Thursday, October 29, 2009

Correlation and Causality 3: Smoking

I personally believe that smoking is a noxious, loathsome habit, and that anyone who smokes is both stupid and rude.  But that does not mean I will tolerate bad data or journalism about smoking.  This chart and the accompanying text are extremely misleading, for several reasons.

The WHO data actually says that there are a few countries where over 20% of people over 30 years old die of smoking-related illnesses.  But that got turned into "Nearly one in five deaths in rich countries is caused by smoking..."  This is wrong in three ways.  First, a few countries got turned into 'the rich world'.  Look at the chart and you will see plenty of rich countries with lower listed mortality from smoking.  Second, the article omitted the important bit about 'people over 30'.  And third, dying from a smoking-related illness is very different than getting killed by cigarettes.

The research defines a 'smoking-related illness' as anything that smokers are more likely to have than non-smokers.  If someone who has ever smoked dies from one of those diseases, that person is recorded in the statistics as 'dying from a smoking-related disease'.  But this is wrong, because it confuses correlation with causation.

People who smoke generally have a lot of other bad habits.  Smokers tend to be more obese and less likely to exercise.  They generally have worse nutrition and more alcoholism.  The bad effects from all of these other things are getting mixed in with the smoking deaths because people are not doing the statistics right.

For example, delerium tremens would be classified in these statistics as a 'smoking-related disease'.  A random smoker is more likely to have DT's than a random non-smoker, because smokers are more likely to be alcoholic.  So if someone died from DT's, these numbers would add him to the list of people 'killed by cigarettes'.

If you run the numbers properly, you find that smoking actually kills about 150,000 people in the USA every year, instead of the 400,000 figure that keeps making the headlines.  And most of those people are very old when they die, meaning that smoking actually just took a few years off their lives.

So the government really should focus more on fighting crime than fighting smoking.  Yes, secondhand smoke is a problem, but you can fix it with the appropriate cigarette tax.  There is no need to keep lecturing people, who usually know the risks and are willing to cut their lifespan a few years to get whatever benefit they get from cigarettes.  The proper role of the state is to prevent innocent people from getting hurt and killed by the actions of others.  Cigarettes should be a low-priority concern.

No comments: