1 require(Hmisc)
   2 
   3 x <- spss.get("session3/ZA5240_v2-0-0.sav")
   4 ## y <- stata.get("data/ZA5240_v2-0-0.dta")
   5 
   6 ## the column V417 contains the net income,
   7 ## calculate the mean using the mean() function!
   8 ## What is the problem?
   9 
  10 
  11 nrow(x)
  12 ncol(x)
  13 
  14 mean(x$V417,na.rm = T)
  15 
  16 mean(x$V417,trim=0.05,na.rm = T)
  17 
  18 
  19 ## summarize the net income using summary(), quantiles() and fivenum()!
  20 
  21 summary(x$V417)
  22 summary(x$V81)
  23 
  24 quantile(x$V417,na.rm = T,probs = c(0,0.1,0.2,0.3,1))
  25 
  26 quantile(x$V417,na.rm = T,probs = seq(0,1,by=0.05))
  27 
  28 fivenum(x$V417)
  29 
  30 
  31 ## make a boxplot by using the following syntax!
  32 require(ggplot2)
  33 ggplot(x, aes(x=V86, y=V417)) +
  34     geom_boxplot()
  35 
  36 
  37 ggplot(x, aes(x=V86, y=V417, fill=V81)) +
  38     geom_boxplot()
  39 
  40 
  41 ######################################################################
  42 ####################### T-Test #######################################
  43 ######################################################################
  44 
  45 
  46 ## one sample
  47 set.seed(1)
  48 x <- rnorm(12)
  49 
  50 t.test(x,mu=0)
  51 
  52 t.test(x,mu=1)
  53 
  54 
  55 ## two sample Welch or Satterthwaite test 
  56 set.seed(1)
  57 x <- rnorm(12)
  58 y <- rnorm(12)
  59 g <- sample(c("A","B"),12,replace = T)
  60 
  61 t.test(x, y)
  62 t.test(x ~ g)
  63 
  64 t.test(x, y, var.equal = T)
  65 
  66 
  67 ######################################################################
  68 ####################### Exercises  ###################################
  69 ######################################################################
  70 
  71 ## do a t-test of income (V417): male against female (V81)!
  72 
  73 t.test(x$V417 ~ x$V81)
  74 
  75 ## and compare the bmi (V279) in smokers and non-smokers (V272)
  76 ## and between the people with high and normal blood pressure (V242)
  77 
  78 t.test(x$V279 ~ x$V272)
  79 t.test(x$V279 ~ x$V242)
  80 
  81 
  82 summary(x$V242)
  83 
  84 
  85 ## bmi by smokers/ non smokers
  86 ggplot(x, aes(x=V272, y=V279, fill=V81)) +
  87     geom_boxplot() +
  88     facet_wrap( ~ V86)
  89 
  90 
  91 
  92 
  93 summary(x$V272)
  94 
  95 ## bmi by high/normal blood pressure
  96 
  97 
  98 
  99 
 100 
 101 ## Alter
 102 summary(x$V84)
 103