文档介绍：244 Resampling: The New Statistics
CHAPTER The Statistics of
Hypothesis-Testing with
17 Counted Data, Part 2
Here’s the bad-news-good-news message again: The bad news
is that the subject of inferential statistics is extremely difficult—
not because it plex but rather because it is subtle. The
cause of the difficulty is that the world around us is difficult
to understand, and spoon-fed mathematical simplifications
which you manipulate mechanically simply mislead you into
thinking you understand that about which you have not got a
clue.
The good news is that you—and that means you, even if you
say you are “no good at math”—can understand these prob-
lems with a layperson’s hard thinking, even if you have no
mathematical background beyond arithmetic and you think
that you have no mathematical capability. That’s because the
difficulty lies in such matters as pin-pointing the right ques-
tion, and understanding how to interpret your results.
The problems in the previous chapter were tough enough. But
this chapter considers problems with plications,
such as when there are more than two groups, or -
parisons for the same units of observation.
Comparisons among more than two samples of counted data
Example 17-1: Do Any of Four Treatments Affect the Sex
Ratio in Fruit Flies? (When the Benchmark Universe Propor-
tion is Known, Is the Proportion of the Binomial Population
Affected by Any of the Treatments?) (Program “4treat”)
Suppose that, instead of experimenting with just one type of
radiation treatment on the flies (as in Example 15-1), you try
four different treatments, which we shall label A, B, C, and D.
Treatment A produces fourteen males and six females, but treat-
ments B, C, and D produce ten, eleven, and ten males, respec-
tively. It is immediately obvious that there is no reason to think
that treatment B, C, or D affects the sex ratio. But what about
treatment A?
Chapter 17—The Statistics of Hyp