Standard-Matrix-Analyse
Coombs, C.H. (1941). A criterion for significant common factor variance. Psychometrika 6, 267-272.
In: Guttman, Louis (1966). Order Analysis Of Correlation Matrices, p. 446.
In: Cattell, Raymond B. (1966, Ed.). Handbook Of Multivariate Experimental Psychology. Chicago: Rand & McNally.
Zusammenfassung - Summary - Abstract: Die Matrix vom Coombs
ist nicht positiv semidefinit, sie ist indefinit mit zwei größeren
negativen negativen Eigenwerten (-.02013 und -.07025) entgleist,
die durch Rundungsfehler nicht erklärbar sind. Das heißt, die
Korrelationen müssen in unzulässiger Weise gewonnen worden
sein. Damit ist die phänomenologisch erscheinende Matrix im
mathematischen Sinne keine "genotypische" (echte) Korrelationsmatrix
(Produkt-Moment oder Bravais Pearson). Die Entgleisung zeigt sich sehr
drastisch durch einen multiplen Korrelationskoeffizienten von 1.5715
(!) und insgesamt 5
multiplen Korrelationskoeffizienten > 1.
Samp _Ord_ MD_ NumS_ Condition_Determinant_HaInRatioR_ OutInK_ Norm_ C Norm ? 22 -1 --2 452.8 0 8.34D-11 44.8 .013(0) -1(-1) |
|
********** Summary
of standard correlation matrix analysis ***********
File = coombs41.k22 N-order= 22 N-sample=?
Rank= 22 Missing data = ?
Positiv Definit=Cholesky successful________= No with 2 negat.
eigenvalue/s
HEVA: Highest eigenvalue abs.value_________=
9.1144109647408429
LEVA: Lowest eigenvalue absolute value_____=
.020127036090946055
CON: Condition number HEVA/LEVA___________~=
452.84417057516329
DET: Determinant original matrix (OMIKRON)_=
6.0403604739505258D-11
DET: Determinant (CHOLESKY-Diagonal^2)_____= -999 (not
positive definit)
DET: Determinant (PESO-CHOLESKY)___________= -999 (not
positive definit)
DET: Determinant (product eigenvalues)_____=
6.0403604739505181D-11
DET: Determ.abs.val.(PESO prod.red.norms)__=
6.0403604739505258D-11
HAC: HADAMARD condition number_____________=
4.1650445190459257D-18
HCN: Heuristic condition |DET|CON__________=
1.3338717524570506D-13
D_I: Determinant Inverse absolute value____=
16555303352
HDA: HADAMARD Inequality absolute value___<=
1.9843086250570484D+20
HIR: HADAMARD RATIO: D_I / HDA ____________=
8.3431091024188185D-11
Highest inverse positive diagonal value____=
14.081193508
thus multiple r( 10.rest)_________________=
.963837794
Highest inverse negative diagonal value____= -.680449043
thus multiple r( 22.rest)_________________=
1.571501754 (!)
and there are 5 multiple r > 1 (!)
Maximum range (upp-low) multip-r( 22.rest)_=
2.509
LES: Numerical stability analysis:
Ratio maximum range output / input _______=
44.751549794654189
PESO-Analysis correlation least Ratio RN/ON=
.013422 (<-> Angle = .77 )
Number of Ratios correlation RN/ON < .01__ =
0
PESO-Analysis Cholesky least Ratio RN/ON__ = (Not positiv definit)
Ncor
L1-Norm L2-Norm Max Min m|c|
s|c| N_comp M-S S-S
484 190.2
10.17 1 -.07
.364 .209 26565
.231 .186
class boundaries and distribution of the correlation coefficients
-1 -.8 -.6 -.4 -.2 0
.2 .4 .6 .8 1
0 0 0
0 12 98 170 138 22
44
Original data with 2, input read with 2, computet with
19,
and showed with 2 digit accuracy (for control here
the analysed original matrix):
1 .59 .5 .3
.21 .13 .16 .12 .16 .18 .21 .18
.18 .14 .12 .04
.59 1 .74 .59 .46
.22 .24 .2 .21 .14 .13 .14
.14 .08 .03 -.07
.5 .74 1 .51 .37
.27 .25 .27 .24 .16 .14 .16 .18
.09 .03 0
.3 .59 .51 1 .51
.45 .47 .45 .44 .32 .33 .33 .36
.17 .13 .08
.21 .46 .37 .51 1
.54 .52 .5 .5 .38 .32 .35
.35 .14 .11 .06
.13 .22 .27 .45 .54 1
.91 .85 .79 .4 .38 .38 .36
.26 .21 .16
.16 .24 .25 .47 .52 .91
1 .94 .91 .46 .44 .44 .42
.33 .33 .25
.12 .2 .27 .45 .5
.85 .94 1 .89 .49 .47 .48
.45 .36 .4 .3
.16 .21 .24 .44 .5 .79
.91 .89 1 .51 .49 .48 .48
.36 .4 .33
.18 .14 .16 .32 .38 .4
.46 .49 .51 1 .92 .96 .91
.47 .43 .37
.21 .13 .14 .33 .32 .38
.44 .47 .49 .92 1 .96 .92
.47 .44 .32
.18 .14 .16 .33 .35 .38
.44 .48 .48 .96 .96 1 .95
.48 .46 .39
.18 .14 .18 .36 .35 .36
.42 .45 .48 .91 .92 .95 1
.48 .49 .41
.14 .08 .09 .17 .14 .26
.33 .36 .36 .47 .47 .48 .48 1
.74 .6
.12 .03 .03 .13 .11 .21
.33 .4 .4 .43 .44 .46 .49
.74 1 .72
.04 -.07 0 .08 .06 .16
.25 .3 .33 .37 .32 .39 .41
.6 .72 1
.13 .25 .32 .43 .44 .49
.49 .5 .52 .64 .59 .62 .59
.32 .27 .27
.08 .01 .18 .29 .31 .41
.41 .46 .47 .6 .29 .62 .6
.33 .35 .31
.12 .15 .22 .28 .31 .34
.33 .39 .39 .44 .45 .43 .44 .36
.3 .29
0 -.02 0 .18 .24
.26 .26 .33 .32 .51 .53 .54 .48
.38 .28 .35
.09 .05 .07 .14 .17 .23
.24 .15 .26 .33 .34 .35 .31 .36
.24 .31
.13 0 .01 .12 .13
.16 .19 .27 .25 .45 .44 .47 .49
.33 .29 .33
.13 .08 .12 0 .09
.13
.25 .01 .15 -.02 .05 0
.32 .18 .22 0 .07
.01
.43 .29 .28 .18 .14 .12
.44 .31 .31 .24 .17 .13
.49 .41 .34 .26 .23 .16
.49 .41 .33 .26 .24 .19
.5 .46 .39 .33 .15 .27
.52 .47 .39 .32 .26 .25
.64 .6 .44 .51 .33 .45
.59 .29 .45 .53 .34 .44
.62 .62 .43 .54 .35 .47
.59 .6 .44 .48 .31 .49
.32 .33 .36 .38 .36 .33
.27 .35 .3 .28 .24 .29
.27 .31 .29 .35 .31 .33
1 .57 .48 .44 .33
.32
.57 1 .45 .46 .34
.34
.48 .45 1 .58 .64
.53
.44 .46 .58 1 .8
.71
.33 .34 .64 .8 1
.8
.32 .34 .53 .71 .8 1
i.Eigenvalue Cholesky i.Eigenvalue
Cholesky i.Eigenvalue Cholesky
1. 9.11441 1
2. 3.05858 .8074
3. 1.95686 .668
4. 1.65374 .7967
5. 1.46789 .8355
6. .81712 .7963
7. .65813 .4056
8. .51057 .3266
9. .48039 .3872
10. .41222 .8242
11. .3872 .3799
12. .36244 .1931
13. .27688 .2961
14. .2338 .8485
15. .20248 .5943
16. .17469 .6333
17. .16373 .6816
18. .07963 -1.0387
19. .05279 .6538
20. .02683 .5897
21.-.02013 -.0641
22.-.07025 -.2761
The matrix is not positive definit. Cholesky decomposition
is not successful.
Eigenvalues in per cent of trace = 21.999999999999999
1 .4143 2 .139 3 .0889
4 .0752 5 .0667 6 .0371
7 .0299 8 .0232 9 .0218
10 .0187 11 .0176 12 .0165
13 .0126 14 .0106 15 9.2D-3 16 7.9D-3 17 7.4D-3
18 3.6D-3
19 2.4D-3 20 1.2D-3 21-9D-4 22-3.2D-3
analysed: 12/05/05 22:04:36 PRG version 05/24/94
MA9.BAS
[Interne Verwaltung: File = C:\OMI\NUMERIK\MATRIX\SMA\coombs41\coombs41.SMA
with data from C:\OMI\NUMERIK\MATRIX\SMA\coombs41\coombs41.k22
Date: 12/05/05 Time:22:04:36]
Suchen in der IP-GIPT,
z.B. mit Hilfe von "google": <suchbegriff>
site:www.sgipt.org
z.B. Numerisch instabile Matrizen site:www.sgipt.org. * Korrelation site:www.sgipt.org |