Wednesday, December 12, 2012

Correlation and Causation

Gullible and uneducated America is always falling victim to misleading headlines that suggest that correlation implies causation.  Two synchronized variables definitely do not imply any sort of relationship between the two.

A recent study out of Australia had the headline "Earlier Bedtimes Keep Children Slimmer."  This study showed that children who go to bed early and wake up early are less likely to be obese. Could this be true? How would you explain this?  You could quickly jump to the conclusion that the reason many kids are obese is because of the amount of sleep they're getting.  Let's examine this study a little more.

This study recorded the sleeping habits of 2,200 Australian children, ranging from ages of 9 to 16, and compared their weights and uses of free time for a little less than a week.  Each kid got roughly the same amount of sleep but the kids who went to bed late and got up late were 1.5 times more likely to become obese than the kids who went to bed early. 

Now let's consideration some possible lurking variables in the study.  What could a kid between the ages of 9 to 16 be doing late in the evening/night?  Your average kid probably won't be playing sports or doing any kind of physical activity after dinner.  So what else is there to do?  We're in the age of online gaming and social networks so one could assume that a kid could be engaging in these sedentary activities late at night.  

Hopefully I've got you thinking a little bit and you could probably think of other lurking variables that could contribute to the obesity in children that were in the study.  My conclusion: early bedtimes do not imply causation of obesity in children; rather, it is the sedentary activity late at night that causes the obesity.  

Another poor assumption about correlation and causality is polio a century ago.  It was a horrible mystery that was claiming a lot of lives, especially in children.  At the time, there was a strong line of research that suggested that ice cream was causing polio in children.  The reason for this was because polio spiked in the summer time, for reasons unknown, and ice cream sales also spiked during the summer.  Researchers saw that when more ice cream was being consumed, more kids were getting polio.  This started  an ice cream persecution trying to stop polio.  This may sound ridiculous but you see it all the time now when people are against or trying to build up something that they are sure is connected to something else but turns out it isn't. 

Last but not least; rain and umbrellas are almost always found together (they are correlated), does that mean that the umbrellas are causing rain (causation)?  No, umbrellas do not cause rain. Quite a discovery!

Hook 'em

No comments:

Post a Comment