1. Correlation does not imply causation. A common media blunder is to infer a causeand-

effect relationship between two variables simply because there is a strong correlation between them. Don’t fall into this trap! A strong correlation implies only that the two variables tend to vary together in a predictable way, but there are many possible explanations for why this is occurring besides one variable causing changes in the other.

For example, the article “Ban Cell Phones? You May as Well Ban Talking Instead”

(USA Today, April 27, 2000) gave data that showed a strong negative correlation between the number of cell phone subscribers and traffic fatality rates. During the years from 1985 to 1998, the number of cell phone subscribers increased from 200,000 to 60,800,000, and the number of traffic deaths per 100 million miles traveled decreased from 2.5 to 1.6 over the same period. However, based on this correlation alone, the conclusion that cell phone use improves road safety is not reasonable!

Similarly, the Calgary Herald (April 16, 2002) reported that heavy and moderate drinkers earn more than light drinkers or those who do not drink. Based on the correlation between number of drinks consumed and income, the author of the study concluded that moderate drinking “causes” higher earnings. This is obviously a misleading statement, but at least the article goes on to state that “there are many possible reasons for the correlation. It could be because better-off men simply choose to buy more alcohol. Or it might have something to do with stress: Maybe those with stressful jobs tend to earn more after controlling for age, occupation, etc., and maybe they also drink more in order to deal with the stress.”