Beruflich Dokumente
Kultur Dokumente
250
Approaches to Multivariate Inference
251
Review of Univariate Hypothesis Testing
Is a given value 0 a plausible value for the population mean
?
We formulate the problem as a hypothesis testing problem.
The competing hypotheses are
H0 : = 0 and Ha : 6= 0.
252
Univariate Hypothesis Testing (contd)
253
Univariate Hypothesis Testing (contd)
If we fail to reject H0, we conclude that 0 is close (in units
of standard deviations of X) to X,
and thus is a plausible
value for .
The set of plausible values for is the set of all values that
lie in the 100(1 )% confidence interval for :
s s
tn1(/2) 0 x
x + tn1(/2) .
n n
254
Hotellings T 2 Statistic
255
Hotellings T 2 Statistic (contd)
If the observed T 2 value is large we reject H0 : = 0.
256
Hotellings T 2 Statistic (contd)
As we noted earlier,
1 1
2
T = (X 0 ) 0 S 0) = n(X
(X 0)0S 1(X
0 )
n
has an approximate central chi-square distribution with p df
when 0 is correct, for large n, or when is known, in which
case the distribution is exact when we have normality.
Note that
(n 1)p
Fp,np() > 2
p ()
(n p)
but these quantities are nearly equal for large values of n p.
257
Example 5.2: Female Sweat Data
258
Example 5.2: Sweat Data (contd)
259
Example 5.2: Sweat Data (contd)
260
The Female Sweat Data: R code sweat.R
261
Invariance property of Hotellings T 2
The T 2 statistic is invariant to changes in units of
measurements of the form
Yp1 = CppXp1 + dp1,
with C non-singular. An example of such a transformation
is the conversion of temperature measurements from
Fahrenheit to Celsius.
262
Invariance property of Hotellings T 2 (contd)
y Y,0)0Sy1(
Ty2 = n( y Y,0)
x 0))0(CSC 0)1(C(
= n(C( x 0))
x 0)0S 1(
= n( x 0).
263
Likelihood Ratio Test and Hotellings T 2
Compare the maximum value of the multivariate normal
likelihood function under no restrictions against the
restricted maximized value with the mean vector held at 0.
The hypothesized value 0 will be plausible if it produces a
likelihood value almost as large as the unrestricted maximum.
265
Derivation of Likelihood Ratio Test
1X
0|n/2 exp {
| 1(xi 0)]}
(xi 0)0
2 i 0
1X
=
|0| n/2 exp{ 1(x0)}
tr(xi 0)0
2 i 0
n/2 1 1
(xi 0)(xi 0)0}
X
=
|0| exp{ tr0
2 i
1
= 0|n/2 exp{ tr
| 1 0n}
2 0
np
= 0|n/2 exp{ }.
|
2
266
Derivation of Likelihood Ratio Test
1X
n/2 exp {
|| )0
(xi x 1(xi x)]}
2 i
n/2 1X
=
|| exp{ tr(xi x)0
1(xx)}
2 i
n/2 1 1 )0}
X
=
|| exp{ tr (xi x
)(xi x
2 i
1
= n/2 exp{ tr
|| 1n}
2
np
= n/2 exp{ }.
||
2
267
Derivation of Likelihood Ratio Test
| 0|n/2 exp{ np }
= 2
n/2 exp{ np }
|| 2
| 0|n/2
=
n/2
||
n/2
||
=
| 0|n/2
!n/2
||
= .
|0|
268
Relationship between and T 2
269
Relationship between and T 2
270
Union-Intersection Derivation of T 2
272