7.2 The Chisquare distribution  7. Hypothesis testing  7. Hypothesis testing

7.1 Purpose

Chapters 4 and 5 discussed methods for determining the best-fit values of parameters, but did not consider if those values provided a good fit to the data. However, it is equally necessary to consider if the selected representation of the data is adequate or consistent, and if the number of parameters is sufficient to represent the observations. The statistical basis for addressing these questions will only be treated in cursory fashion here. This is a branch of mathematics that has received extensive application in meteorology, especially in connection with weather modification experiments; see, for example, Dennis (19XX).

Here the discussion is limited to the chisquare test, the F test, and a short warning concerning likelihood ratios. The chisquare test and the Student t test (discussed in Chapter 2) are the tests most often needed in analyses of experimental data. The F test is included here because, among other applications, it is useful when considering how many parameters must be included in multiple-parameter fits. However, it is expected that these sections are only short reminders of familiar material. If not, the reader will benefit from study of a more complete text, such as one of those included in the bibliography.

   
 7.2 The Chisquare distribution  7. Hypothesis testing  7. Hypothesis testing