38 



decline of nicotine and tar from the 1950's to the 1990's; for both 

 tar and nicotine there has been a 69 percent reduction. 



This graph also shows that nicotine levels are a function of the 

 tar levels in tobacco. When tar levels are set, nicotine levels follow. 

 As manufacturers have reduced tar levels over the years, nicotine 

 levels have concurrently fallen. That is why tar and nicotine levels 

 have been reduced by exactly the same amount, 69 percent. 



Therefore, this graph demonstrates correlation between the re- 

 duction of tar and nicotine levels. 



As I am sure you know, correlation is a very common statistical 

 measure of the dependence between two factors. For example. 

 Scholastic Aptitude Test scores are strongly correlated with first- 

 year grades in college. In other words, a high SAT test score is an 

 excellent predictor of success in college. 



In statistics, correlation coefficients of 0.50 are evidence of sig- 

 nificantly strong relationships. In this graph, the correlation is very 

 close to one, meaning that the relationship between tar and nico- 

 tine is very, very strong. 



What does this mean? First, that nicotine levels follow tar levels; 

 second, that no nicotine manipulation has occurred in the manufac- 

 ture and process. Why? Because if nicotine was being added back 

 into cigarettes the data would have shown one of three things: Nic- 

 otine levels would have remained relatively constant over the 40- 

 year period while tar levels would have declined significantly; sec- 

 ond, nicotine levels actually have declined very little compared to 

 the 69 percent reduction in tar levels; finally, nicotine levels could 

 have increased while tar levels declined. 



In all these cases, the correlation between the levels of nicotine 

 and tar would be very weak or even negative. Instead, this data 

 shows an almost perfect positive correlation. In other words, nico- 

 tine is not being added. 



Chart three, what about the tar and nicotine levels in 1994? This 

 table presents the correlation analysis of the latest tar and nicotine 

 results to be reported by the Federal Trade Commission. The cor- 

 relation coefficient of the amount of tar and nicotine in domestic 

 brands is 0.975. This is essentially a perfect correlation between 

 the levels of nicotine and tar. Again, this is very strong statistical 

 evidence that nicotine is not being manipulated. 



[The charts follow:] 



