013, 09:52 PM
Loach Loach is offline
The Central Scrutinizer
Moderator
 
Join Date: Apr 2003
Location: Pork Roll/Taylor Ham
Posts: 23,183
I'm surprised that no one has brought up the video game/violence political argument.
  #100  
Old 01-14-2013, 10:47 AM
CC CC is offline
Charter Member
 
Join Date: Apr 2000
Location: not elsewhere
Posts: 4,291
Here's a nice one from today's paper: "College students who watch reality television beauty shows are at least twice as likely as nonviewers to use tanning lamps or tan outdoors for hours at a time, according to a U.S. study."

Watching reality beauty shows does not cause students to use tanning lamps, nor does using tanning lamps cause students to watch beauty shows. There is, however, reason to believe that certain students are more likely to do both. That's a case where a third factor is the common link, and possible causative factor, to the observed behaviors.
Reply

Bookmarks

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is Off
HTML code is Off

Forum Jump


All times are GMT -5. The time now is 06:16 AM.

Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2017, vBulletin Solutions, Inc.

Send questions for Cecil Adams to: cecil@chicagoreader.com

Send comments about this website to: webmaster@straightdope.com

Terms of Use / Privacy Policy

Advertise on the Straight Dope!
(Your direct line to thousands of the smartest, hippest people on the planet, plus a few total dipsticks.)

Publishers - interested in subscribing to the Straight Dope?
Write to: sdsubscriptions@chicagoreader.com.

Copyright 2017 Sun-Times Media, LLC.

 
Copyright © 2017