59 



of their variances: w(i,j) « N(i,j)/ SDR(i,j)'2. (These are 

 not quite the same weights as in analysis 1 because in that 

 analysis, the final weights are scaled to add up to 1 over 

 the 10 years within each tar category; in this analysis, the 

 weights are scaled so that the sua of all 3 weights over 

 year and categories is 1.) 



Findings were that nicotine/tar ratio was increasing with 

 time in all three tar categories to a statistically significant 

 extent. The three tar categories also had statistically 

 significantly different mean nicotine/tar ratios when adjusted 

 for year. 



P-values for H,: There is no change in the ratio over time, 

 are tabled as follows: 



P-values 



