Ch5-Inference About Mean

CHAPTER 5: Inferences about a
Mean Vector
5.1 Introduction
We shall now use the concepts and results set forth in
Chapters 1 through 4 to develop techniques for
analyzing data. A large part of any analysis is concerned
with inference—that is, reaching valid conclusions
concerning a population on the basis of information from a
sample.
At this point, we shall concentrate on inferences about a
population mean vector and its component parts.
5.2 The Plausibility of μ0 as a Value for a
Normal Population Mean
Let us start by recalling the univariate theory for determining whether a specific
value μ0 is a plausible value for the population mean μ.
Hypotheses:
H0 is the null hypothesis and H1 is the (two-sided) alternative hypothesis.
This test statistic has a student's t-distribution with n-1 degrees of freedom (d.f.).
 Consider now the problem of determining whether a given p×1 vector μ0
is a plausible value for the mean of a multivariate normal distribution. We
shall proceed by analogy to the univariate development just presented.
where
The statistic T2 is called Hotelling's T2 in honor of Harold Hotelling, a

pioneer in multivariate analysis, who first obtained its sampling
distribution.
At the α level of significance, we reject H0 in favor of H1, if the
observed
Example 5.1 : Let the data matrix for a random sample of size n = 3
from a bivariate normal population be  6 9
X  10 6 
 8 3
Evaluate the observed T2 for μ0' =[9, 5]. What is the sampling distribution of T2 in
this case?
 One feature of the T2-statistic is that it is invariant (unchanged)
under changes in the units of measurements for X of the form
5.3 Hotelling‘s T2 and Likelihood Ratio Tests （※）
 There is a general principle for constructing test procedures called
the likelihood ratio method, and the T2 -statistic can be derived as
the likelihood ratio test of H0:μ=μ0.
 We know the maximum of the multivariate normal likelihood as μ
and Σ are varied over their possible values is given by
Where
Under the hypothesis H0:μ=μ0, the normal likelihood specializes to

The equivalent statistic called Wilks' lambda.
We reject H0 if
where cα is the lower (l00α)th percentile of the distribution of ʌ. (Note that

the likelihood ratio test statistic is a power of the ratio of generalized
variances.) Fortunately, because of the following relation between T2 and ʌ,
we do not need the distribution of the latter to carry out the test.
Result 5.1.
 Let X1, X2,..., Xn be a random sample from an Np(μ,Σ) population.
Solving for T2,
General Likelihood Ratio Method
5.4 Confidence Regions and Simultaneous
Comparisons of Component Means
Beginning at the center X , the axes of the confidence ellipsoid are

Example 5.3 (Constructing a confidence ellipse for μ)
 For the n=42 pairs of transformed observations, we find that
Example 5.3 (Constructing a confidence ellipse for μ)
Simultaneous Confidence Statements
Intervals are wider than intervals of (5-21) for a specific choice of a.
 Example 5.5 (Constructing simultaneous confidence intervals and
ellipses) The scores obtained by n= 87 college students on the
College Level Examination Program (CLEP) subtest X1 and the
College Qualification Test (CQT) subtests X2 and X3 are given in
Table 5.2 on page 228 for X1 = social science and history, X2 =verbal,
and X3 = science. These data give
 The individual t-intervals:
The Bonferroni method of multiple comparisons:

Often, attention is restricted to a small number of individual confidence statements. In
these situations it is possible to do better than the simultaneous intervals of Result 5.3.
If the number m of specified component means μi,or linear combinations a’u is small,
simultaneous confidence intervals can be developed that are shorter (more precise)
than the simultaneous T2-intervals The alternative method for multiple comparisons is
called the Bonferroni method because it is developed from a probability inequality
carrying that name.
5.5 Large Sample Inferences about a Population
Mean Vector※
 When the sample size is large, tests of hypotheses and confidence
regions for μ can be constructed without the assumption of a normal
population.

Ch5-Inference About Mean

Uploaded by

Copyright:

Available Formats

Ch5-Inference About Mean

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ch5-Inference About Mean

Uploaded by

Copyright:

Available Formats

CHAPTER 5: Inferences about a

The statistic T2 is called Hotelling's T2 in honor of Harold Hotelling, a

Under the hypothesis H0:μ=μ0, the normal likelihood specializes to

where cα is the lower (l00α)th percentile of the distribution of ʌ. (Note that

Beginning at the center X , the axes of the confidence ellipsoid are

The Bonferroni method of multiple comparisons:

You might also like