CHAPTER CONTENTS
        CHAPTER CONTENTS
        12.1 Introduction .................................................................................................. 590
        12.2 Nonparametric Confidence Interval ................................................................. 592
        12.3 Nonparametric Hypothesis Tests for One Sample ............................................. 597
        12.4 Nonparametric Hypothesis Tests for Two Independent Samples ........................ 609
        12.5 Nonparametric Hypothesis Tests for >  2 Samples ........................................ 618
        12.6 Chapter Summary .......................................................................................... 627
        12.7 Computer Examples ....................................................................................... 627
        Projects for Chapter 12 .......................................................................................... 635
                       Objective of this chapter ：
      To study tests that do not require distributional assumptions about the population such as the normality.
                   N(mean, variance), Uniform(a, b, 1/(b-a))
                        Jacob Wolfowitz
     It is in this paper by Wolfowitz in 1942 that the term nonparametric appears for the first time.
        Wolfowitz made important contributions to Information theory.
                     12.1 Introduction 
    Sometimes we may be required to make inferences about models that are difficult to parameterize,
     or we may have data in a form that make, say, the normal theory tests unsuitable.
      to parameterize =  to identify a classical probability distribution that will characterize the data’s behavior. 
                  Nonparametric methods are appropriate to estimation or hypothesis testing problems 
                  when the population distributions could only be specified in general terms. 
                  The conditions may be specified as being continuous, symmetric, or identical, differing only in median 
                  or mean.
              Nonparametric methods:
                Classical ：     based on ordering, ranking, and permutations                                 Ch 12
                Modern：         based on resampling method                                                   Ch 13
  Nonparametric methods:
   The distributions need not belong to specific families such as normal or 
    gamma. 
   Are distribution-free  methods
   Depend on a minimum number of assumptions (So, the chance of their 
    being improperly used is relatively small.)
   Involve ranking data values and developing testing methods based on the 
    ranks.
   When the assumptions of the parametric tests can be verified as true, 
    parametric tests are generally more powerful than nonparametric tests
   Some information is lost because the actual values are not used.
   Less powerful than their parametric counterparts when parametric tests 
    can be used