a multivariate analysis of certain biochemical components of … · 2016-06-02 · cnco o co m en...
TRANSCRIPT
A MULTIVARIATE ANALYSIS OF CERTAIN BIOCHEMICALCOMPONENTS OF EQUINE AND FELINE SERUM SAMPLES
AS REPORTED BY AN AUTO-ANALYZER SYSTEM
By
Robert Rothnick Jorgensen
United StatesNaval Postgraduate School
rsnrHESISA MULTIVARIATE ANALYSIS OF CERTAIN BIOCHEMICAL
COMPONENTS OF EQUINE AND FELINE SERUM SAMPLES
AS REPORTED BY AN AUTO--ANALYZER SYSTEM
by
Robert Rothni ck Jo rgensen, Sr.
Thesis Advisor A. F . Andrus
March 1971
Apptiovnd faon. public A.e£ea.6c; cUAtnlbiition 'Mitlmltzd.
T.1 37477
A Multivariate Analysis of Certain Biochemical Components
of Equine and Feline Serum Samples
as Reported by an Auto-analyzer System
by
Robert Rothnick Jorgensen, Sr.Lieutenant Colonel, United States Army
B.S., University of Minnesota, 1957D.V.M. , University of Minnesota, 1959M.P.H., University of Minnesota, 1965
Submitted in partial fulfillment of therequirements for the degree of
MASTER OF SCIENCE IN OPERATIONS RESEARCH
from the
NAVAL POSTGRADUATE SCHOOLMarch 19 71
LibraryJJAVAL POSTGRADUATE SCHOOLMONTLT . 93940
ABSTRACT
This thesis contains a multivariate statistical analysis
of the results of an automated analysis of serum samples
from the horse and cat. In the horse, 12 biochemical com-
ponents plus body weight and age are recorded; thus,
observations are made on 14 random variables. In the case
of the cat there are observations on 13 random variables.
Ninety-one (91) pair-wise correlation coefficients are
computed from the equine data and 78 pair-wise correlation
coefficients are computed from the feline data. Extensive
hypothesis testing concerning these correlation coefficients
is conducted and the results are presented. A discriminant
analysis for 2 groups, male and female, is conducted for
each species. In this analysis the vector of sample means
of the biochemical components plus body weight and age for
males is contrasted with the corresponding vector from
females. Tolerance limits for each biochemical component
measured are presented for both . species
.
TABLE OF CONTENTS
I. • INTRODUCTION 6
II. A BRIEF CONSIDERATION OF THE AUTOANALYZER 8
III. THE MAMMALIAN SPECIES UNDER STUDY ANDCOLLECTION OF SAMPLES 9
IV. TABULAR PRESENTATION OF DATA 10
V. SAMPLE STATISTICS 16
VI. HYPOTHESES TESTED 30
VII. TOLERANCE LIMITS 44
VIII. SUMMARY 48
IX. RECOMMENDATIONS 56
LIST OF REFERENCES 57
INITIAL DISTRIBUTION LIST 59
FORM DD 1473 60
LIST OF TABLES
Table
I - Serum Biochemical Values in 44 Horses asReported by an Auto-analyzer System 11
II - Serum Biochemical Values in 30 Cats asReported by an Auto-analyzer System 14
III - Table of Sample Mean Vectors 20
IV - A Variance-covariance Matrix for 44 Horses 22
B Variance-covariance Matrix for 21 FemaleHorses 23
C Variance-covariance Matrix for 21 FemaleHorses 24
D Variance-covariance Matrix for 30 Cats 25
E Variance-covariance Matrix for 15 MaleCats 26
F Variance-covariance Matrix for 15 FemaleCats 27
V - Critical Values r-j ^ for the Specified afor 44 Horses and 30 Cats 31
VI - A Correlation Matrix for 44 Horses 32
B Correlation Matrix for 30 Cats 33
VII - Critical Values of r j ^ for the Specifieda for 23 Male Horses and 15 Male Cats 34
VIII-A Correlation Matrix for 23 Male Horses 35
B - Correlation Matrix for 15 Male Cats 36
IX Critical Values of r j ^ for the Specifieda for 21 Female Horses and 15 FemaleCats 37
Table
X - A Correlation Matrix for 21 Female Horses 38
B Correlation Matrix for 15 Female Cats 39
XI - Statistically Significant Differences inSample Correlation Coefficients in Maleand Female Horses 41
XII - Discriminant Analysis Between Males andFemales 43
XIII-A Tolerance Limits in the Horse 46
B Tolerance Limits in the Cat 47
I. INTRODUCTION
There is a basic oneness of the human and veterinary
medical professions [1] . Many diagnostic and therapeutic
devices developed for use in one of these professions have
sooner or later found equally significant application in
the other. In particular, the advent of automated devices
for the analysis of human serum has greatly facilitated
biochemical profiling [2]. However, as Coffman reports,
restricted availability of these devices in the past has
limited their practicality in clinical veterinary medicine
[3]. Coffman goes on to point out, and this writer has
encountered no statement to the contrary, that the appli-
cability of these devices, in enhancing the diagnostic
capabilities of the clinician, will be cause for their
increased utilization by the veterinary medical profession
[4] . It is therefore logical to assume that automated
biochemical profiling will be of increasing value in
describing the normal intervals about the true biochemical
parameters in the serum of domestic animals.
Although the auto-analyzer devices are capable of
generating a large volume of data concerning biochemical
values in human and animal sera, the proper analysis of
these data should not be treated superficially. The lack of
proficiency in statistical methodology by members of the
bio-medical professions is an unfortunate fact [5].
Additionally, qualified statisticians are often asked to
make meaningful conclusions from biological data that have
been generated in meaningless fashion [6] . The increased
reliance of the veterinary medical profession upon automated
analyzing devices will yield little addition to our
knowledge of serum biochemical parameters and their con-
comitant physiological significance unless the data are
meaningfully collected and analyzed by sophisticated
statistical techniques. It is this conclusion which has
motivated the research effort reported herein.
II. A BRIEF CONSIDERATION OF THE AUTOANALYZER
The autoanalyzer used in this study was a Technicon
Sequential Multiple Analyzer, SMA 12/60, which was first
introduced in 1967 as a successor to the widely used
SMA 12 [7] . The SMA 12/60 analyzes a serum sample
quantitatively and reports on the concentration of 12 of
the serum biochemical components on an easily interpreted
pre-calibrated chart paper. The 12 biochemical components
quantitatively analyzed are: calcium, inorganic phosphate,
glucose, blood urea nitrogen (BUN), uric acid, cholesterol,
total protein, albumen, total bilirubin, alkaline phos-
phatase, lactic dehydrogenase (LDH) , and glutamic oxalacetic
transaminase (SGOT)
.
The primary value of the autoanalyzer is that it provides
a reasonable biochemical profile of the subject from a
single serum sample within 60 seconds; it is invaluable as
a screening device in that certain metabolic aberrations,
whether suspected or unsuspected, will be manifested in
abnormal serum chemistry charts.
III. THE MAMMALIAN SPECIES UNDER STUDYAND COLLECTION OF SAMPLES
The horse has recently enjoyed a resurgence of popu-
larity although it is more in demand as a pleasure and
recreation animal rather than a source of agricultural
power. It was selected for this study because it was felt
that the economic value of the horse justifies the most
sophisticated diagnostic techniques by consulted veterin-
arians. While some data are available as to equine
biochemical parameters the work thus far reported does not
fully exploit the statistical potential of these data [8].
In this study, 44 horses of various breeds, 23 males
and 21 females, were selected. All were clinically in
good health. Blood samples were obtained by vena puncture;
the serum was separated from the clot and was then centri-
fuged, frozen, and maintained in a frozen state until
submitted to the laboratory.
The cat was selected for this study because information
concerning biochemical parameters in the serum of this ani-
mal is particularly lacking [9]. Thirty (30) cats, 15 males
and 15 -females, all of mixed breed but of varying ages were
included in this study. All were free of clinical illness.
Blood samples were obtained by cardiac puncture and the
samples were then processed as described above.
IV. TABULAR PRESENTATION OF DATA
In consideration of the equine data 14 variables were
under consideration; in addition to the 12 serum biochemical
components reported by the autoanalyzer , body weight and
age were also considered. A previous report indicates
that equine serum albumen is not reliably reported by the
autoanalyzer [10] . Albumen is nonetheless included in this
study to form a basis for substantiation or repudiation of
this fact as may be justified by further research.
A similar approach was employed with the feline data
with the exception that 13 variables were under considera-
tion; inasmuch as no cat demonstrated measurable total
bilirubin it was deleted from consideration.
The data as reported by the SMA 12/60 Auto-analyzer for
the 44 horses and 30 cats are presented in Table I and
Table II respectively.
10
gCO Wfa EhCO cofa XO COK
fa•<* w^ n
><
S JH <2
CO <fa 1
H D OA Eh
pa < DJ > <CQ
<Eh 3 <
uH >H
£ CQfak au wO EhH faCQ O
fa£ faD 2fafa COCO <
* K * * * * Hexas fa fa' £ fa £ Cu £ £ £ £ fa fa 2 fa fa fa
in in in
bBv coi-H
00 o o inrH
o\ •^ CO inrH
rH r^ 00 in ^ orH
O o o o in o o o o o O o o o o o C^q6T3M in o o o r^ o in in o o in o o O o o in
00 o in in o rH o er> CM rH o 00 rH rH CM rH rHrH rH rH rH H rH rH rH fH rH rH r—i
o O o o o O O o O O o o O O o O oiODS vo VO in •^r vo <T> ro in VO "^ ro o ro ro ro o> C">
•<^ in vo r-» in CMrH
in r^ in 00 rorH
r- vo vo vo in in
H ** o in in r^- o m CA o o o o o m rH CMhcti rH (N in in o 00 • CM in in in ^r in in vo ^ in
(N CM (N CM CM rH ro ro CM CM CM CM CM rH CM rH CM
•SOLId r- in O o rH 00 00 in CM O CO in in rH r^ VO CMco o O in rH rH r^ CM r-« CM ro ro CTi CM o ^r CTi3UTXP>[XV' rH •^ ro rH rH rH rH rH -H ro rH H rH
urqnjTxxg CTi o O 00 ro o CM O O 00 vo o^\ o 00 CM 00 rH
*JL O rH O o rH CM rH rH rH O o o rH o rH O rH
VO VO vo vo in •<T •^ in in VO in •^r in in VO m inuauinqxv
O o o o o O o o o O o o o o O o o
UlS^OJd ^r cr> r-» in in 00 vo ro vo in ro vo 00 ro o •"vj" in
T^lOi VO vo in in vo vo vo vo vo vo vo vo in vo vo vo Vi)
Xoxeq. O in in CM rH «=r o r» CM CM rH rH o 00 o •"^ C4rH ro ro in O o^ cr> o ^ rH rH o o rH rH o rH-saxoi{3 iH rH rH rH rH rH H rH rH rH rH rH rH rH H
pxov LD in in o ^ *r *r "3« *r r^ •^ •«* co •^r 00 ro C-)
OTJfl O o o rH O o o O o o o O o o O o o
ms O rH ro CM VO <j\ ^r 00 rH ^r vo ro rH vo in r^- r-H
rH rH H rH rH rH CM rH CM rH rH rH iH rH tH H rH
VO m 60 CM VO CT» 00 in ^ ro o m <J\ rH VO m CMssoonxD 00 00 cxi O
rH>> r- »> 00 O^i 00 CO o
rHr^ 00 00 00 r^
•soqd CO rH •*r >> o -H ro ^ r^ r^ r-~ rH CM r-» CO !>• r-
•joui ro ro vo vo ^r ro ro ro ro ro ro in ro ro CO CM ro
CM 00 o ro CM in •*r O rH vo H CM 00 CM c^ VO inuiriTOXE3
rH rH CN rH rH o rH rH CM rH H CM rH CM rH rH rHrH rH rH H rH rH rH rH H rH rH rH rH H rH rH rH
OJ OJ OJ 4J d)
CO CO CO C o CO
XI U u U •H c ucu o O CO •H oa) d) ffi ffi K CU fa e OJ a) OJ S*H HD T5 • • • T3 •
1 1 1 | 1 -r) T3 'O •
CQ fd rd u u U 03 u rH n3 rG cd Sh
u >H -p -p -P U -p fd u U u -Po O a a a O a fa o O o a
•
rH CN ro *r in vo r^ 00 a\ o rH CM ro •^ in vo •>
2 rH rH rH rH rH H rH rH
11
X3S1
fa fa*s fa
*s
*fa fa
*fa g
*s
*fa fa s
a5v CN CTi VD CT. <J\ co r- COrH
VD •*r VD CNrH
CNrH
orH
in• ^ CN
rHin
^qb-paMOOOi-H
oorHrH
OoCTi
OOOrH
OoorH
ooCNrH
oinorH
OOCn
OoorH
ooorH
OOen
OmCO
OOorH
OorHrH
ooCTi
oin
OOOrH
ooorH
10DSOin
oin
OCOin
o00orH
orH
rH
Oinr-
oinCO
oCOVD
OOinCN
ooVD
orH
oCO
O0~\
*3»
OOCO
oVD
oVD
OCOVD
OorHCN
HCfloinin
in00rH
ooCN
OCOCN
inCOCN
HCNCN
VDoCN
VDOCN
OCN
corHCN
inVDH
VD
rH
enorH
•**
CNCN
oCTi
rH
oVDCN
COVDrH
inCOCN
•soqa oCN
00COrH
oinrH
inCOrH rH
inCOrH
morH
enCX>
inrHrH
VDorH
in
rH
COCO
inCNrH
VDcn
inCTi
rH
rHr-rH
VDCOrH
UTqnjrrF9•1
rH
rH rH
00
oin
oCO
OrH
VD
mCN
00
rH
CN
rH
CN
rH
orH
CO
rH
CO
CN
rH
CN
orH
CN
rH
VD
rH
CO
rH
uauinqxvin
o oin
omo
mo
rH
O O oVD
Oin
oin
o o O Oin
oin
oin
omo
UTsq.oad rH
VD
in
VD
in
VD
CN
VD
rH
VD
in
VD
in
VD VD
oVD VD
o VD
VD
o rH co
VD VD VD
in
vfi
-[OJtsq.
-saxoqo
oCNrH
ocorH
OrH
Oen
COOrH
oCNrH
VTi
rHrH
OCOrH
rHCNrH
oH
orH
coen
COrHrH
orHrH
oinrH
CNCTi
inCTi
CO3\
pTDV in
o oco
oin
oin
o<*
o o O o o O oin
o o oVD
O OCO
O
una•>*
rHinrH
VDrH
rHrH
CNrH
COH
VDH
VDrH
inrH
CO CNrH
enrH
COrH
VDrH rH
inrH
COrH
asoonxoCO
in00
inCO CO
in rHCO
inCO
enCO
inCO
oCO
O00
in
rH Cn COOCO
men
•soqa• ioui
CM
in
rH CN
co
in
CO
CO
CO
•"3"
^in
co co
CO
CN
CTi
CN
en
CN
CN
CN
Cn
CN co
rH
CO
in
CO
UIRTDX^Dco
rHrH
rHrH
rH
rH
CO
CNrH
CNrH
in
orH
rH
CNrH
in
rHrH
CO
rHrH
en
rHrH
CN
CNrH
00
rHrH
OrHrH
CO
CNrH
rHrH
rHrH
CN
rHrH
rH
CNrH
<U
(1)
Um
CD
wuo55
•
U4->
a
CD
rd
5-i
u
rd
CO
rH05
CU
<
<D
fd
UID
oC5
-Hg
rHcd
CO
OrHrd
a,
<
CD
13fd
U
CD
13fd
Uo
T3CD
Um
U rHCD T3g fd
< en
cfd
•HX!fd
u<
CD
fd
uU
Oc•HsorHfd
(D
CO
u
55
U-Pa
13CD
UCQ
134-)
in
CD
rd
U
CD
fd
u
CD
13rd
Uo
i a)
o uu rd
b a
2COrH rH
oCN
rHCN
CNCN
COCN CN
inCN
VDCN CN
COCN
CnCN
oCO
rHCO
CNCO
COCO CO
mro
12
X9S fa fe
* *2
* *fa
*fa
a6v in cx» as <r> r> 00 CM 00 cr>
H\q6xaM mCTi
in
rH
inCNrH
CNrH
rHrH
OO00 rH
rH00
LODS 00rH
Oroin in
Ovo
OO
Om
OVO
OvoVO
HOlIT)
inrH
OsOCN
00coCN
00rH
ininCN
mmrH
in
CN
«3"
inrH
inasrH
•soqa Os
rH
0000
ro
rHCT»
inCNCN
rHrH
OrH
OrH
OinrH
UTqnaTXTe CO•
CN
VO•
rH
as•
CN•
in•
rH•
rH
cr> VO•
rH
ro•
rH
ueuinqxvin in VO
OVO
O O Oin •<3*
O O
uxaq.oj:dVO
VO
as
VO
rH rH 00
VO
CN
vo VO
rH
VO
Xoasq.-saxoqo
rHrH
CNCNrH
inrHrH
in
rH
CNro
rH
ro
rH rHrH
pxovO O
00
Oin
Oro
wna rorH
O Or-i rH rH rH
asrH
00 CNCN
8SOOUXD in00
OOS 00
O00
rorH
COas
in00
00CO
•soqd•JOUJ ro
mCN
O 00
CN
as
CN
VD
CN
CN
ro
CO
CN
uinxox^Dro
CNrH
r>
CNrH
rH
rorH
VO
CNrH
rHrH
OCN CN
rHCNrH
rH
rHrH
mCD CD
rd
UO
<d
wu
•
u-p
a
CD
V)
UOK
•
U-Pa
•H£OrHrd
CD
u
• CD
-P >iOEh
rd
•H
rd
Jh
<
CD
rd
Uu
CD
CO
u
CD KTS •
rd UU -Pa
•
o VOro
00ro
asro "3"
rH CN ro
rH ro h^rCN cn]-^II II
fa 2
13
w
aCO faEh Eh
<; coU X
cooro «W
04 eg
O >*Cm JCO 2w <D I
H* O< Eh
> D
u <Hw mKU Qo wH EhCQ tf
O5 faD Wfa fafaCO CO
CQ
Pc TD
mme
0)
to
Qau CO
X3S a a S fa a fa fa S fa s s fa S fa fa a a
m in r^ 00 ro CO inaBv • • • • • • •
CN •^ o H rH iH ro ^ rr o CN o o o ro CN rH
^qBxaM r- oi-H
** r^ 00 r^ CN in in r* r~ in ro ro r- 00 r^
m o o o CN 00 in ^r o m 00 CN CN 00 00 ^r ro
103 S m en r- VO VO in CO r-> o Is* -^ in r~ in r^ vo
r-H
CTi
O LD in in in o in m o o o o o in o in Ohctt ^ VO ro CTl CO r» r^ rH in rH m CN rH CN r^ ^
CN CN ro iH r-H CN ro CO ro r^ 00 CN CO CN CN CN ro
•sona CN LD r* !>• CN in o in r~ CN o o o in r-\ ro CNr-» ^r CO CTl VO in ro r~ a\ 00 <D r~- •=T •** •rf •^ CO
sotip^tv rH r-i
oo vo CN r^ VO r^ m T~\ m ^ in in «=r VO r- CO vo
usuinqxvrH iH H H i-H rH iH *-* rH rH rH r-i rH r-\ t-i <-{ •-{
UTaq.o^d •^ r» in in "^ VO r» CTi rH ro ro in CN •*r t-\ <r vo
T^^oj, VO VO in v£) VD VO r* VO r- U3 vo VD vo V£> r^ i> vo
TOJa} LD •<* vo in in in r* r- CN CTl in KD in CO CN o rHtH ro r» r-« r» 00 o CTl r-\ VO oo y& •^ in in o CO
-saxoqD iH iH rH r-i ^ <-{ rH
pxov CM VO r^ VO 00 00 CTt CTl o 00 o CN CN o CTl r» r-
of^n H o o O o o O o <-\ o t-\ rH rH r-\ o o o
"<* ro VO in VD CN m in r^ ro V£> r~ 00 VO r-\ rH rH
Nna CN CN CN «-H CN CN CN CN CN CN CN CN CN CN ^f CN CN
CTi CO O ^r ^ O O O CO O CTl 00 O CO in CN 00
asooni3 VO in r~ VO in VO VD in vo ro V£> in 00 CTl vo VO ro
•soqa ro CO 00 00 CO 00 O o r^ 00 O cr. CTl "a< <T O m•Bjoui r^ vo r» VO r- VO VO c\ CO [** 00 r^ r- r^ in 00 CO
•^ CO in O H o\ CN o <^> rH CN •^ »x> r^ rH VO ro
uinTOx^D o cr> <j\ Ot-H
OiH
CTl CTl i-H
rHCTl o O CTl CTl CTl CTl CTl CTi
•
o r-\ CN ro •^ in VO r- CO CTi O >-{ CN ro ^ in VO r-
2 r-i f-\ rH •-A r-^ r-{ <-{ <-i
14
pc<D
OU
T3O>H
OJ
•PPCD
CO
xss fa s fa fa 2 £ 2 fa 2 fa fa fa fa
©6v iH CM CN CN m vo CM
VO•
O VD in 00 rH
in*
rH
^qBxaM vo rHH
m VO orH
00 00 in CT> CMrH
r* in en
I.ODScooiH
i>rH
CN coorH
CN o O CMiniH
VDCOVD
CM COCO
morH
HOIin00iH
incoCO
00voco
VOrHCN
ovoCN
rHinCM
CMCOrH CM
VO
rH
in oenCM
VDCM
•soqd OCM
ovo
oco
VOCM
CNro
rHco
oCO
CMVD
CNin m
CMin
voin
CTi
CM
ueuinqxvco 00
i-H rH rH rH
in
rH
CO
rH
VO
rH
VD
rH
CO
rH
vo
rH
CO
rH rH
uxaq-O^d rH in
VO
CN
VD
o o o O CO
vo
CO CO CO
VD
o
XOjaq.-SBioqD iH
iH
vo
rH
00co
VDCM
mrH
oorH
COVO
CTl
VDinin
rHin
rHco
oCO
CMvo
p-cov 00
O00
o orH
rH
orH
OH
OrH
CTi
OCO
ocy>
oorH
CTi
oorH
nhhOCN CN
VDCN ro
CNco
CTi
CNrHCO
VDCM
rHCN
VDCM
CNCN
COrH
CTl
CN
asoonxDLDLD
CNrHrH
VOmin
o Ocr>
o00
OrHrH
VDVDrH
r-rHrH
CTi
inCTi
r--
cri
•soqd•Bjcoui
00 CN
CO in
rH
VO VO
CM
in
00
in
vo
vo CO
uiriTox^DoCTi
VO
00
in
00
co
00
CO
00
CTi OOrH
CN
CO
rH
00
o00
CO
00 00
•
o 00H iH
oCN
rHCM
CMCM
COCM CM
inCM
VDCM CM
00CM CM
oCO
15
V. SAMPLE STATISTICS
The analysis of the data is amenable to multivariate
methodology. In the case of the equine data we have obser-
vations on 14 random variables:
X. . 1 = 1,2,.... ,44
where "i" refers to the identification of the animal and j
refers to a variable.
In particular we have:
Random Variable Definition
X. , Calcium Level of Horse Number "i"1/1
X. _ Inorganic Phosphate Level of Horse "i"l , z
X. - Glucose Level of Horse "i"l , j
X. „ BUN Level of Horse "i"
X. _ Uric Acid Level of Horse "i"i/5
X. c Cholesterol Level of Horse "i"i / o
X. _ Total Protein Level of Horse "i"i / '
X. Albumen Level in Horse "i"I/O
X. n Total Bilirubin Level of Horse "i"i/9
X. , A Alkaline Phosphatase Level of Horse "i"1,10 v
X. ,, LDH Level of Horse "i"i/ll
X. , SGOT Level of Horse "i"
X. , .. Body Weight of Horse "i"1 / 1 J
X, ,
4Age of Horse "i"
16
In the case of the cat we have observations on 13 random
variables
:
Y. . i = 1,2, ,301 1 J
j = 1,2, ,13
where "i" and "j" again refer to the identification of the
animal and the characteristic respectively.
In particular we have:
Random Variable Definition
Y.,1
Y.,2
Y.1 ,3
Y.l ,4
Y.l ,5
Y.1 ,6
Y.l ,7
Y.l ,8
Y.l ,9
Y.1 ,10
Y.l ,11
Y.l ,12
Y.1 ,13
Calcium Level of Cat "i"
Inorganic Phosphate Level of Cat "i'
Glucose Level of Cat "i"
BUN Level of Cat "i"
Uric Acid Level of Cat "i"
Cholesterol Level of Cat "i"
Total Protein Level of Cat "i"
Albumen Level of Cat "i"
Alkaline Phosphatase Level of Cat "i'
LDH Level of Cat "i"
SGOT Level of Cat "i"
Body Weight of Cat "i"
Age of Cat "i"
In the description of sample statistics that follow, the
computational formulae are given for x. .. The same formulae
apply for y. and it is understood that N = 44 and N = 30
for the horse and cat respectively.
17
In multivariate terminology, each horse may be repre-
sented as a 14 dimensional vector X . For example,
referring to Table I we see that horse number 1 is repre-
sented as follows: "
X (1) _
— — — —xl,l
11.2
Xl,2 3.3
31,3
86.
Xl,4
10.
Xl,5
0.5
*1,6 110.
Xl,7 = 6.4
Xl,8
0.6
Xl,9
0.9
Xl,10
87.
Xl,ll
211.
Xl,12
460.
Xl,13
850.
Xl,14
13.
The sample mean, x..
, is required for each of the set of
observations on the 14 variables and is computed as follows:
Nx. . = Z
3 i=l
2-1N
The sample means of the observations on the j variables
may be represented as a j -dimensional sample mean vector,
x, where
18
X.
X.
X.
X.
X.
_1_
N
NEx
i=l lfl
NEx. ~
1=1 1*2
NE x. -.
i=l L ' 3
NE x. .
i=lX '=>
NE x.
i=l X 'P
The 6 sample mean vectors representing 44 horses, 23 male
horses , 21 female horses, 30 cats, 15 male cats, 15 female
cats, are presented in Table III.
2The sample variance, s . or s . .is computed for each
3 3 i 3
set of observations on the j variables as follows:
3 N" 1 i=l
NE (X_. i - x. .)
The sample standard deviation is computed by simply obtain-
ing the square root of the sample variance:
s . = Jl
The sample covariance, s. , , between the observations x. .
and x. , , i = 1,...,N, is computed as follows:1 , K
19
rH rH
1 oo r» r* VD CO r^ in CO o r~ CO CO CO u h\rH CO o o 00 *J" a\ •^ en in < 00 •^ cr> r^ o' 0) Ein (d -P • • \ • • • • • -P
4->
a3h E m <T» r» r>- in o rH VO rH 55 r^ a\ CM VD CM CO
<D CJ
1
r^ CM r» in CO 00 a) rHft CM rH
uCJ 13
CU
CU1 ° ON o co 00 CO O co co o O r-» colCD CO 1 "* co CM in 00 r^ VD in < in KD VO •*r •^ x: -p 4-> CU
in iH +j • • \ • • • • • V CU •H rHrH fd fd <T> r» CO in o r^ VD rH Z co CM <T\ r* CM •H rH 2 CU
S U1
r- CM 00 VO OCO
CO 5 CU
nials
'
mg%
dCO
rH(0 CP
r^ co co o o o o\ co CO CO r- co r- o VD| g g •HCO 1 CN CM in in <7\ vo r-» co r* m < VO o CO rH CM •H c oO -P • • • • • • • • • • \ • • • • • C O (0 VD
co <a CTi r» in m O <ti rH r-» VO H 2 O rH vo r^ CM d inU r- CM r-- en en VO r^ CO CTi V
* * CM VD v ** * *
1 ° in rH c\ 00 en CO O CM CO in sj" o CM oj T3CO cu co 1 r^ r^ 00 CM T CM •<* m CM CO <7\ CM O VD ^ CU
ft H Q) T3 -MO rH fd CO rH CO •*r •^ o r^ VO o rH VD a\ in in r~ r~- CU CU
EH <N g 5h H 00 rH rH •^r CM VD rH •^r +J T3 rH
U CU rH rH CM CO VD en CU CU CU
w ft ft * rH -M 13
> **
CD CD
rH •
ws
0)
T3D
I—I
HH D
•
E^
1f"" Cn r- 00 CO CM in r^ VO CM •^r in vD CO °M •
CO 00 ^r in r- •^r in in ^r in in r^ co co CO CO Eh D ow w Q) CD • oJ w- CO rH CO rH co 00 •^ o 00 vo o rH •** r» •^ rH CM VD O Eh CM
n ft CM (13 ^ rH 00 rH o ^* iH rH in CM o rH
< 2 S rH rH CM r* vd O CM o AB ^ » Jc rH rH oCO * A
CMrH
A
EH
OCJCO
ftO
Eh
OCJCO
1 °° CN r^ in in O en CO o CT\ r-» vd in CO * Eh
W CD ' i-» VD r*> in ^r r- "* •^ •^r CO in CO o a\ rH CO O X!kh" "* CO CJ -P
m ^ u rH CO VD •^ c CM in o rH in CO VD VD <D r- -C CO •H
< 1^ 00 rH rH -«* CM co co CO -p 2
Eh ffi rH rH CM r*» vd en -H x:CO
•H CU
rHCO 3:
rH fd
^-^ • to CU gdP D g rH CU
& • •H fd IH
g. Eh c g
^_^ W(a
rHm*
w <*p CO VD * *EH CO < * * *< ,-«. g Ehft <*> CJ> <
<#>
ftCO
OK o\° g
5 2H CO
2Hftft
ftftCOO
CP ft 0i »-* ^ W g ft ft Ehg g 9 EH CT> H ft • ft*»*
•
U — *-~. Q ft O *
—
J • D CJH OP M w ft H w D • H
s Z w C7> u Eh cu S ft s • Eh wD "^ CO S < CO w M Eh sM cj o — ft ft s ft ^u ft CJ CJ t-q 5! D < <* Eh >H
Hi o p z H o Eh ft Eh ^ ft O Q w< 2 IH" P ft S3 O
SO J a CJ O CJ
I u M cj « D CJ EH Eh < •H" CO ft <
20
s .
N(x. . - X. . ) (x. ,
- X.. )J,k N-l i=1 i,D 3 i/k k'
In multivariate analysis the sample variances of the
observation on the variables and the sample covariances
between each pair of these variables is presented as a
sample variance-covariance matrix (S) with dimensions p by
p. The elements of the sample variance-covariance matrix
are as follows:
(S) =
Sl,l
sl,2
Sl,3
•** Sl,k
s2,l
S2,2
S2,3
**• S 2,k
'1,P
;
2,p
3,1 2,2 3,3 D,k j ,p
p 7 l p,2 p,3 p,k P/P
In the main diagonal of the matrix j is equal to k and the
main diagonal elements represent the sample variances for
each of the j variables. Since s . , = s, ., for the sample
variance-covariance matrices computed, those elements above
the main diagonal will be deleted and the matrices will be
lower triangular. The variance-covariance matrices for 44
horses, 23 male horses, 21 female horses, 30 cats, 15 male
cats, and 15 female cats are presented in Tables IV, A
thru F.
21
«o
o
x;M«
W HhJ UCQ 2< <
K<>OuI
wu
<>
1CO
s£v CO
ojcoH
•^r cn *r
^qfixaM00 U3
• CNrH rH
ApoeCN
VDH
CN ro VD
K^rH ro
• •<* CTi
J.OOSiniHCO00H
00COro
HCN
1
• CN • ^r •<£• LO
HOT r~ rH <x> in HC • roin o ca r- HC r~ coin ro VD 00T ro 1 l
tn . ^ . rH • <x> M"
•soqd cc'^CO
"1
CTi •"
1
CO ^KD CNro cti
rHrHcn ro
ro
roin•PXTV CXi H o O
in H CN1
001
Tr* U3 H o ^r
irrqru.L[Lg o00 " 00
• CNH in . 00
cCN
1*101 o H!
H1
1—
1
ro O00 o
00
>»c -c in
uauinqxv oo • VOo o °5 ro ^r in o o
o 1 o o ^r 1 I
UTsq-O^dCNVOH
•
O
CT>
• oo o
1
enCNrH
c
. HO 00H IX)
1
c. CO
CTi CN1
in• m
oo cn1
ro roCN
co-.
^J1 o ro r^ vo V£> • iH . ro HXO^aq. * o o 00 CO • VD ^3" in oc rH CN • 0>
^r ro ro <-\ ^T ro r» 00 \o en O VO ro H-saxoqo MrH
1
• . •^ CN r- in HCM o o ro r~» l i 1
pTDVi—
i
(NVOH LO
#t-\
s CN• H m
CN 3 a-,in
o• CN
^T^nOO
r-•
oo o
1
oo
o o1
Oro
i—
i
<N inco ro
1
O H1
inro CN ro CN vo in • in
CO00 • rH CN O
inCNMns 00 •«H . CO • CN • o • ro CN 00 CN O O • H
• 0<H r-» in O i—
1
o o O VD ro cn co ^r Hr-roH 1 I 1 1 1 1 VO H ro O
CO ro -I CO CN iH 00 o • CN . cc • <c r^• 00 H ro vo "^ -=r CO • CN ro nr o^* ro cc H
asooniD ld ro ro ro r- in • o "^ oro ^ m o "* r*» ro . roin i • • • o • o rH 1> >* ^rrH VO o CO
ro o 1 o CN 1 1 1 l
r- CN sr o r» ^ CN r> in CN H i—
1
«*
•soqd *» vo ^r in ^ 00 i-H vo r^ O • C •r- roX 1—
1
• ro o VD •rH o • o • 00 • H in cc ^r vo • CTi
•6.10111o o
ol o "^
o1 o
o1 in
inCN l
COl
H1
Is- "* T o ro H ^3* o CO CTi o 00 CO COr» • VD • ro . CO o • in T <N • vo a\ • r^ • CN . o • VD
uiriTox^D no
O C1
O iHl
O L.O
1
co
O CO
l
OO
oo
O H1
mrH 1
rHHH
KD VO1
O CN1
<~\ CJ
O cn
U QJ S-i c 03
-H 4-> a) •H O 4-1
e c ro cu jj C c X! C rd
3 rd X (0 co •H 0) P -hx: 4J
•H CP c o cu H d) £ H M H Dj Xi^ to O'O H rd 4-> 3 rd -h rd co rH >i C"
H D 2 •H-H -P -Q «P H ^ *r* O T3 -H 0)
« C JC -I D U U r" n 1—
1
0-H H SZ 3 o O tn
' u m a. o a ^< u H CU < h a <Ch j CO C3 S <
22
towCO«ore
wXI
ooCM
o
XM«
Wu<M
>ou
I
wu<
>
r»
a6v rH00
en
Apoe25733.
664
00
• o00
• 00 oo or-» oo cn VjO
J.ODS o HcnrHrHrH
inCN
• cn<j\
r--
H1
• oo • ^ • KD CO\o rr r- oo p-» oo . «rp
senCOCN
^* en
m 001
o oor-
i
r~ o t • CO • CN•soqd
6708.
22VX5
rH^oinoo
I
^COCO
1
r- ooH OrH
1
00CN
• cno
•CNOO
irrqn.it ij.h
T^oj, CN
rH
in <a<H1
r^jin
1i
oo Ln rH
oCN O ^r 00 V£) r^ o
uauinqjY rHO•
o• OO rH
1
rH
CN
00
o rH
en• CO
I
• enO CM
l
H 00 <£>
rH1
o • CN oo oUT3q.OJcJ ro
rH•
o• oo o
1
OrH
•
O. om »
i
oo co
rH ^1
m inCN
in
oftf o en CN CO o CN O "tf
1 00TOja^ • CN M1 ^r <£> • m r» kd r^ • o . rH
-seToqo Cn o• co CN m cn
o• CO
00o ooH
in "sf
eno enCN
rH1 OO U3 r^ 1 1 1
pxov rHCNo
Cno o OO
o o po HrH CN
CN"5j<
OTjn O• in o o o m <X>
inC
o o o o o rH O ! i oM 00 00 r-- CN CTi
rH OO VOro oo V£) o CTi O • cn rH rH o <D
Nna • KD . rH • o rH O • in en r-» * CN * r*• ^D • O
LD O r» • • H UD <£) 00 <T\ en O CN
rH 1 i o o 1 1 CN ^r UD 1
m 00 r^ Cn CN 00 CN oo • r- • in . rH KD• CO CO 00 H ^ <* 00 VD r- cn \£> o CO <3" SP
esoonxo r* en IT) CN . -=J< <£> • o o> • «* rH in H CN o en CN
CN • • CN • o • CT> oo CN oo •
CN r» o in rH 1 rH ^1 1 l CTi
oo •^ 00 oo 00 r^ r- ^ •^ CTi -=r CN rH
•soqa rH in CO rg r^ CO H CN O r- C7\ • «!* CN
•6joui r-•
• CNOO
• ino
o•
o•
• rHO
o•
o • 00CTi
• CN • CT oo inrH
o 1 1 o m11 o o "Sj" rH rH i 1
CM 9• oo
10 CN «* CN COCNO
<o CN KD U3 CO • CN OO
uinTox^Coo
o ^r• CNO 00
Ho O "5*
ooo•
. ino oo
enrH
00rH CN^
r* cti
CN V£>
inCN
o 1 1 1 o l O o 1 sr O CN 1 OrH CD
COU CD u c Cd•H 4-1 0) •rH CD 4J
g C fd d) +1 C C X! £ cd
3 as x CO CO •H CD 3 •HX 4->
•H en D. CD H CD £ rH >H rH a XU ^ CO O u T3 H Cd 4-1 3 ftj -H <d CO Eh >1tJrH 3 S •H -H 4-> X! 4-> rH ^ 2 O T3-H CD
id c x H P u o X! O U rH -H h x; a u CD CJ>
u M f^ CD PQ Z) < u Eh CU rtj Eh CQ XI CO CQ S <
23
wCO«OKW
WEm
CM
#O
U EnI
> XH H
ft
Wu
>ouI
wu<H
>
1 in
!: a6vH
• CM O
J
Apoa 2636
901
rHCO
00 in ro
c^S• in r*
CO o CM
|
10DS COonCOCOCM
•
00HH
otH
• T . VO en • LOCM CM H CO • r- «* o
HOT <0 COin
tH CMon
•^r co H CTi
HCO 1 H 1
. O • ^r • en . r^ ^o•soqd in H CO iH CO o
CO 00en coO rH
• rHH O
•P>ITY r- H o Oin H
1
CO1
CM1
urqrurpxTgHCO
CO• CO
en enCO
eno CM
CM
TB?OA CM•
O00 o
1en"
00
I
• 00COH
CO
oin CO o o O o o
uauinqxv oo•o
• iHO O
1
• 00O if)
1
Hen
o
• ino ^
i
o oI
UTaq.ojrainOn
CO• o
00CO • 10 CM CO
CO CM. f>
COro
"[e^oj,H
•o o H
•oo en H on H r- o
HCO
O 1 o l i *o c"<* «3« o r» on ^T
s?• ^r H
Xojsq. • rH 00 00 CM en H H 00 CM
-sajoqD CM OCM
. U3CM
!
CM•o
• otH
1
• in
CO
• COcoH .
CM OH
1
rH r*co
l
. CO
1
pxovenCMO
VO O• O
-Co-coo
CTnH onen>0 m
• ooo en
• rH
oxan • • o • o ° • «* <x> en in o CO• o o 1 o 1
^ CO H l 1
«* ^r co ^r in CO o »0• r> «*H r- CO iH o o in 00
• CO oNna • ^r • o • o • ^r • H CM 00 1—1 on m cm en oo en
CM o CO o o • • H r^ ^r CM CMH l 1 l 1 o r- 1
CO 1 oCM r> <D r- ^r in o m o • o • ^r "=r
VO o in in Cn CO r^ • *o *tf kd in cm r» • en
asoonxD • LO CM in • r» • Cn o • r^ en co • CO r- "^ co en CO 00
in • • CO H • H en in on 00 H1
r^ «4< o ^1 o 1 CO vo l l 1
^r CO H CO CTi «tf CM CO CO CM • CO en CM
•soqd•Baoui
Cn CO CM r- "^T i> o CM CO H CO 00 • CO in
en ^ • O o CM • H o • H • Cn . CO co r~ ^r oo • ^t. o . • O • o CM •^r H 00 CM
o "^1 o "*
1 o 1<0 CO 1 1 1
oCM CM <o en ^r "* CO VO CO CO in r~- CO 00
1 CO • G\ • CM • co • o CT. in o • O • o • o • CM CO • ouinTox^o
oO O
1
O rH
1
o enI
o o1
d *I
oo
oo
O O1
H O1
en H1
CO COH
CM
en
O CO1
H CD
O CO
U CD S-l C iti
•H +J CD •H CD +J
6 C 03 CD -P C c & C 03
P rO .c CO CO •H CD P h x: M•H CT> Ci o CD rH CD e H M h a £U >-l CO U TJ H rd 4-> p 03 -H n3 co rH >i CT
rH O P 2 -H -H -P £1 4J iH M K O T3 -H oid c x; H D M O x: U H -H H £ Q u CD CT>
u H cu U pq D < u Eh ft < & CQ <C eu J en CQ S <
24
Eh
<Ooro
Ohi
XQ «1 En
>M si
w W.-3 UCQ 2< <H H
«<>ouwu<
>
CTi
e6v CO
•
ro
o r-^qBxeM rH 00
Apog •
in
in•
CN
in-^ CTi X)
L0S6
orH
mCO
CNrH
CN•
VD1
* 0000 CO • CT\ ro CN
hot in kd oo ro . CO • CNr» ro in CN rO(Ti rH^ -^ <* CTi
H 1 1
KO r- • r-> r» rH• »x> . ro ro iH • rH • <J\
*soq<3 cn ro VD CO <J\ CN o in X) VC
•B3(I^inCT
in rH1 l
CN1
ueuinqxvrooo
<T
oo l
CNCO[*-
rH
rHHCN
OrH
• rHo
m cr 00 o O ro r^UTaq.ojci CN
rHO CT' o «3 in ro
in in
I^^Oi •
O•
oCO
1
(N1
•
ro o•
oCN ° CTi o rH <£> .<""> 00 CO
TOia^ ^ U5 • CO m . «3< • O roro CN rH
-saxoqo ro vx>
HO CN
1
in
rH2 N
1
CO CO • iX)
rH
. inCT.
I
pxov CNOin rH
rH • o in
CN ro•in
"ST
O x>oopn •
o CN °1
o•
oo o
1 O• 00
rH1
OfOl
• rHO1
o•
o»x> 00 m o CN P- rH CN <X> [*-
CO CO cr> <X> CN • o • ro X) 00 X)
Nna • o m r» ** • O O CN O ^r • ro o 00LD • • • O ro ro o • •
CN o CN o 1 1 l 1 o oro <£> oo • rH r* V£> >X> O CN <* CN ro
• «tf <£> CO <Ti in CN ^D CN UD ro • k> r- cnasoonxo r- oo • <T> ^ *£> CTi ro rH • rH ^ in ro o • r- • o
sr rH • CN • • CO rH rH CN ro
r- 1 o 1 rH .rH 1 1 H rH CN
LDCN rH » rH rH ro o CN ^ O m
•SOLld CN 00 CT o vo KD ro CT <y\ <* rH X)
•6joui • • X> • LO • o • CN • rH • O • O • r—
1
« CN . rH » orH in rH o rH O O in ^ ro r-i rH
1 1 1 rH 1 1 rH in r-\ 1 1
as ro CT. o CN in rH ro o m in x>IT) •^ • in *£> CN rH r- o ,H CN rH in CN
uiriTOXED X)•
LD•
o .-h
rHCO
o• OO 00
* rHoo
CN ^• rH CN • ino
• mo cr>
O O 1 1 1 rH 1 o rH <* ro I l
rH CD
CO
U CD u rrj
H 4-> CD 0) 4-)
6 G rrj CD +J C C c td
P <tf £ CO CO •H CD •h .c 4->
•H tr> ft o CD rH CD B rH ft -CO U CO CJ U TJ rH CO -P rd co fH >i CT
rH o 2 -rH -H 4-) 43 M o EC o XI -H CD
fO C X! rH D ,M U xi 5h <-\ rH X* Q CJ CD CnH fn o CQ D < u t-< dt < < a. J w PQ S <
25
CO
a6voVD
•
ro
^qBxaM 00ro
voCN
Apoa CO•
VD•
CN
CN rHrH ro CN
• ro ^ 00
10DS<*rH
CN•
cnCN
•
ro1
• r-Cn CN ro ^r • *r
HOT 00 CN • rH •rH ro vo
CO VD r~ in CTi in HOEh Cn r^ VO rH
< rH I 1 !
Uin • iHI • VD in ro
Wh|
in
•solidro cn
r^ iHa> in
CN 00ro ro
•CTi00 rH
• roVOCTi
•Bi(IV rHCO
inrH
rH1 i
ro1
vo 00. o
rH <*
roo in
rH %H usuinqxv o • rH•
ro•
rH•
OSO
o 1 1 in o o
uxaq-O^d oCN
rHroo
VOro
00 °°0000
3in
rHin
w1 H
OS
T^oi •
o•
o00
I
rH ^1
• •
o•
o> #
M Eh CO vo ro o o • 00 • rH vo ojoaaq. CO ^ CTi m • en rH CN o ^r r~ • <T\
S -ssioqo CN ^rH
Oo
HCN
cn ^j.VO ^
1
•^r oCN
VO CTi 00*^H
oo ro
1
CQ w< u CO ro rH CN VD 00 ro vo r^
Eh !2 pxov CO . VD • o O r- r- ro CN o< o CN ** o o ^ ° r—
)
• o • r- • rH • oOTjrn o 1
1
O1 rH
<* o o o1
<l 1 1
si" r» en Cn CO CO • CN o ro 00
> CN ^ • rH CN in • ^r r- r- ro 00
o NQ9 . iH ro r~ cn O • rH co cn cn r> SrH CN r-
u CN • in • O rH rH ro • •
rH o I o • |l 1 1 o o. VO O rH • -^ PO VO . CO ^D 0° rH rH VD
2<H
in ir O H CO rH cn H r- cn ^r co . r-» r^ 00
asoonxo vo ^ VD r- r-» m r^ m 00 CO CN OO CTi ^T • <T\ . in
o • • rH • • rH CN ro rH in
<>
i-H VO o 1 rH rH 1 1 rH CN ro
roVC
CO r- CN VO 00 CN rH o in o•soqd ro • "=? CN CN •CO ro rH CN O CTi vo r-
•Bjcouicn VO . ro • o • o • rH • O . ro • rH • CN • cn • o
•
rH rH rH o CO o o VO O rH o rH
1 1 1 rH 1 1 H CO rH l 1
-3" ^ r~ o CO rH VO CN o r-1 O a- cn<N ^ • r-
r^-o 00 ^r CN CO rH r« CN CT\
uinxox^D 00• •
m r-
rH• CN
rH• oo
• roro
• rHo
• OO «3*
CN• CO 00 o M
- o o l 1 1rH 1 1 rH m ro I 1
rH 0)
CO
U QJ u CO
•H -P 0) <D 4->
g c fd CD -p c C C rC
3 rd £ w en r— 0) •h xz 4- J
•H cn - QJ -H a g rH D Xu M en o u t: S rH a 4- 3 rd cn E^ >i t 1
iH C 3 2 •H -r- \ -P c A X X O T3 -r a)
CCi C X -H O 5-1 t_> x: o »- rH rH £ Q O o a ; cn
u h p. o CQ D <t : u E-" 0. < < cu Hi cn CQ S : <1 '
26
<Uw
WEli
iniH
OP4
ftI
>H Pi
X
u<H
<>Ouw
*c
<>
in
a6v Eh•
^q6T3M inCTi
cn
Apog•
KD
in•
CN
m in KDvo o rH
1,00S o•
KDcnKD
ro•
in1
•
OH
1
• cn in iH • rH
Hen t*- o•^ HCO
• r-CO KDin
• 00CO o-**
1
<=r hO CNH
1
• -^ CO• CN KD o
•soqa H 00• CNocoCO
oo ^r • 00 • ^r
BXIV rr ooH
oo a\CN
cn ^H
H •^1 1 l
•^ H 00 m ^ oCN r- •CO • in CN o
ueuinqxv O•
Oo
•
CN1
H rrl
H•
OrH
•
o>^> r- CO CN
•rHror^
3 OCO 00UTS^OJcJ
CMoo
• inr*» cn
KD•
[-> KD
T^h-oj, O o i
HI
"sf o O• in CO Cn KD o ^
xoaaq. o ro 00 HCN ^
• inr-^ <7\ O
LD 00 ro H • H • "vT • 'cr-sajoqo ro CN• r- -I C7M>- CO in «*
H H 1
KDCTi H H
pxovCTirH Cn KD
H o >00 Cn
CO H CO •
CN
oT^nOo rH
1
O•
O• oo
1
• oo
1 rH
H•
o• oo
l
o•
oCOro
00in CN
00to 3 CO
* CNrHCO
m H KDcn
jsina • COCn
<*•
• iHro •
H•
<^ COCN
• oCA H
• COo o
en•
ro o KD o o 1 KD CN l o
a LD CN • tr CN ro 00 • CO CA cn CO
• 00 • o 00 m o Is* 00 • CN H rH • iH H ro
asoonjo H C O LD H ^r o o CO in co CN CN H CT\ CO • rH
p* rH • CO • • 00 KD H • co
*r 1 o l o o H 1 H isD H
in <Tl O CTi CO Cn o ro KD H CO in
•soqd LP) 00 CO H H ro in r- in (T\ in cn
•Bjoui ro <* • cn o in • H • o • cn • in ' CT\ • ^r • iH
• • rH • o o CO H "31 H HH LD 1 o CM 1 I H CN H 1 1
LO *>D £ *£> cn o ^D en KD 00 CN r* •^r
uinxox^O U5 rHCO
—
i
• ro • inCN . iH
H 10• ro CNo • Cn • CO CT\ • r* • ^r
• • ^ o oo O H • H KD h r- CO om o en- o O 1 I 1
H 1 o H CN H i 1
_j CD
CO
U CD S-l CO
•H +J o CD 4->
e c rC 0) -p c C C cd
3 fd X! CO CO •H CD H x! 4-1
•H Cn a CD H CD g h a XiU >H CO u O t3 H rcj -P 3 crj co Eh >i tr
rH p 2; •H -H o -P £1 ^ K O TD -H Vid C -C rH D 5-4 U Xi U H H Xi a O CD tr>
cj H Cu o CQ D < u Eh fH < < a. ^ c/l CQ S <
27
The primary purpose of computing the sample variance-
covariance matrix is in computing the Pearson Product
Moment Correlation Coefficient, r. . . This statistic is a3 /*
measure of the tendency of 2 variables to vary together in
linear fashion. It is described as the ratio of the sample
covariance between 2 variables to the square root of the
product of the sample variances of the 2 variables; that is,
s . ,
3 /kr . . =D/k
V (s:,:
)(s*,k>
/l4\For the equine data there are _ I =91 pair-wise correla-
tion coefficients to be computed and for the feline data
there are I „ 1 = 78 pair-wise correlation coefficients to
compute and consider.
The tests of hypotheses conducted in this research
effort rely on certain assumptions concerning the sample
statistics. The central limit theorem, which permits us to
a2
assume a normal (y, — ) distribution of sample means in the
univariate case [11] , also permits us to make the assumption
of normality in the multivariate case; namely, the distri-
bution of the sample mean vectors, x, is normal with mean y_
and variance-covariance matrix -=- Z [12] . This assumption
will be employed in testing the notion that the vector of
sample means of the males of a given species does not differ
significantly from that of females of the same species.
Additionally, if the population correlation coefficient,
p. , , between 2 variables equals zero, then the sample3 r&
28
correlation coefficient, r. , is distributed as3 r*.
t(N-2) v/(l- r* )
/N-2
where "t,„2 v represents the Student "t" distribution with
N-2 degrees of freedom [13] . Equivalently , the statistic
r. , /N-23' k
is distributed according to the Student "t" distribution
with N-2 degrees of freedom, provided that the hypothesis
that p . , = is true.3 /-K
29
VI. HYPOTHESES TESTED
A "null" hypothesis, denoted as H , can be defined as a
statement or notion for which it is possible to compute a
statistic and the corresponding probability of a more extreme
value of that statistic. Alternatively, it is common to
establish a probability corresponding to the risk of
rejecting H when, in fact, it is true and denoting this
probability as a. H is then rejected only when the test
statistic is more extreme than that associated with a.
In this study the following null hypotheses were tested:
1. H : The true correlation coefficient (p . . ) betweeno H D,k'
variables j and k is equal to zero. This hypothesis was
tested for both equine and feline data irrespective of the
sex of the animal.
Since the sample correlation coefficient can be converted
into a "t,„ ox " statistic, providing H is true, it is(N-2) e o
possible to solve for the "critical" value of r . , by3 / K
establishing a type I error (a) and locating the appropriate
"t., „ " value from a statistical table. In fact, this(N-2)
critical value of r. . , denoted r . , , is obtained asJ , K CCJ ,K
follows
:
ta
raj,k r~2
V< + (N " 2)
where t is the 1-a percentile point of the "t" distribu-
tion with N-2 degrees of freedom. Conveniently, values of
30
r . kare already tabulated [14]. Those appropriate to the
testing of this null hypothesis are presented in Table V.
TABLE V
CRITICAL VALUES OF r . , FOR SPECIFIED aD_rk
N a = 0.05 a= 0.01
44 (Horses) 0.298 0.385
30 (Cats) 0.367 0.470
Triangular matrices of sample correlation coefficients
for the 44 horses and 30 cats are presented in Tables VI-A
and VI-B respectively. In these tables those values of r . ,
3 / K
for which H : p. , = can be rejected at a = 0.05 areo H j,k J
indicated by a* and those for which the hypothesis can be
rejected at a = 0.01 are indicated with **.
31
qqBxaM r-
Apoe CM•
oLD "^LD H
LODS O•
OO
•
o1
CO LD CMrH 8 CO
HCTI iH•
oo
1
•
o1
CO H CO H•soqd CM
CDo•O
CO * LO *
•B2[T\fCM
•
<X> *•o
LO *•
oO 11 1
1
O o CO CO ro
H o>rH
roo CM OiirqriaLLLS
ieq.o.1•o
1
O1
•
o O•
oCO l> o r^ CTi HO
CO
(Ti
w CM * CMH
LD O CMCO uauinqxv r- *
•
H•
rHd
HO o o o O i OK i
"
CM CO CO CM H LD«3< ^ m "31 5* LD ID CO^ UTSqOJTd CM CO* CO *
•
ro *•
O•
CO *•
CTi *CM
1
T^qOJ, O1
o o1
o1
O1
O •
oH X H
i—
i
LO o o> H O CM o o rH o -31
« XOjraq CM ro CO <J\ r^ H CM CM
W E-<
CQ S-saxoqo
•
O1
Ho
o•
oCM
•
Ooo
•
O1
•
o1
•
o1
< ^ <Ts o rHEh 2 CM «X> r~- CO a\ LO r^ LD CO
O PTDV CM CM lo* o KD H CO CO * CMHEh
pt^a ro *O
^r*o
CM•
CM o •
o•
o< o 1 o
1 O O o1
1
J r- CM rH CO o CMW o ro CO rH CTi H LO CM CM en«« Nne CM H O O t-\ H CM
OHH
LDO
Ho
o O o O O O O • • • •
u 1 l 1 1 1 1 o O o o\D rH r-| CM K£>
VD ro ro H ro 5* r^ [*•» ro ro CTi
ssoonxD co 00 o rH o <tf H rH H CM OH rH CM O . o CM • • • •
• • • • o • • o o o oo o o o 1 o O 1 1 1 I
"*oO CO o en
CT>o H
CO o «* LO,—
1
r_t *LO *
CO *•soqd rH rH r* ro <£* 5* o i> * o* H LO *
•Baouio • co* ro* .* H r- * "tf * # •
• O • • o • o • O o oo 1 o o 1 o
I o o 1 1 1
uiriTox^D
COHi-l
•
COrHO
•CM
roroo
oo rH
CMinCO*
LOOCO * ro
O
ro
3CM
O oCM
o1
o1
O1
•
oo
1
•
o•
o o1 o
o1
•
od
I
o1
rH <v 4J
O CO du a u c fd tP•H 4- <D •H a) 4-) •H
e G fc 0) +J G C XJ C <d (U
3 id x to co •H CD S H jC £•H CT> c . a; H <L> 6 H M h a,u !h Cf U U TS H fd 4-> 3 03 -H id to ^> >1rH c ) P 2; •H -H -M O £! 4-> H .* M O T3 cy
id c X : -h D U U 43 U H -H m x: a O Cnu H P- u CQ D < U Eh Cm < Eh CQ < Cm rH" CO CQ <
LOODCO
OAM
OoII
3
fd
u-HLH•HGt7>
•HCO
I
CO
IN
oA||
ino
II
a
4->
fd
4->
Ctd
o-H<4H
•HCtj>
•HCO
I
32
'
o^q6xaM rH
*° *r-
Apoe•
O•
o
CO CTi A||
V£> CTv
I.ODS rH•
oO
•
O1
XV
CN 9u
HOTrHrH
•
O
rH•
o1
«
o1
rHO
•
•soqj in
rH3^ *
OII
•^XTV rH
O O1
•*
O*1
oi
CNo
<Ti CO COtd
uauinqxv rHO CN
• CN00 * CO
COPc
enEh
<U
• o • • • ccj
o 1 o o o o
UTeq-o^d o in *
CN 00n CO
"3*
rH *
•H
CN • * • CN ^r * in * Co T^oi, •
oo o
i
•
O•
o•
oCn
1
I
X loja^ 3oon 8o
CNCN
CO(/)
>Eh
-saxoqo oCN
orH
•
co*•
CN•
•
o1
O 1 O o o 1
IH*
rnpxov in
co co
oCNrH
CN in
800
OCN
CN
COrH r^
< 2OM
W53ou
oxjrn oI
rH•
o•
O1
O•
o•
o1
O1
*
O1
OO
CO•
Nfia
CN00 *^r *
inrHo
00r*rH
CNO•
masrH
•
oino
•
CNOO oo
CTi
COo
o
aJ!
• • • o o o O 1 • Mo o o 1 1 l 1 o o
asooriTD rHoorH CN CI
CNCO
9oo 00
CN cr> COU
rH O CN • * H rH ^ -a
O • O • • o O • • •
1 o 1 O o 1 1 O o o ino
•
o•soqd 00rH
rH00CN
Oo
CN 00
CNmrH
00 in oirH*
CO *
•5J.OUJ •
O1
•
o1
•
o1
CN«
Oo
1
•
o1
•
om *
•
o•
o• *o
1 oII
SCN
COin 3 rH 00 00 o o r^ CN © CN
uinTox^D in* • *CN
•rH
•
in<* *
CNo
00 * CN CN ^* -Pid
• o o o « o • • • • O oo l 1 1 o 1 o o o o 1 1
4->
cidrH o -P
O V) x: oU 0) U fd Cn •H•H-P CD CD 4-> •H IH
g CfO cd -P c G C 05 CD •H3 (T5.C to W •H a) •h x: "3 c•H CnD o CD rH a B rH a CnO >-i en o O Ti rH ccj +- 3 id w EH >i •HiH 00 p "2 •H -H -P c J3 M K O T3 CD en
id cc rH D H U & rJ rH rH J3 Q u O tT>iu H&< O CQ D < u E-» a < < CU rJ 03 CQ <*
33
m2. H : The true coefficient of correlation, p . . ,
between variables j and k for the males is zero. This
hypothesis is tested separately for horses and cats.
The critical values of r . , are presented in Table VII.3 /K
TABLE VII
CRITICAL VALUE OF r. . FOR THE SPECIFIED a3>k
N a = 0.05
0.413
0.514
a = 0.01
23 (Male Horses)
15 (Male Cats)
0.526
0.641
Triangular matrices of sample correlation coefficients for
the 23 male horses and 15 male cats are presented in
Tables VIII-A and VIII-B respectively. Significant values
are indicated as previously described.
34
^q6T3MApOQ
enoo
10DSCO
od
CN
-JoI
Hen
COVOm
m
o
rHOr-\
O1
dI
•soqd*^TV
oCO
•
o
00
o•o
1
d*1
in
i
rH•o
1
ocnr~A
•oi
oo
•
o1
o"^CN
•
o
ino
•
oCO
ULqnj.LL.LU
wCO
ouauinqxv
in
00 *
o1
CN•
O
mrH
•
ooo
CNO
•
oI
COo•
o1
I-A-
MALE UTsq.oj;a;
T^q-oi,
enoCN
•
o1
eninCN
•
o
CN
^.*O
1
cn00CN
o1
rH
•
O1
•
o
cnoin *
o
LE
VII
ATRIX -saioqoCOo
•
o
CNrooo
I
oCN
•
O
cnCNm *
o
00cno
•
o
inCNCN
•
O1
CN
o•
o1
CN
. *o
1
PQ SEh 2OMEh
PTOV ooro
o
rHCN
o ,
CMoro
o
mCNoo
ininrH
o
orH
o
CN^>rH
•O1
r-H
Oro
o1
ro
o<
PSo
wna ft•o
i
U3CNrH
O1
LDr-o
•
o
COrHCN
•
o
r-mcn
dI
r-\
CN
O1
CN
•
O
mo
•
o
orH
O
rHo
•
o1
u
asoonxDr-<NrH
•
O
CMrorH
*
oCN
•
O
Oro
o
orooo
1
ro
•
o
oo
•
o
mencn
oI
mod
I
CO
•
o1
inenrH
•
o
•soqj•Bjloui
LDCN
•
O1
•o1
enCNCN
•
o
rH*?cn
*
o
o
• *O
1
co
o
inCNo
•
o
CO
•
o
cninm
•
o
i OO
•
O
00
in *• *o
1
rH•~\ *
O1
uinxox^D
IT)
rooo
I
oo
l
CNrH
•
o1
CXI
rH
o
rHino
•
ol
Cn^-{
<-\
o
CN
in *•
o
CNr-
• *o
i
CO
oo
moo
•
o
r-H
O
o
CNKDCN
d1
CO
Pj
o
g
•r-i
urHrd
U
U <L
•H 4-
C re
rd XCn CH cr
C
H P.
CD
CO
- OUPrHO
DPQ
•rH -r
D <
uCD
-MCO
Q)
O> x:: u
cr-
t-H ard 4-
4-> cO r-
Eh p.
CU
e
rH
<
c•H£1
rH Ufd-H4->rH0-Hir> CQ
Alkaline
Phosphatase
Q
Eh
OOCO
-Mx:
•HCU
>1Ti
m
CD
<
CNin
oA||
ooI!
«
-Prj
4J
fd
-Hl+H
•H
•HCO
rorH
oA||
mo
-Prd
4->
Crd
U•H
•HcCn•HCO
I
35
COEh
<Uw
CQI I
HH XH M
w
Eh
Eh
12OHEh
<WPi«OU
Apog
CN<£> *
m
o
•LODSoinco
•
o
COino
•oI
HOIHO
•
o1
inCMCM
•
o1
CM
•
O1
•soqa en
CO•o
CMCMH
•
O1
00
• *o
1
CM00
•*o
1
uauinqxvHHCO
•
o1
COCMCM
•
O1
HCO>X> *
•
o
<vO *
O
CO<£>
CO«
o
uxeq.oj:dHCMCO
•
o
o
• *O
1
CO<y\
CM
o1
CM00CM
•
O
•Jj1
Hin *
•
o
CMCNin *
•
o
lojtat).
-saxoqoooo
00r-CM
o
00
g•
o1
00
oo
r-inm *
o
00roCM
•
o
COCNH
•
o1
pxo\[r-
ro•
oI
HO•
o1
ooo
1
CMCM
•
o
en
•
o1
oH•
oI
HCO
•
o1
HCMO
•
o1
ma00
LD *•
O
CO<£>
•
o1
00Ho
•
o
PiCM
•
o1
rH
•
O1
CMvoCM
•
o1
CMCM
o1
oCOo
•
o
CT\
HrH
O
ssoon"[D
00ino
•
o
orH
•
O
00
rH•
o1
HCMHO
HCM
•
O
CMoCM
O1
ocr>
•
ol
CMHH
•
O
oCO
•
o
r-m *
•
o
•soqd* bJLOUJ
mLO
•
ol
"5J1
no
oI
CNro
•
o1
in
•
o
CM
O1
CO
o•
o1
Hin*
•
o
Hin *
•
oCM
O
mCO
ol
00
8•
o1
uiriTDx^Dco *
•
o
CMro
• *o
I
CMO
•
o1
00"ST
oo
1
rHH
o
COinCO
•
oI
r-\HHo
1
o00
•
o O
CMHHO
CMVOCO
•
o1
00CO
•
o
e
•HUHfd
U
O Q)
•H-PC rd
(t)£CT>C1>H CO
ccH Cu
OJ
CO
oU
rH
UDCQ
TJ•H -H
D <
H5H
d)
-PCO
0)
rH
u
CHfH CD
rd -P-P o
UEh O,
(D
£
H
OJ
CO
03
OJ -Pc fd
•H X-Hrd to
X O
< tu
ECQ
Eh
OUCO
-p
C7>
•HOJ
>1
CQ
OJ
rt!
oAll
M•r-i
OII
3
-Prd
-Pcfd
oHMH•H
rj>
-HCQ
I
*
Hin
oa ii
mo
•
oII
a
-prd
-PCrd
O•Hm-rH
c
•Hco
i
36
3. H : The true correlation coefficient, p. ., betweenO ] fK
variables j and k for females is equal to zero. This
hypothesis is tested separately for horses and cats. The
critical values of r . , are specified in Table IX.3-f ^
TABLE IX
CRITICAL VALUE OF r. . FOR THE SPECIFIED aLdS
N a = 0.05 a =0.01
21 (Female Horses) 0.433 0.549
15 (Female Cats) 0.514 0.641
Triangular matrices of sample correlation coefficients for
the 21 female horses and 15 female cats are presented in
Tables X-A and X-B. Significance is indicated as pre-
viously specified.
37
ro CT\
^q6T3M r^ •^T
Apoa •
oin
o
o tHin
A||
I09S iH•
Oo
•
oCN rH CD UCN rH ^
hot OO
1
O•
oro
•
o1
r-i
OIT) 00 ^ CN
•soqaCN
•
O
orH
•
O1
VD*VD*
•
O1
CO *^*o
i
OII
PI 00 CTi ro ^r
ro CO O in -Purqnxr[Te • CN CN CN rH d
CO ieq.o.1 o1
•
o •
o•
o•
o 4Jw 1
r:CO o rH o^ ^r in CN <3ct;
CD rH in ro CD ^ uo ueumqxv ro
oiH
OrH
oHo
oo
CN
O•H
•HW1 ! 1 1 !^
^ CO roCNVD r» CO "* -H
IT)
CN
O<-i *CD *
•
CN•
oro
•
oCNO
•
roCN
00rH
•
10
!
1
X 1
W X
1 O 1 1 o o o
XO^aq.oCTi
ro m 3CDin
rorH
CTs
,H
CN
CN
roCNrH
t:
m r4 -saioqo oCN •
Oo
•
o •
o O O .„
TA
ON
MAT
1 o 1 o o 1 1 1
rorop-cov ro
00roin *
cnoCNCN
CN00
roin
CDVO
CO
ro
rH
oxjin cn • CD * • ro CN rH • « •
• o • O •
?• o O oM
Eh
<W
ou
o l o 1 o O l 1
Nna
LDCN
o3
U3CDCN
rHCN
rorH
oro
CN
OCDO
rHmo
roCD
A||
• • • rH O • CN • o r^O o O o • • o • o • ».
1 1 1 1 o o 1 O l o •r—
o 00 "vfVDO r*> o -=r
>>rH
oCTi ,H
U
ro r> CN LD in ro CN * CTi CN in* LO*~"
ssoonjo H ro ro • * o • CD * o • .* •*• • • O • o • • O o O ino o o 1 o 1 o o 1 1 1 o
•soqd rH
vooo00
ro
00CN
rH
CN
CNID *
inCN
coVOCN m*
rH
9*VD *
•
o
•Baoui ID *•
Oo
1
•
oCN
•
Oo
1
O•
oO
1
00 *
O•
oO
1
oI
• *O
1
II
ro VD o ro CO CN "^ CO cn. VD CN r^ ^ IT) o CN CN CN 9
•
CD o "Bf -p
uinTDX^o rH O ^r * o O rHCN
COrH
O•
O• O
orH
ro rd
o1
o1
ol
o1
o1 O O
O1
o1
o1 O o
O1
Prd
rH cu -P uCO x: •H
u o M C rd en 4H•H-P cu •H CU -P •H •H
g c fd cu -p c c .Q C rd 0) G3 rd,C CO CO •H cu P •h x: 5 Cn•H enp CD rH 0) £ H U -H CL •HU >H CO u U TS rH rd -P rd -H rd CO ^ >1 CO
rH a 2 -H -H O -P -Q P rH X EC O ^ CU
03 ex: iH D Sh U x: Sh rH O -H rH x: Q O tn 1
U M CU o CQ D <C u Eh Ch < Eh CQ < cu »-Q CO CQ <*
38
I
XwCQ
<EH
CO
<Uwr-q
p4
XM
2OHEh
<
2«ou
! H\q6xeMApoa
COCTi
•
o
IODS
CTt
r*o•o
1
COCTvrH
•
o1
HalinoCO
•
o
rHCOiH
•
O!
in
"I*o
1
•soqd•
oi
"St1
CNco
o1
rH-ST
•
O1
CNCOCN
•
o1
UBiunqxvC^CO
o
oin
oI
mCO
oi
inrHco
o
r-rHCO
o
uxsq.ojd rH
O•
o
o•
o1
CN•
o1
COCOCO
•
o
in
in *•
o
COCO
•
o
-saxoqoCNrH
O
r-
rH•
O
COino
•
ol
COoCN
•
o
rHorH
o
in
rH
o
CNcr»Ho
I
PTOV cx>
CN•
oI
rHCN
•
O
00
r»CN
•
o1
rHo•
o1
00COrH
•
oo
•
o
rHoCN
O1
00cr>
o•
o
NnaroCNin *
•
o
CNr-CN
•
o
COCOCN
•
o
inrHrH
•
o
oCN
O1
OrH
O
inr-rH
•
o
moo
•
o1
o•
o
asoonxor-o
•
oI
o<£>
O•
o
CNCO
ol
OOO
CN
O
CNinCN
o
oCO
O
inCTi
rH
o
rHrH
O
C7\
CN
O
•soqd•Baoui
r»rHCN
•
Oo
1
oCNrH
•
O
CTi
o•
o
00CNCN
•
oI
r-
CN
oI
minco
•
o
CO
rH•
O
00CO
o
rHCXl
oI
CNOmo
I
uinxox^D00
ro•
o
CN
CN
O1
CNCNrH
•
O1
oro
o1
o
•
o
oCOco
•
o1
COr-CN
•
o
COoin
•
o
rH"Si"
CO•
o
r-oo
CNCOCO
O1
COr- ..
• *O
1
B
•HUHu
ocu•H-PGfdrdx;ct>q5-^00ex
ca
PrH
u2D
O T3H -H5-1 OD <
rHO5-(
CD
-Pto
CD
rHOX!U
Total
Protein
CCD
B
XIrH
<
Alkaline
Phosphatase
QEh
OCO
-Px•HCD
>1
cq
CD
CT«
<
OA||
I!
a-p
«a
*j
crrj
o•H4-1
•HC
-Hen
rHL0
OAll
Xn
5-1
ino
-Pcd
•PCfd
o•H
•HCtj>
•HCO
I
39
4. H mj.k pja = P j/k
This hypothesis states that the sample correlation
coefficients between variables j and k for males and
females, considered as separate entities, are actually
estimating a common population correlation coefficient,
p. . [15]. Alternatively, the hypothesis states thatJ * K
P- v " P- u= 0. Note that there is no assumption that
p j,k = °-
To test this hypothesis Fisher's Z transformation is
employed where
.m
j,k \ in1 + r
mj.k
1 - rmj»k
and
j,k I-1 + r
j rk
1 - rj,k
It will be noted that each Z . , is the inverse hyperbolic3 /K
tangent of rjrk
[16]. If H is true, then
1/k i ,k
vNm~ 3
+
Nm-3 N
f-3
has a standard normal distribution [17] . In the equine data,
thereforeNm = 23 and N
f = 21. The quantity
becomes
/-±_ + i
"VNm-3 N
r-3
./•^r + ^r - /0. 10555. . . - 0.325. To reject H atV 20 18 J o
40
the a = 0.05 level (|z. . - Z. ,| ) , must be greater than
or equal to (1. 96) (0 . 325) = 0.637. To reject H at the
a = 0.01 level this absolute difference must be greater than
or equal to (2.575) (0.325) = 0.837. It should be recalled
that 1.96 and 2.575 represent the (1-a) =0.95 and
(1-a) = 0.99 percentile points of the standard normal
distribution, respectively.
In the equine data, statistically significant differences
between male and female sample correlation coefficients are
presented in Table XI.
TABLE XI
Paired Variables rm
. Z™ . rf
, Z? . I Zm
. - Z? . I
SSS^ri^-I "°- 254 -°' 260 °' 515 °' 570 °' 830 *phate & Glucose
Glucose and Age 0.195 -0.198 -0.517 -0.572 0.770*
Tota/protein °* 216 °- 220 "°- 534 "0-596 0.816*
TotalS
protein °* 301 * 311 -°' 506 "0-557 0.868**
* - significant at a = 0.05** - significant at a = 0.01
In the feline data there were no statistically signifi-
cant differences between any of the male and female corre-
lation coefficients. Therefore, H cannot be rejected ino J
the case of the cat, at a = 0.05.
41
5. H : y - y =0; that is, the samole mean vectoro — —
for males and the sample rear, vector for females car.e frcr.
the same parent population with mean vector u. To test
2the validity of H , Ma:.= l = r.:c:3' D statistic is computed
as follows
:
Um - xf
]
Ts'
1^ - xf
] = D2
— — — — _where [x — x ] is the trar.sccse cf the 14 by 1 vector o:
cr s a z z—— a z z ~ z z z z — z — ~ z a b s "*e s **
~""" e z a 1 a a *~ c f e— a 1 —
vectors of sample r.ear.s , S is the inverse cf the ceded
variance-covariance matrix S,
where S = (/-DS 111
+ (Nf-l)S
f
n™ + h* - 2
arc.r.S"
-
is the variar.ce-ccvariar.ee matrix fcr male data
S~ is the variar.ce-ccvariar.ee rr.atrix fcr ferale fata
w is the male sample size
N~ is the female sample size.
2If H is true, then D' is distributee"
z ::"-::- .:r-vr -2)~- —
in acczrcar.zs v;izh the T cistributicr. chat has ~ >"-" -::~-p-l)
degrees of freezzr. [IS]. The results cf this hype thesis
test are presented in Table XII.
42
TABLE XII
DISCRIMINANT ANALYSIS BETWEEN MALES AND FEMALES
Species ComputedD2
Degrees ofFreedom
ComputedF Value
F ata = 0.05
Equine
Feline
2.537
5.748
(14,29)
(13,16)
1.374
1.895
2.10
2.40
Thus, for each species, the computed F statistic is not
significant at a = 0.05 and H cannot be rejected at that
level of significance.
43
VII. TOLERANCE LIMITS
Upper and lower tolerance levels based on the observed
data are determined as follows: each x. . is estimating a
2population mean, y . ; each s. is estimating a population
2 2variance, a., u . and a. are fixed but unknown parameters.3 3 3
r
2The statistics x. . and s. are random variables. It is3 3
possible to determine a constant, K, such that in a large
series of samples from a normal distribution, a fixed pro-
portion, y, of the intervals x. . ± K Is . will include
100 (l-a )% or more of the distribution. Thus, statistical
tolerance limits for a normal random variable, X. ., are
given by the following:
- K Is2. and U = x. . + K Is
2
«V ] 3 <*v :
These limits have the property that the probability that
the interval includes at least a specified proportion (1-a)
of the distribution is equal to a preassigned value a [19]
In this paper, values of 0.95 and 0.05 were assigned to y
and a respectively. The following K. _,. values such that
the probability is 0.95 that at least 95% of the distribu-
n~tion will be included between x. . ± K- ._ Is. arej 0.05V ]
presented:
44
NK0.05
44 2.415
38 2.464
30 2.549
24 2.651
21 2.723
These values are employed in constructing the tolerance
limits for the biochemical components of each species.
These limits are presented in Tables XIII-A and XIII-B.
45
TABLE XI I I-
A
TOLERANCE LIMITS IN THE HORSE
BiochemicalComponent N x. .
3S.J
K *05 x. . ±
J: K.
Q5S
Calcium 44 11.78 0.61 2.415 10.31 - 13.25
InorganicPhosphate
44 3.62 0.92 2.415 1.40 5.84
Glucose 44 86.77 12.47 2.415 56.65 - 116.89
BUN 44 14.55 3.72 2.415 5.57 - 23.53
Uric Acid ' 44 0.45 0.15 2.415 0.09 0.81
Cholesterol 44 112.70 15.31 2.415 75.73 - 149.67
TotalProtein 44 6.49 0.40 2.415 5.52 7.46
Albumen 44 0.48 0.09 2.415 0.26 0.70
TotalBilirubin 44 1.40 0.90 2.415 3.57
AlkalinePhosphatase 44 145.39 77.39 2.415 - 332.29
LDH 44 223.57 67.51 2.415 60.53 - 386.61
SGOT 44 786.36 433.77 _ _ — —
38 636 05 127.80 2.464 321.15 - 950.95
46
TABLE XIII-B
TOLERANCE LIMITS IN THE CAT
Variable N X S K *05 KS x ± KS
Calcium 30 9.27 0.81 2.549 2.06 7.21 - 11.33
InorganicPhosphate 30 7.23 1.13 2.549 2.88 4.35 - 10.11
Glucose 30 78.53 27.35 2.549 69.72 8.81 - 148.25
BUN 30 25.50 5.01 2.549 12.77 12.73 - 38.27
Uric Acid 30 0.90 0.16 2.549 0.41 0.49 - 1.31
Choles-terol
302421
79.6091.7997.33
36.6429.9127.76
2.5492.6512.723
93.4079.2975.59
12.50 -
21.74 -
173.00171.08172.92
TotalProtein 30 6.78 0.52 2.549 1.33 5.45 - 8.11
Albumen 30 1.53 0.18 2.549 0.46 1.07 - 1.99
AlkalinePhos
.
30 60.67 30.86 2.549 78.66 139.33
LDH 30 271.03 121.49 2.549 309.68 580.71
SGOT 30 86.37 32.35 2.549 82.46 3.91 - 168.83
47
VIII. SUMMARY
-
A summary of the significant findings of this study will
be presented according to species. In this summary,
"significant," as applied to sample correlation coefficients,
refers to an ability to reject H: p • , = at a ! 0.05.2 J o K j,k
A. EQUINE DATA
In the case of the data pertaining to the 44 horses,
significant positive correlation coefficients were obtained
for the 2 3 males in the following pair-wise observa ions:
Calcium and Albumen
Inorganic Phosphate and Alkaline Phosphatase
Cholesterol and Alkaline Phosphatase
Total Protein and Body Weight
Total Protein and Age
Significant negative correlation coefficients for the 23
male horses were obtained in the following paired obser-
vations :
Calcium and Total Bilirubin
Inorganic Phosphate and Total Protein
Inorganic Phosphate and Body Weight
Inorganic Phosphate and Age
Cholesterol and Age
Total Protein and Alkaline Phosphatase
Albumen and Total Bilirubin
48
Alkaline Phosphatase and Body Weight
Alkaline Phosphatase and Age
LDH and Age
Significant positive correlation coefficients were
observed in the sample of 21 female horses between the
following paired variables:
*Inorganic Phosphate and Glucose
Inorganic Phosphate and Uric Acid
Inorganic Phosphate and Alkaline Phosphatase
Glucose and Alkaline Phosphatase
Uric Acid and Albumen
Total Protein and Total Bilirubin
Body Weight and Age
Significant negative correlation coefficients were observed
in the sample of 21 female horses between the following
paired variables:
Calcium and BUN
Inorganic Phosphate and Body Weight
Inorganic Phosphate and Age
Glucose and Body Weight
*Glucose and Age
*Uric Acid and Total Protein
Uric Acid and Age
Glucose and Total Protein
Alkaline Phosphatase and Body Weight
Alkaline Phosphatase and Age
50
Those correlation coefficients in the female found to be
significantly different from the corresponding correlation
coefficient in the male are indicated with a *.
Significant positive correlation coefficients in all 44
horses as a group that were not significant in either of the
sub-groups considered separately were observed between the
following paired variables:
Inorganic Phosphate and Cholesterol
Uric Acid and Cholesterol
Inorganic Phosphate and LDH
Significant positive correlation coefficients in all 44
horses as a group that were also significant in both the
male and female sub-groups were observed between the follow-
ing paired variables:
Inorganic Phosphate and Alkaline Phosphatase
Significant negative correlation coefficients in all 44
horses that were also significant in both the male and
female sub-groups are as follows:
Inorganic Phosphate and Body Weight
Inorganic Phosphate and Age
Alkaline Phosphatase and Body Weight
Alkaline Phosphatase and Age
All other positive and negative correlation coefficients
that were observed to be significant in all 44 horses appear
to have reached significant levels only because of a very
high correlation coefficient in one of the sub-groups.
51
In considering the fifth hypothesis, it is concluded
from this study that there is no difference between male
and female with respect to the mean values of the 12 serum
biochemical parameters as reported by the autoanalyzer in
the population of horses from which this sample of 44 horses
is a representative sample.
B. FELINE DATA
With respect to 15 female cats, the following positive
bivariate correlation coefficients were found to be
significant:
BUN and Uric Acid
Total Protein and Body Weight
Total Protein and Age
The following negative bivariate correlation coefficients
were found to be significant in the data obtained from the
15 female cats:
Calcium and Age
LDH and Age
The 15 male cats showed the following positive correla-
tion coefficients that were statistically significant:
Calcium and Inorganic Phosphate
Inorganic Phosphate and Alkaline Phosphatase
Inorganic Phosphate and LDH
Glucose and Age
BUN and Uric Acid
52
Cholesterol and SGOT
Total Protein and Age
Albumen and Body Weight
Body Weight and Age
The following statistically significant bivariate cor-
relation coefficients were demonstrated from the male data:
Calcium and Glucose
Total Protein and Alkaline Phosphatase
Alkaline Phosphatase and Body Weight
Alkaline Phosphatase and Age
None of the correlation coefficients in one sex differed
significantly from the corresponding coefficient in the other
sex. Admittedly, the relatively small size of the sub-groups
is a limiting factor.
The influence of the number of observations on the sig-
nificance of correlation coefficients is evidenced when we
consider the 30 cats as a single group. Significant posi-
tive correlation coefficients in all 30 cats were obtained
for the following paired variables:
Calcium and Inorganic Phosphate
*Calcium and Cholesterol
*Calcium and Alkaline Phosphatase
Calcium and LDH
Inorganic Phosphate and Alkaline Phosphatase
Inorganic Phosphate and LDH
Glucose and Age
53
BUN and Uric Acid
Cholesterol and SGOT
Total Protein and Body Weight
Total Protein and Age
Albumen and Body Weight
Body Weight and Age
The following significant negative correlation coeffi-
cients between paired variables were observed in the 30 cats:
Calcium and Glucose
Calcium and Total Protein
Calcium and Age
Inorganic Phosphate and Body Weight
Inorganic Phosphate and Age
Glucose and LDH
Total Protein and Alkaline Phosphatase
Alkaline Phosphatase and Body Weight
Alkaline Phosphatase and Age
LDH and Age
Significant correlation coefficient in the combined sample
of 30 cats that were not significant in either males or
females considered separately are indicated with a *.
Certainly, it would appear that additional observations
in both male and female cats could result in an increased
number of bivariate correlation coefficients attaining
significance; this is particularly true of the female group.
54
With respect to the fifth hypothesis of equality of the
mean vectors, it is concluded that there is no difference
between males and females in the means of the 12 biochemical
components in the feline population from which this sample
of 30 is representative.
55
IX. RECOMMENDATIONS
The analysis of the equine biochemical data suggests
future work is desirable, particularly in the measurement
of albumen, LDH, and SGOT as reported by the autoanalyzer
.
The significant bivariate correlation coefficients , par-
ticularly as they occur in the female, should be exploited
for possible diagnostic significance.
The analysis of the feline biochemical data demonstrates
that animals in apparent robust health can nonetheless be
hypercholesterolemic , at least as reported by the auto-
analyzer. Even when the 9 most extreme cases were ignored
the lower limit of the cholesterol tolerance was consider-
ably below those values previously reported by Cornelius
and Kaneko [20]
.
This work has been undertaken as a pilot study in
multivariate analysis of biochemical components of the serum
of two of the common species of domestic animals. Emphasis
has been placed on tolerance interval estimations, possibly
significant bivariate linear association as measured by the
Pearson Product Moment Correlation Coefficient, and a
discriminant analysis of the male and female sample mean
vectors. With the increased use of the autoanalyzer and
increased availability of data, this method of statistical
analysis can and should be expanded.
56
LIST OF REFERENCES
1. S.chwabe, C. W. , Veterinary Medicine and Human Health ,
p. 3, Williams and Wilkins, 1964.
2. Technicon Corporation, SMA 12/60 , p. 1, 1967.
3. Coffman, J. R. , "A Perspective on Clinical Chemistriesin the Horse ,
" Veterinary Medicine/Small AnimalClinician , vol. 65, No. 11, p. 1092, Nov., 1970.
4. Ibid, p. 1092.
5. Schor, S. S., "The Mystic Statistic," Journal of theAmerican Medical Association , vol. 95, No. 8, p. 615,1966.
6. Worcester, J., "The Statistical Method," New EnglandJournal of Medicine , vol. 274, No. 1, p. 27, 1966.
7. Technicon, Op. Cit. , p. 1.
8. Wolff, W. A., Tumbleson, M. E. , and Littleton, C. A.,"Serum Chemistry in Normal and Diseased Horses,"Advances in Automated Analysis , vol. Ill, Mediad, Inc.,White Plains, N.Y., pp 179-135, 1970.
9. Kaneko, J. J., Personal Communication, October 22, 1970.
10. Coffman, Op. Cit., p. 1024.
11. Mood, A. M. , and Graybill, F. A. , Introduction to theTheory of Statistics , pp. 149-150, McGraw-Hill, 1963.
12. Anderson, T. W. , An Introduction to MultivariateStatistical Analysis , p. 2, Wiley, 1958.
13. Ostle, B. , Statistics in Research , p. 225, Iowa StateUniversity Press, 1963.
14. Snedecor, G. W. , and Cochran, W. G. , StatisticalMethods , p. 557, Iowa State University Press, 1967.
15. Sokal, R. R. , and Rohlf, F. J., Biometry , p. 520,Freeman, 1969.
16. Ibid, p. 519.
17. Ibid, p. 521.
57
18. Dixon, W. J., BMP Biomedical Computer Programs , p. 190,University of California Press, 1970.
19. Bowker, A. H. , and Lieberman, G. J., EngineeringStatistics , p. 228, Prentice-Hall, 1964.
20. Cornelius, C. E. , and Kaneko , J. J., Clinical Bio-chemistry of Domestic Animals , p. 100, Academic Press,1963.
58
INITIAL DISTRIBUTION LIST
No. Copies
1. Defense Documentation Center 2
Cameron StationAlexandria, Virginia 22314
2. Library, Code 0212 2
Naval Postgraduate SchoolMonterey, California 9 3940
3. Department of Operations Analysis, Code 55 1
Naval Postgraduate SchoolMonterey, California 93940
4. Assoc. Professor A. F. Andrus , Code 55As 1
Department of Operations AnalysisNaval Postgraduate SchoolMonterey, California 93940
5. LTC Robert R. Jorgensen 1
246 Ardennes CircleFort Ord, California 93941
6. Office of the Surgeon General 1
Attn: MEDVSDepartment of the ArmyWashington, D. C. 20314
7. Dr. J. J. Kaneko 1Professor and ChairmanDepartment of Clinical PathologySchool of Veterinary MedicineUniversity of California - DavisDavis, California 95616
59
UNCLASSIFIEDSecurity Classification
DOCUMENT CONTROL DATA -R&D^Security classification ol title, body ol abstract and indexing annotation must be entered when the overall report is classitied)
I originating ACTIVITY (Corporate author)
Naval Postgraduate SchoolMonterey, California 93940
Za. REPORT SECURITY CLASSIFICATION
Unclassified2b. GROUP
3 REPQR T TITLE
A MULTIVARIATE ANALYSIS OF CERTAIN BIOCHEMICAL COMPONENTS OF EQUINEAND FELINE SERUM SAMPLES AS REPORTED BY AN AUTO-ANALYZER SYSTEM
4 DESCRIPTIVE NOTES (Type ol report end, inc. I us i »t dares;
Master's Thesis; March 19715 AUTHOR(S) (First name, middle initial, last name)
Robert Rothnick Jorgensen, Sr. ; Lieutenant Colonel, United States Army
6 REPOR T D A TE
March 1971la, TOTAL NO. OF PAGES
617b. NO. OF REFS
20Be CONTRACT OR GRANT NO.
6. PROJEC T NO.
9a. ORIGINATOR'S REPORT NUMBER(S)
9b. OTHER REPORT NOISI (Any other numbers that may be assignedthis report)
10 DISTRIBUTION STATEMENT
Approved for public release; distribution unlimited.
II. SUPPLEMENTARY NOTES
13. ABSTRACT
12. SPONSORING MILITARY ACTIVITY
Naval Postgraduate SchoolMonterey, California 93940
This thesis contains a multivariate statistical analysisof the results of an automated analysis of serum samplesfrom the horse and cat. In the horse, 12 biochemical com-ponents plus body weight and age are recorded; thus,observations are made on 14 random variables. In the caseof the cat there are observations on 13 random variables.Ninety-one (91) pair-wise correlation coefficients arecomputed from the equine data and 78 pair-wise correlationcoefficients are computed from the feline data. Extensivehypothesis testing concerning these correlation coefficientsis conducted and the results are presented. A discriminantanalysis for 2 groups, male and female, is conducted foreach species. In this analysis the vector of sample meansof the biochemical components plus body weight and age formales is contrasted with the corresponding vector fromfemales. Tolerance limits for each biochemical componentmeasured are presented for both species.
DD FORMi nov es
S/N 01 01 -807-681 1
1473 (PAGE 1)
60UNCLASSIFIEDSecurity Classification
1-31408
UNCLASSIFIEDSecurity Classification
key wo ROSROUE *T ROLE W T ROLE W T
MULTIVARIATE ANALYSIS
SERUM - EQUINE AND FELINE,AUTOMATED ANALYSIS
DD ,
Fr»"..1473 'back, UNCLASSIFIED
S/N 0101-807-6921 61 Security Classification 4-3 1409
Thesis
J825c.l
2 2 2 7 ti
126356|
jorgensen
A multivariate
^nalvsis of certain
^hemica! components
of equine and feline
serum samples as re-
ported by an auto-ana
lyzer system.?
g
? ft
ts
Thesis 126356J825 Jorgensenc.l A multivariate
analysis of certainbiochemical componentsof equine and felineserum samples as re-ported by an auto-ana-lyzer system.
thesJ825
A multivariate analysis of certain bioch
3 2768 001 03038DUDLEY KNOX LIBRARY