Marginal and Conditional distributions
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Marginal and Conditional distributions
1
2
Let q
p q
xx
x
Theorem: (Marginal distributions for the Multivariate Normal distribution)
11 12
12 22
have p-variate Normal distribution
with mean vector1
2
q
p q
and Covariance matrix
Then the marginal distribution of is qi-variate Normal distribution (q1 = q, q2 = p - q)
iiwith mean vector i
and Covariance matrix
ix
1
2
Let q
p q
xx
x
Theorem: (Conditional distributions for the Multivariate Normal distribution)
11 12
12 22
have p-variate Normal distribution
with mean vector1
2
q
p q
and Covariance matrix
Then the conditional distribution of given is qi-variate Normal distribution
1- ii j ii ij jj ij
with mean vector 1 = i j i ij jj j jx
and Covariance matrix
ix
jx
12 1 22 12 11 12The matrix
is called the matrix of partial variances and covariances.
th
2 1The , element of the matrix i j
1,2....ij q
is called the partial covariance (variance if i = j) between xi and xj given x1, … , xq.
1,2....1,2....
1,2.... 1,2....
ij qij q
ii q jj q
is called the partial correlation between xi and xj given x1, … , xq.
112 11the matrix
is called the matrix of regression coefficients for predicting xq+1, xq+2, … , xp from x1, … , xq.
12 1 1 2 12 11 1 where x
Mean vector of xq+1, xq+2, … , xp given x1, … , xqis:
Example:
10
15 and
6
14
Suppose that
1
2
3
4
x
xx
x
x
Is 4-variate normal with
4 2 4 2
2 17 6 5
4 6 14 6
2 5 6 7
1
10 and
15
The marginal distribution of1
12
xx
x
is bivariate normal with
11
4 2
2 17
1
10
15 and
6
The marginal distribution of1
1 2
3
x
x x
x
is trivariate normal with
11
4 2 4
2 17 6
4 6 14
Find the conditional distribution of
11
2
15
5
xx
x
given
11
4 2
2 17
1 2
10 6 and
15 14
Now
and
32
4
xx
x
22
14 6
6 7
12
4 2
6 5
12 1 22 12 11 12
114 6 4 6 4 2 4 2
6 7 2 5 2 17 6 5
9 3
3 5
112 11
The matrix of regression coefficients for predicting x3, x4 from x1, x2.
14 6 4 2
2 5 2 17
0.875 .250
0.375 .250
1 22 1 1
1 2
0.875 0.250 6.5
0.375 0.250 6.5
x xx
x x
6 0.875 0.250 10
14 0.375 0.250 15
6.5
6.5
12 12 11 1
2 1
0.875 15 0.250 5 6.5 7.875
0.375 15 0.250 5 6.5 13.375
Thus the conditional distribution of
11
2
15
5
xx
x
given
is bivariate Normal with mean vector
And partial covariance matrix
32
4
xx
x
2 1
7.875
13.375
2 1
9 3
3 5
Using SPSS
Note: The use of another statistical package such as Minitab is similar to using SPSS
The first step is to input the data.
The data is usually contained in some type of file.
1. Text files
2. Excel files
3. Other types of files
After starting the SSPS program the following dialogue box appears:
If you select Opening an existing file and press OK the following dialogue box appears
Once you selected the file and its type
The following dialogue box appears:
If the variable names are in the file ask it to read the names. If you do not specify the Range the program will identify the Range:
Once you “click OK”, two windows will appear
A window containing the output
The other containing the data:
To perform any statistical Analysis select the Analyze menu:
To compute correlations select Correlate then BivariateTo compute partial correlations select Correlate then Partial
for Bivariate correlation the following dialogue appears
Correlations
1.000 .080 .253** .372** -.069 .009 .210**
. .281 .001 .000 .357 .899 .004
183 183 183 183 183 183 183
.080 1.000 .481** -.007 -.013 .147* .106
.281 . .000 .930 .863 .047 .153
183 183 183 183 183 183 183
.253** .481** 1.000 .131 -.235** .072 .291**
.001 .000 . .078 .001 .330 .000
183 183 183 183 183 183 183
.372** -.007 .131 1.000 .075 .269** .294**
.000 .930 .078 . .313 .000 .000
183 183 183 183 183 183 183
-.069 -.013 -.235** .075 1.000 .454** .039
.357 .863 .001 .313 . .000 .603
183 183 183 183 183 183 183
.009 .147* .072 .269** .454** 1.000 .178*
.899 .047 .330 .000 .000 . .016
183 183 183 183 183 183 183
.210** .106 .291** .294** .039 .178* 1.000
.004 .153 .000 .000 .603 .016 .
183 183 183 183 183 183 183
Pearson Correlation
Sig. (2-tailed)
N
Pearson Correlation
Sig. (2-tailed)
N
Pearson Correlation
Sig. (2-tailed)
N
Pearson Correlation
Sig. (2-tailed)
N
Pearson Correlation
Sig. (2-tailed)
N
Pearson Correlation
Sig. (2-tailed)
N
Pearson Correlation
Sig. (2-tailed)
N
AGE
HT
WT
CHL
ALB
CA
UA
AGE HT WT CHL ALB CA UA
Correlation is significant at the 0.01 level (2-tailed).**.
Correlation is significant at the 0.05 level (2-tailed).*.
the output for Bivariate correlation:
for partial correlation the following dialogue appears
- - - P A R T I A L C O R R E L A T I O N C O E F F I C I E N T S - - -Controlling for.. AGE HT WT CHL ALB CA UACHL 1.0000 .1299 .2957 .2338 ( 0) ( 178) ( 178) ( 178) P= . P= .082 P= .000 P= .002ALB .1299 1.0000 .4778 .1226 ( 178) ( 0) ( 178) ( 178) P= .082 P= . P= .000 P= .101CA .2957 .4778 1.0000 .1737 ( 178) ( 178) ( 0) ( 178) P= .000 P= .000 P= . P= .020UA .2338 .1226 .1737 1.0000 ( 178) ( 178) ( 178) ( 0) P= .002 P= .101 P= .020 P= .(Coefficient / (D.F.) / 2-tailed Significance)" . " is printed if a coefficient cannot be computed
the output for partial correlation:
Correlations
1.000 .080 .253** .372** -.069 .009 .210**
. .281 .001 .000 .357 .899 .004
183 183 183 183 183 183 183
.080 1.000 .481** -.007 -.013 .147* .106
.281 . .000 .930 .863 .047 .153
183 183 183 183 183 183 183
.253** .481** 1.000 .131 -.235** .072 .291**
.001 .000 . .078 .001 .330 .000
183 183 183 183 183 183 183
.372** -.007 .131 1.000 .075 .269** .294**
.000 .930 .078 . .313 .000 .000
183 183 183 183 183 183 183
-.069 -.013 -.235** .075 1.000 .454** .039
.357 .863 .001 .313 . .000 .603
183 183 183 183 183 183 183
.009 .147* .072 .269** .454** 1.000 .178*
.899 .047 .330 .000 .000 . .016
183 183 183 183 183 183 183
.210** .106 .291** .294** .039 .178* 1.000
.004 .153 .000 .000 .603 .016 .
183 183 183 183 183 183 183
Pearson Correlation
Sig. (2-tailed)
N
Pearson Correlation
Sig. (2-tailed)
N
Pearson Correlation
Sig. (2-tailed)
N
Pearson Correlation
Sig. (2-tailed)
N
Pearson Correlation
Sig. (2-tailed)
N
Pearson Correlation
Sig. (2-tailed)
N
Pearson Correlation
Sig. (2-tailed)
N
AGE
HT
WT
CHL
ALB
CA
UA
AGE HT WT CHL ALB CA UA
Correlation is significant at the 0.01 level (2-tailed).**.
Correlation is significant at the 0.05 level (2-tailed).*.
Compare these with the bivariate correlation:
CHL ALB CA UACHL 1.0000 .1299 .2957 .2338 ALB .1299 1.0000 .4778 .1226 CA .2957 .4778 1.0000 .1737 UA .2338 .1226 .1737 1.0000
Partial Correlations
Bivariate Correlations
In the last example the bivariate and partial correlations were roughly in agreement.
This is not necessarily the case in all stuations
An Example:
The following data was collected on the following three variables:
1. Age
2. Calcium Intake in diet (CAI)
3. Bone Mass density (BMI)
The data
Age CAI BMI Age CAI BMI Age CAI BMI
25 75.2 147.7 45 62.5 239.8 65 66.8 298.925 83.6 166.7 45 84.6 257.5 65 53.5 280.625 112.2 254.9 45 107 317.2 65 64.9 287.225 99.8 193 45 82.3 280.3 65 63.8 302.225 93.1 199 45 69.9 232.8 65 52.6 263.325 97.7 202.6 45 100.6 270.5 65 58.8 296.225 103.7 231.7 45 74.1 228.6 65 61.4 294.225 101.8 199.7 45 60.2 231.8 65 59.6 294.325 99.6 182.4 45 94.6 252.5 65 62.9 250.425 94.9 202.8 45 80.7 254.9 65 52 26525 99.6 204.7 45 94.4 266.3 65 60.4 267.625 100.2 206.6 45 73.1 227.9 65 61.2 287.325 116.9 280 45 81.2 245 65 67.4 299.825 97.3 186.9 45 106.1 297.8 65 51.5 273.225 98.8 217.9 45 79.3 217.7 65 60.7 284.225 90.6 198.7 45 85.1 263.9 65 56.3 290.225 101.7 190.4 45 81.9 280.5 65 72.9 306.725 98.6 221.3 45 98.7 281 65 40.3 258.825 93 191.8 45 89.1 275.4 65 47.1 283.625 108.1 216.2 45 71.6 225.2 65 76.9 323.525 78.9 161.3 45 76.9 240.9 65 64.7 303.425 87.1 188.6 45 79.7 252.1 65 59.9 297.935 96 248 55 61.1 238.1 75 37.8 277.935 97 261 55 60.5 250.1 75 43 287.535 94.8 237.5 55 82 285.9 75 33.8 305.935 78.1 225.8 55 70.7 267.1 75 41.5 320.335 93.1 239.7 55 71.9 258.2 75 71.2 353.335 74.3 205.7 55 64.4 245.3 75 58 34535 100.1 255.7 55 66.3 284.9 75 53.4 325.535 95 244.3 55 64.3 281.7 75 37.5 268.535 77 202.3 55 59 249.9 75 41.6 312.135 94.5 231.6 55 83.8 318.1 75 50.9 282.835 108.6 288.2 55 68.9 274.4 75 57.5 346.935 92.4 221.2 55 77.8 266.6 75 51.8 32335 104.3 262 55 63.9 274.9 75 64.9 343.835 87 218.7 55 75.8 277.6 75 44 299.235 88.6 232.3 55 78.8 291.5 75 49.4 313.435 97 252.7 55 82.6 302.8 75 54.4 294.935 85 213.8 55 65.4 270.7 75 53.9 321.335 96.1 232.8 55 59.5 231.3 75 48.7 26235 111.2 288.6 55 54.4 239 75 51.4 325.735 83.9 230.3 55 56.6 247.8 75 49.2 345.535 100.1 248.2 55 55.6 219.9 75 36.5 260.935 91.2 258.6 55 64 249.2 75 61 333.8
Correlations
1.000 -.863** .800**
. .000 .000
132 132 132
-.863** 1.000 -.447**
.000 . .000
132 132 132
.800** -.447** 1.000
.000 .000 .
132 132 132
Pearson Correlation
Sig. (2-tailed)
N
Pearson Correlation
Sig. (2-tailed)
N
Pearson Correlation
Sig. (2-tailed)
N
AGE
CAI
BMI
AGE CAI BMI
Correlation is significant at the 0.01 level (2-tailed).**.
Bivariate correlations
- - - P A R T I A L C O R R E L A T I O N C O E F F I C I E N T S - - - Controlling for.. AGE CAI BMI CAI 1.0000 .8057 ( 0) ( 129) P= . P= .000 BMI .8057 1.0000 ( 129) ( 0) P= .000 P= . (Coefficient / (D.F.) / 2-tailed Significance)
Partial correlations
0
50
100
150
200
250
300
350
400
0 50 100 150
Ca intake
BM
IScatter plot CAI vs BMI
(r = -0.447)
0
50
100
150
200
250
300
350
400
0 50 100 150
Ca Intake
BM
I
25 35 45 55 65 75
3D Plot
Age, CAI and BMI
Transformations Theorem
Let x1, x2,…, xn denote random variables with joint probability density function
f(x1, x2,…, xn )
Let u1 = h1(x1, x2,…, xn).u2 = h2(x1, x2,…, xn).
un = hn(x1, x2,…, xn).
define an invertible transformation from the x’s to the u’s
Then the joint probability density function of u1, u2,…, un is given by:
11 1
1
, ,, , , ,
, ,n
n nn
d x xg u u f x x
d u u
1, , nf x x J
where
1
1
, ,
, ,n
n
d x xJ
d u u
Jacobian of the transformation
1 1
1
1
detn
n n
n
dx dx
du du
dx dx
du du
ExampleSuppose that x1, x2 are independent with density functions f1 (x1) and f2(x2)
Find the distribution of
u1 = x1+ x2
u2 = x1 - x2
Solving for x1 and x2 we get the inverse transformation
1 21 2
u ux
1 22 2
u ux
1 2
1 2
,
,
d x xJ
d u u
The Jacobian of the transformation
1 1
1 2
2 2
1 2
det
dx dx
du du
dx dx
du du
1 11 1 1 1 12 2det
1 1 2 2 2 2 2
2 2
The joint density of x1, x2 is
f(x1, x2) = f1 (x1) f2(x2)
Hence the joint density of u1 and u2 is:
1 2 1 21 2
1
2 2 2
u u u uf f
1 2 1 2, ,g u u f x x J
Theorem
Let x1, x2,…, xn denote random variables with joint probability density function
f(x1, x2,…, xn )
Let u1 = a11x1+ a12x2+…+ a1nxn + c1u2 = a21x1 + a22x2+…+ a2nxn + c2
un = an1 x1+ an2 x2 +…+ annxn + cn
define an invertible linear transformation from the x’s to the u’s
1 or u Ax c x A u c
Then the joint probability density function of u1, u2,…, un is given by:
1 1
1, , , ,n ng u u f x x
A
1 1f A u c
A
where11 1
1
detn
n nn
a a
A
a a
Theorem
Suppose that The random vector, [x1, x2, … xp] has a p-variate normal distribution with mean vector and covariance matrix
x
u A c
then
has a p-variate normal distribution
with mean vector
and covariance matrix u A A
u Ax c
Theorem
Suppose that The random vector, [x1, x2, … xp] has a p-variate normal distribution with mean vector and covariance matrix
x
u A c
then
has a p-variate normal distribution
with mean vector
and covariance matrix u A A
u Ax c
Proof
11
2/ 2 1/ 2
1e
2
x x
pf x
then 1 1 g u f A u c
A
1 1 11
2/ 2 1/ 2
1 1e
2
A u c A u c
p A
1 1 11
2/ 2 1/ 2
1 1e
2
u c A A A u c A
p A
since
1/ 2 A A A A
11
2/ 2 1/ 2
1e
2
u A c A A u A c
pg uA A
1 1 A u c A u c A
Also
1/ 21/ 2 1/ 2= A A A A A and
11 1 1 A A A A and
hence
QED
Theorem
Suppose that The random vector,
has a p-variate normal distribution with mean vector and covariance matrix
x
Ax A
with mean vector
and covariance matrix Ax A A
then has a p-variate normal distributionAxLet A be a q p matrix of rank q ≤ p
proof
u
AC
B
then
is invertible.
and covariance matrix
, = u
A A A A BA B
B B A B B
A Axu Cx x
B Bx
Let B be a (p - q) p matrix so that
AC
B
is p–variate normal with mean vector
AThus the marginal distribution of
and covariance matrix A A
Ax
is q–variate normal with mean vector
Related Documents
