According to the American Lung Association, smoking contributes to approximately 80% of lung cancer deaths among women. So you would expect the relationship between cigarette consumption and lung cancer death rates to be unmistakable. However, as the scatter plot on the left shows, the connection between the two variables is very weak. The correlation coefficient for the plot is a negligible .22 — revealing an astonishingly tenuous link.
Per capita cigarette consumption:
()
Lung/bronchus cancer deaths per 100k women:
()
Per capita cigarette consumption:
()
Lung/bronchus cancer deaths per 100k women:
()
The dots appear chronologically on both chart, but the cigarette consumption data along the x-axis spans from 1930 to 2006 on the left and from 1930 to 1981 on the right. The reason for this is because the second chart includes a 30-year delay between cigarette consumption and the lung cancer death rate. This is also the reason why the correlation on the left is so weak.
When I first plotted these two data sets, I didn't realize the importance of a latency period, or the time it takes between when someone starts smoking and when they might first develop lung cancer. Researcher William Weiss found that a 30-year gap exists between when cigarette smoking increased in the U.S. and the subsequent increase in lung cancer deaths. When I adjust the original chart for this latency period, the relationship grows much stronger. The correlation coefficient increases from .22 to .93.
This 30-year lag also exists in other countries. In Australia, the Sydney Morning Herald noted that in 2014 lung cancer deaths continued to increase among women, even as it dropped among men. The reason for this is that the increase in the number of female smokers peaked decades after it did for men.