# YouthRisk 2007 data from Division of Adolescent and School Health, National Center for HIV/AIDS, Viral Hepatitis, STD, and TB Prevention

install.packages("Stat2Data") # only if you have not done so already or the the package has not been installed as part of R installation
libarary(Stat2Data)
attach(YouthRisk)
head(YouthRisk)
number16and17yearolds<-length(which(age4>16))
ages<-levels(as.factor(age4))
tapply(age4,smoke,sum,na.rm=TRUE)
#     0      1 
# 96400 112756 

tapply(female,smoke,sum,na.rm=TRUE)
#   0    1 
#1857 4642 

YouthsByAge<-apply(agetable,1,sum)
#   14   15   16   17   18 
#1183 2883 3287 3262 1970 

agetable<-table(age4,smoke)
agetable
#    smoke
#age4    0    1
#  14  739  468
#  15 1595 1397
#  16 1579 1798
#  17 1349 2001
#  18  774 1248

sextable<-table(age4,female)
sextable
#age4    0    1
#  14  903  280
#  15 1709 1174
#  16 1587 1700
#  17 1190 2072
#  18  558 1412

smoketable<-table(female,smoke)
smoketable
#      smoke
#female    0    1
#     0 3871 1979
#     1 1857 4642

femalesOnly<-subset(YouthRisk,female==1)
femaleSmokersByAge<-table(femalesOnly$age4,femalesOnly$smoke)
femaleSmokersByAge
    
#        0    1
#  14   84  188
#  15  315  823
#  16  493 1173
#  17  557 1468
#  18  401  970



ProportionFemaleSmokeryByAge<-femaleSmokersByAge[,2]/sextable[,2]
ProportionFemaleSmokeryByAge
#       14        15        16        17        18 
#0.6714286 0.7010221 0.6900000 0.7084942 0.6869688