Here’s a news piece :
I know what you’re thinking : It doesn’t make sense! The article merely says that whenever the Himalayas have received heavy snowfall, it has been followed by a drought the coming summer. ToI instead prefers to say that heavy snowfall causes drought.
Which brings me to another one of my pet peeves, confusing correlation with causation.
Correlation is — whenever event A happens, it is highly likely that event B happens.
Causation is — Event A happens because of event B.
At first glance, they might seem similar, but they are not. I was talking about this to a friend, who is far more sensitive to misplaced causations and correlations. He quoted a nice example to illustrate the difference :
Let’s say — since 2007, petrol consumption has increased and iPod sales have increased. Let’s assume the reason is just a boom in the economy. But someone might make a case that because of a new iPod, people like driving since they can enjoy all their songs in the car. They consequently drive more and longer, so petrol consumption has increased.
There could be many theories. The fact is — unless there is a strong and obvious cause-and-effect relationship, it is mere correlation.
*****
XKCD has a nice take on it, and illustrates it better than I did :
Link via Raghu.
7 comments
Comments feed for this article
April 23, 2010 at 12:10 am
maverickshree
Liked petrol consumption and iPod sales corelation
You are correct there can be many more such theories
April 24, 2010 at 8:08 am
Sriram
Very true. I remember a similar example from my Maths textbook in school. In the 19th century (I think), the production of pig iron in the UK and the child birth rate in the US had a correlation coefficient of -0.98. In simpler terms, there cannot be a better correlation between any two quantities. By no stretch of the imagination, can the increased production of pig iron in one continent bring down the child birth rate in another. And yet, there are people who peddle such theories in other contexts.
April 24, 2010 at 12:02 pm
Deepak Iyer
@Sriram : Hmm .. well .. say the increased production of pig iron led to a boom in the industry and increased job opportunity. That would lead to many from the U.S. migrating for jobs, which could keep them away from home. Assuming faithful wives, that would lead to lower child birth rate.
Phew.
(of course, it’s silly and incorrect)
April 25, 2010 at 6:29 am
The Perfect Boob
haha. love XKCD!
April 27, 2010 at 3:01 pm
dev
have you read Freakonomics… Correlation and causation is pretty much what the entire book is about. decent read.
April 27, 2010 at 8:57 pm
Deepak Iyer
@dev : I haven’t, but I wouldn’t want to read an entire book on such stats [:)]
April 28, 2010 at 2:15 pm
dev
Deepak,
Instead of using variables , the book uses stats from society to find surprising correlation. For example, crime in LA went down in the 1980. Some attributed that to increase in police force but this was disproved because other cities with increased police force did not notice drop in crime. Some attributed it to better economy since people have jobs but this was disproved because earlier periods in history where economy took off did not lead to lower crime. But if you consider that abortion became legal in 1960, and than most criminals are born to single mothers + mothers in areas where crime was highest… everything is solved. So the correlation and causation does form the skeleton which is then topped by many example as above.