NON- PARAME TRIC TESTC bei ng analyced consts merely

Non- pa rametric terts are tatbat or clasyication

Yonkin
do not depend on he particular
form4he

|
dkotribution. Dísad vartages o on- parametric
They don4 make any arumption tects

regarding ihe dorm dhe populoti warte da ihe doa can

an d arp no bared ondhe parameler be handled by parametric

o
This ic unlike dhe tade we looked
previouly uohich
on gctimadinq or
were
eting
pcued
hypolhasid
ome
labourouc
thee der may be
for large vampler.
about one or more popu farlion
parameters DeN-TEST (Taet 40c randemne
Suppase , [Link] l an erdered
Advandaqet
tec
They allow
Non-parametric

for the derting

sample from
|donaity
1enady
gl) )nll d
and lel [Link]
be qn independpnt ordere
ion wrth

ample
yoa

h,Polher 4h at are n atementifrom onoher pepulelirn olh

about populat ion para meler valuar denoty ty)
hey moy be wied hen he form |e
uch odedl he campler
Ah vampled populotion is un known |have been drawn from he same
They Aend to be computaton aly |populelion or have been random
easier and consequ eny more selecled.
quickly apliod bah paramelric |We combine he wo samplps oand
procedures obrer vations prder
ofrange Ahe in

hey may be used hen dhe data | ihe com bined

magntudo togive Ahe
A i
Ayalyo imop
orderedsgmple
os Test
Z0
claliic;
h-r
acPouence lter
one kind currounde d by a Decicion Rule.
|equence leler {he ther Rejecl Ho yi
|kind and Ahe
run
no. el empr in a 120 7 Za
he
Length
bh
Is usually rejerred
he
vample come
run . to

1rom
as

he ane Comptolion dest olalkticj

populotion hen Ahere will be zo:19- ai

thorouqh minqling ond and

cOncegu enlu -he no run uill
be large Satitical decicipn,
yhe ca mplas came rom diferen o.1106 1.q6.
populaion so he ranges do not We faíl to rejecl Ho.

over lap hen hore wll be on ly Awo Conclurion;

runs randon.
Dala is

hoc been choun -hol he meon

no D
runci SIGN TES T

Ihesqn Aest oc uses on he

aod varianCe
|medon rather hon ho mpan q
0 meaure
d,an,nalan,na - n-na)
(n tna) (ntna -)
locion
The
. ceml rol {pndan or

Ahe gnlg umplion underlying

ic Ahal he ditribut

v0ciable reract i cominu

The run theory can akso be used This assumption ruler ou ke
Ho test yor
manner.
ron doon eusin a cimilar ue nominal dala
The iqn Aer qer rle nome from
Hhe ae Abal lhe (t and (
Examplp 1 rodher han ihe numeri cal value
|he win locu reord o Cprtain providecl in he rqw data are
run
Hoolball leom pr Hte la 4 eol in tho calculotipns.
concequedive gamar u ac pllowcLed x , XaXn be aample ro

a conlinuous dalibution f
wlww w w w wlw w w lL and
Tat ue are

wmlorn lkding the hypolhef

funetion
he
|1arl dhe median 1)voy
Can we that
conc lude
Peorma equol o o dpecyied value mo
nce 1his eom ic random ? m=m
Dadaj Conni adacl he om,
[Link]
n,: 35 Ho, m = mo
0a 14 ., mmo
n= 19 where f m):o.5
r -ax35 4 tl 2 |hshypolhesis can earily be
Heied b noting hal ca ch 1hel
0-2x354(aa5x14-a5 -4) buervaiond ull independ eelly
(25+ l35t4-) be less than mo orlh probabilty
4)
mo), hence f uwe leli

Hypothacic,
Ho, bla i randem.
H,Datq i n random. Ahen II,...nare Independent
 bernoulli random variable with wouldimprove 1be appearance
parameer
hypolhasis
P4lm) and co tho null |mentlly challenged gd.1og"
i equivalent to stating celecded at ranom "re ceired
|Ahal p- Ya.
pecial iatructor in Pe ronal
Ihecores oblained are
heloo
4,5 s, 6, ,G/j0, 7, 06
The dast totistics qor
wich to knos 1 we can
gn 1et N the, median eor p
conclude that
etherdhe
Dr the ober ved no.
cbserved no 0
iqs| 4he populariDn neselh ith
(4)
e
The noure (-)
he 59n. |asrme hic sample to bave been
aHlernotive draun diterer K jrom 6
hypolhers deleimine ohich
|here {ect olidt ice is apprepriote
Aasumption;
|na Gven -ler .any the oll own | Dola i cortinuour gq

aternolive hyprthecis is posrible,

)
Hypotherkj
Ho iMm=5K
i ,, m t5
H, p
pl-) Test ctatistis;

Care;
A
The
coueu
Gjiciendly mal no.
a rejedion Ho.
) T+ or
Shaller
Sbecuion rule j
e
T whchever vrlcenty

)
rejecl
teld at rticr Ahe no
)signs
p(x) DD25

A icienly mell (+hsp 6505"0025

cauBs rejertion Hb
The test talicdic ir {he no |x 453s96lo 1cc
t)signs T X-6;
Tt = 8
Slatictial decrion;
Ehor iSuHieiotlly Gmall nd
(4) signG or (gn cau e
Plx1): 2)S^
Kzo
05
1-K (tsig
ns

ecion H -(3)x0.5° 2 0-5"+ (9) x0-5xo-5

Distri buton
kle rejecd Ho
The est claliclis 0rll 4ollouw a
binormial bictibtion Concluion ;
The median ks no 5 Only
ue
DecÁsiDn Rule
nDrmal
We reied Ho th probdbilrty
observinq fPWer obsprvtion
. K or
|ernatively
We can uSe normal approx to istrueted

obt ain Dur tet Gatictic asi

)plT. <K)<« Zo=T- D5n
o.55n

Eample |We rejec oy

Rosearchers wish to knouw 4 |z.|>Zx or
instructions In perssonal groomng
 Zo|- 5x9) Clatotica ll decirion ;
plr2) o-D5
|()os*xos"+ (0)os'yo.6" t ()
Wered Hby o5xo5
=0.a327
Oa37 0.05
Izol>-96.
We reject tb.
R333 7-96 Conclusion,

H was beneficial
Sin Aest paced d ata
hen dhe dala do be ana lyzed Zo:2 - (o 5 xI)
conotc obervations in n atched 051 ngnec

pairc, an d the aruptions underlyig -36 2.l

|4be 1-tert are not met, or
the tet

Mpasurement Ccale wpak, the 3G Z- 164.

|
sign det may be employed to te rant

Silcoxon
.11 l6t Rejed Ho
the hypolhers hat the' median
derEnce
Hypolheric ;=
o. WILCOXON RANKED SIGN TEST
This is ured to Aest a null bypotha
Ho , m
o populorlion mean when nerher
plt:) plt *-) nor t z an appropriae dest
4 we hav e a cmall cample
Example a nc30 qom a po pulalion dhal u
A dendal rerearch team oiched kno wn to be
groly non -nof molly
to knoo
brucdo heir
deachi09 people hoo to | drbuted and cenral
totlh ould ha thearem í not app licable,
ld
|benericial. The fo lleuing dotachout ctatistier is rulbd out
oral hygne GCorR nvbjecte b7he etotkrtia ic n appropriate
recei ving oral hygqåne inglruclionc(x)|because be campled population
and t9 ubjeettafer receiving oral |dopr n4ficiently PProximmte
hyqine instructions (Y). A loo ceore a nor mal dctribuho,

indicode lexel high

hyqine 4 y the daa for onalyris art measured
x22+a4a3s32 sa55l2.d on atleaet an interval ceale, dhe
be undairable
|nce t uilT n moke-ull ue
|Aecumpliong | he inormoti on Comained in
bala is codinuou

|ypclhecie,
, mmy
|
ahe

4he
daa
more oppropr iale procedure t
wilcox on qned rank ted

|uhchmaker uE {he nrude

|TeG oti | the derencRr blon mPndrement
T. and on
ypolhpned locati
Decision ryle, |parameer, rolher han de tbe
We rejec Ho 4i |1gn he d1e rence
Pxk)
KP Cx k) <o os Astumpdions
(2) ogo 5 D The somple N random
la)The varioble is coninuouc.

Compulolion 1ec{ cotidlit )dri The population

a
buted he me an u.
ymmeically
he mpasuremend srale is otleatt
Tt=4 4)
iterval
 X o5A)
Hupoth eris.

|2)Ho, uo
614
13)Ho, u uo V 4 T+ 1-21
7-42
3
10

Tect
(chotícdiac 7.5
Procedure,
)Subdrad the hupot hesies
mean do69
G56
dix-o
from each 1he obrer vation xto 598 O43
'oain di - x-Ho.
Lf an y Xi K
O that di =o,
equal todhe hegn - ag
eliminale that di 5-89
ftom dhe calculai
onxs and reduce
n 2
5-14 o34
3)Rankaccordingly
the ucable di 5-17
from the
Smallect to he largett
a Ke
rRgord toAhe ign
2) Asciqn each rak {he iqn
withoul
(ldl) od m wich do know
the bas
we can conclude
thir dot a 4hal the
d ihel pbpulalion
i dyferP ron
wieldecl hal rank.
meon 5
+)find T+ 4he Cun he rankr urh
Hypotheris)
HoW=5-05
and T the cunm the H, t5 05
rank wHh -Ve ign. |2. Ted laic
When we have 2-sided altern ative 3.
beotion Rulei
hypotheis eidher
Sicient4, cmall pject to

Ho
eject
The Tluo,^
dact clotkeic s T, or whicheve T<25
I-
Smaller. T-34
is
hen Ho; o, we expet our TET-34
cample toyielada larqe value 4 Stotitical decioon ,
, when
4 drve 4 VHicienl
T+ Dll Cae arejpcion
he

mall valuo
Ho and
aHernative hypothesis

5.
34 >Q5
We fail
Concluri on,
to reject Ho.

T i {he tet aielic. The populati on mean is nol

hen HuLo drue,wp eAped dyperent from 5. o6
|our comple to yield Q laqe value
Small
|T- a cuicieny uill" Lapproimatioo do NJeoxDn
caue reieion }o and T. u he
Hert aclic. |4 hos been choun that
n(nt)

T
0
Decicion

T,
Rule.

Tn j

Tn,d.
From example 4)

|Cardiac output w as measured wrth a

mple ran dem camgle 16 post- Cardiad -15 6.3L 310
surgi ca l pasiion in the let nat
 34-6o --.+76
bN3lo o= lad - bc)
14761<l96 ABCS
reject Hoy >

,
Example 5.
Weail to reert Ho and concl ude Leveld meral head corer jr
that the pop-mean is not di from hiqhssch b oy9Yen ar
g o5.
|ban: 35196,27,a, 27,36, 23, 25
Rural: 29,50, 49, 2R A2, 41, 431
MEDIN TEST
Ndest) 60 1, 45,k i3326 46) 5
|hu is non- parametric pro cedure 3 &t, Gl,25, 35. 4
hat will be ued to tee the null
belermine ir{he mediqn the
hypothei that a inde pendent samplecrural menlal heaHh rore and urban
have been drawn
Trom populationc mental healh sore are eq ual.
pith the came median
Median

|Ascumption a median dacd a7,>4,, 33,34 )35,35,3, 38,

he cam pler are rondoml and
ndependently seleced heir rom 50,50
apective populotipns 35134 345
|he Populadipar are 1he ome 10rm
2.
d'Hering, only
The varnable
in location. BL
interest is cpotin uoue Rural

Urhqn

ypadhocs Tod l4 |14

3. Ho, Mx M BCD
,, Mx > My Rs(1a1)-xe)]°
Ho, Mx My 19x9x4 +.

as(s4-14)
33510
&&X4900
The tect etatitias will be a ehi-cquore 33516

0 e5 -3SI +
becii on Rule + 094 3-814
rejecd to

The medon the rural menr al

heal4h c COre and urban nedal

heoHh rore are n equal.
|PcncedureL0 lan
Calcule he mm po mpda co the
|o d mpleu Combined.
he
MANN ALHITNE TEST Wilcoxaa

[Link] e for eg ch numb er Ranked ya Tacd)

oboervoton ollhn obove and belbuo The median tel dogc mat make qull

he common reqon and Cum marisp the |0SP all {he inormation Prerenl in

r esults in q x table as ollows, Ahp wo a mples'when he variadle

Above Belous Todal inere meafured on a leal qn

larb =A ordnal scale
C d normatron
cHd
By using an obser vali on

Total ac= c btd:b N |coenl o manlu1hat ahe her

boye or belou he
 medion
Wann-
iaUade
whitney Het l -herefore Prpa ed
iormation . Hhe
|
alernative
M<M,,woexpedt 1he
hypothari

nce baced on qned to

-lherankr |um {he rank aSri
obrervation:
Hhe peputotion bbservalíon from
ilicer more
m:
to be smaller
median ingor mationthan X population
tat rank
Hhan 1he rum he
|aciqned grom he Y populahon
|dccumptionc o Mann- lal bctne Ted The dart attic is bare d on
|Thedwo cgmple have en independen- My
be bis rionale such a way
in
|y and randomly drawn rom ther |Abat a cuyiciendly smoll
My

arithmetic population. Nalue Tuill cauea

he meaurenient ucale k allead ordinal reedon Ho. ie T<mn
GThe
1
dier
he
variable
POPulolionc
only
intererl

ort heir
dor at all they
i

medans
continuou .|

a
we
Ho;
are
Mx Mc H,
Aeting
valuo
M« > My
T
Ha
HM

cUH icienly large

|wll cauce reje tion vo 1hl
|Hupcthecis. we rejed o T >0m.n,t
i; Mx M. :T>nm- Wmn, a

Ho Mx My: for ; M, # My,

H,
, ;Mx > My.
cOmpurled values o T that
G. Aoi Mx > My
are ether dufjicienty large
or small will cowe eje tion
Ho.
Toct sdaticdics
P Ho i
Jo compude the test statistias,

we combine the two ramples and

nm- Wmn."
rank all observations qrom the
Cmallect to the larqast while Example e
Aack tomples to whidh The table below chow
|keeping,
haemoqlobin level a4ter one
each ohrer vali on belongs
|Ted pbservtion ore cfigned |qroup expoed do cadmium ar 0e
g rank equal to the mean |0xde. Je he
oiurh do knpo

rank positipns wohichhey can conclude 4hal prolonged

are ti pd. expocure Cadnniunn oxide
hen
The tecct otiicr ir reducec hae moqlobin level
Gpared b) Unexpoced (9)
gYen as nlnt) 174
T=S- 14.26
uherp n No. ample x
IG517?7
S obcervati ons
Sum
qsigned
roanks

o cample
[Link]
15
5

obrer vations-from he 2
value 156
pepulorlion 163
The choice u hich Uampler we 153 b5
label Xis arbitrary 1Ç.7
13 7
Decision Rule 153 16.

nedian 1he popu lat on

4he an t he median
 Ho; My Ma Tac (elatietioc
H,Ma <My In Ahisdert,a com parison i
moce betwpen some theor rticol
Tes-n
(ot) eum ulalive dictributfon uncti on
FAtx) and a sample
Ne cump
rejet Ho y ditribution yuneti Dn
T
mna The sample is a rondom uample
T4i5,10,0D5 from come pbpulation oith
|unknown cumulative ditribuhor
functivn
T=45 - (15x16) |4 here is a clase ggrreement
|wn. the theorati col and (comple
cumulotive ditributions,hen lhe
a5 46.:e reje o. |hypheui thol the comple uat
r
, do
Fxposure cadmi um oxide reduce drawn rom t he po pulation
haemoglobin levels pecigied Cumulalive distribulion
Hun etibn
|Let be dhe ccum urported
1he ranks ip |l Ahere isq descr
ipanc Hhon
|CCOmple 1and 4he (un ranks lhe Ahppr atica) ond obcerved
|n
are
cample 2. hen

moder
d etri bution a, brh m and n |cu mul ot ve dherlribution
large
can be,
>s),dhe Ho a qreot erent hen funhon
wel
PProx by q normal distri bution
(
hyptheris is rejeted.
The diere nce betoeen
and fs k) s
he null
(x)
me as ured bythe
F
s, -n,(n tha1l) |Hitics b ohich the
9reates- verica dican ce hn
ts Ca) and F()

Zo -S-us, w {0,1) |Ne reject Ho 4 bKSn,K

Aatumpianc
from exom ple , 3 Sample was randomly
2) Dlg is condinuous
selected

|3) Meqsuremen s cale is atleat

ordinal.
6s, I5 x10 (15 tiOtI)

ypolheci
Ho Fe (x)= Al) 7

145-J95
N325

1165 7164 we conclude hol the b elou

.We
2
|Can
Ho |data is normally dietributed
rejec
(ur a mean and .D6. 80
KOLMoGOROV SMNORV TEST5,42 80780841731
This i q tect used to determ ine |4, 7, Srj 77, 5, S,
how well 4he djerib ution (onne 80,
9a, a, 1, 18,1
cam ple doda coMorme tbcrome |7 E, 77, 9,&e 8.

theo rati cal dietributions

H ic an ternotive
a o
4he Chi |81 6 80
09uQre g00dnero 4t 1erl
 6 CAt)l)
OO5GG
|Ahe smallest value to n which is

7 R|D.020.b38 antigned to the lar qest value

When tLDO or more obervatípns
D.1G67 |083|.2033l0.p36G
7
|have dhe camp value,Pach obrervatjoo
0-2aba|0.7 514 |0.0292 |
g1VPn lhe mean he ranks

&C
o.3&89o-5
0.4723 |0-33 .370|0lc
|3 0.63g9
|
308 5lo-b6o4

|.138
ror
ep

orervations
which it tied.
Q, the ranks assigned
in eadh be k
32co 1922 |:1 056501547qTp
B4a2slo-7718 o7486 D.02 42
ave dded cFparortely give
Ri-CK rankcum)
S4I3| cgu
32o. 88691- 17
Aep 8; Ahe test statistics is

Ahen qiven q¢
1110228
n(n+1)
? - d(nt) ,
tlo bata is normally distibuted
H. Data k not normally distiburled where K" no. campler
Reject Ho4
no o b«ervation in
the ihsaMple
n- np, ober vatiorn i
all camples
dum rank in the
h sauple.

becision Rule:

D0.(541 hen
or
here
4ewer
are 3Lample and 5
obcer vations in each sampl,
O1547 40.a2)
|he gnyicance the computed )
We qail to rejec te dete r mined by
|s
kruskal Walls
bato is nor mall disributed. hable and we rp jed He if
H Reied
KRuC RAL-ALLS KWn,ans, K
TEST.
ver, when there are more than
Ho u e
Ths is red uohen {he Gumpons
|5Dbrervatianc ln one or more
underlyi pg q one- way ANova lcRb) Hhp camples or more tban 3 qrps
o0
|ae not metie Ahe populionc 4rm Hhen Ho ol| be ejeted iY
which he camplar have begn draon
are not notmally distriburled wrth
equal oriancec
|The on ead ion
|Ascumptians n Kruckal dallae tecl Hime o a cert ain timulus was ,2dng
Sampla are candomly celected. |ftudier in 3 uamplef
erperimental
bata is continuous
|animale omple ce Ved ar cortrol 3
8 bala ic Meqsurpd on atl east an |»hile the nimole in oMple 1 uere
ordinal scal e
tr eated wrlh drug A and hae in

Kauple were 4rpaled wllh

Tost sdaditícc
Pipr Ahe pplication the mulur
The n, na, ng obser vations from an uop Contude tha {hethree
|the K samples are combined into a
tompler djpr oith respert to
cingle.
Gerieu Gen and arranqed |reaeion 4ime.
in order
magn rtude 4 I, 40, 21 a0,

malled do the largert 2879,8oG

The observalions are dhen replced 3: 2,6, +)4
by ranks from one which is ascigned
 Ho, a ua.
Model.

3 (n) aa
nlnt) 3
3+ a-
Reject Ho
H> KNaAitit b 25 56
3
>56923.
B
3
123 Te-6x13) 5

-o--
9-8t6. Ho;,=da=ts
., A, +ta?ta
4- 846 >5-69a3

- he
We rejec
4hree compler der
to

uor reaciot k(ke)

R- 3nl(kn)

time Re ect Ho iti

n
FRIEDMAN TEST Xr>Xa,o. D5
Thic det is qppro priade whenever the
data is mecured on allenrt qnd ls't2s?)- 6x9x+)
ordinal ccale qand can be 9x3X

arrang n a 2-way clariticoti

ed o
meaniguly

0iven 4or e

LHupthecis.
RCBb.
. [Link]>6.22
e reied
:. The models
H.
are not preerre
equally
Va,,itj
SPEARMÄN RAN CORRE LATIon K

Ha; bi#bj his is a piocedure hat makor

|05e the tuo cels ranks
Tesl stauclics |Ahat may be asciqned -to the
|We use |ample valuec x and
3nlk)
nk(K) |Assumplionc

Decision Rule
) The sgmple bave been
randomly celected.
Rejet to ifi 12) Variable interert ic
Cominuour
|3)Dorta ir m arured en atloas
Exe mple 4 erdi nal scale

A phycician Conduced a utudy

lou vol
t
compare0modl |Hypelheckc
9 her 1)H: ond are

|o
electricdl cdimulplore mulually

physieal 1hergplts
were acked o yince pendent
rank the ctimulators ordor in H, xand qre not

A rank Ohe indicaies mutuallu ndependert

Prefprren ce.
e.

firc preference .e
wich toknow | 8.) Ho; xand m u{ually are
we conclude ha the mod e ls are independent So
Py
H, There is a endane
preferedequally
|> ho qe valuer *and
 0)Ho; arge valuer bepalred 4 15
543 169
Qndy Qre mutually
53 I7 53 a 15
,There
ovaluec b
is dendancy dor lqrgd
d0 be Paired
55
68 14
8 67
UHh Gmall valuer y G5 1

R348
Tac ctalhciac
Rseceduce
1) Rankthe valuer xTom
Hi Payo.
4 to n.
|0)Rank the Valuer (4 rom n (n)

.'yx
1 to n.
Reed Ho
|4) Compute d or each pai Payl> Pn,a.
obrervations by subdracting > Pao,o-o5
fank
1om >0-316
di Rt-R)tank
t) Square each di and com pute a342
20(20-i)
5) Compute the cogfficlem :-O-7Goq
Py -- @d
n(n-)
O.7Goq> 8189 o

Decision Rule.
We reject Ho.
:. EEG output is in
verely
correlated orh age.

CoNCORDENCE
KENDÀLL TEST OF
Fxample 10, hs used uhen we bich to
tudy 1he relclionchip |Aesl the hypolheir that a
between atk and £EG,dalo oere potlern indicaling adendancy
collected on 20 ubjeck
iuon the | 3or more 3roup4 to 0gree
|aper a0 and e0. The inverlqrtur |re c\gnicand.
|Licher to knpb ean be 42e. Cog Eordancetorl
Concluded lha hic regorming pthete,
par ticular |a 0i9ncance
)
level
EEG orp ic inveksely Correlaledwe cay 4hal a pattern
wth dye
indicating tendoncy {or people
Age(x) EEG d Ao 0greeebi.
dqgeoment9nficant
1515-13 |1evel ettenelontt

q5 11
G9
he hypherir ic 9iven au42
Ho, The 5els rank do
nol indi
19 icatfe93can
651 level Ceacante.
H, 7he so ran k exhibrt
o ignii(anl level of
338 0 2 Dg reememt
359 85 1G
|Tect rlatia
744
68 10 Kn (nt)
4a 19
ohere:
44 1S1 13 R=Sum ran ks arsigned
an teM
 n No. tens ranked.
K N, rple assigning G 5 3
ranks 5
Ditribution: a Kendal|Concardante
Peryorm
tert
reject Ho
>-pn. >1,o.05
avple
Cuppose G people
|Varieier
ranked & dit
wine As rhoun bel ow:
X -3n[kn1
Knln)
R
3
5 atLs3) - 3x6(Bxa)
3x 8 X9

7 1456 > +.06

7 We
W1 reject Ho
Rankings exhibk a gn i Cat
level 0qreement .
Deerm ne 1he raokinqs pxhibh
O

*19nigicant level 0greement . |CHL-SQWAR E ESTS.

domptimes in we Come
Ho Rankings do ndt pxhibit aqreenment| cress cesearch

H; Ronkings exhibit greenem sCale

dada
and wp
meas ured cn nomiml
OIh do makeTome
nferences.

2:zR'-dnlklnt)) | easurel .on nominal

We reject
Kn(nti)
H f
| uay
which doec not conform to
e <Cale
Chunt dota
he
|afumpion mae tertr te
> .0G This leaves us wrlh the
optin
Chi Sguare tet
-

X:12 k(9°110°14Rt9119rIs a' Genera lly,in chi-cquare lat

83)-aXAlaxg) 4he Aes! ctic i
3xSX alwoy
91 ven as
|5 7 )4067
.
We reject Ho
Ranking exhibit a9nicond
where
0: Obterved frequency
level agfeemert E Expecled yrequeng9

Esouple a
|hreeexecuive 0CoMpany rankel becid ion rule
|4he perofmance_ rondom sample We re,jec Ho g
8 manqgers. lheir fanks are
|
9Yen as; M
|4dhquareect
3 )
Tedners
|) Chi-suore
Chi-square Hest
leu for independened
homegene
 The dietribtion ís binomial
Geodeescpded Ho

hic ic appropriate when one with nt00PoD5

wishesto decide an oberved H, The dirlribulion ic n tor

|rctributon 1 rPquencip binomial

n
1ncompalible wrth som prp-concived
r
We hypoheized
distibution (0-E/E
wich to doermine, for

qoodnecr
example. whether or nol a cample
observed yaluec ome OG 79|

V=0-1-9
D
rondomvarioble is compatible H
oth the hypctheis hat H war D.1S43 4 we
draonrom poPularion valuer
thal normall
is ditributed. INe rejec Ho iyi 9:n:
The proce dure for reachin9 a X>v,r
No Ne
rejed

decision concit Pldcing the >An-g o

values int o mutually exclative
cot eqorias or clas imervals
>-,0-05
and clarer

|noing theireguency oceuren Ce

>5.9. Ho,

parom
valuer in each Coleqory.
0.1544 59l vadio

eter

Exanple3i We fail te rejec to.

Fizz company makes comp reued Th dastibution i bin omial. ured

gau ylinden used gor carbonotinq

water ond inglating lye vete.
n
Cylinderc ore roldin boxpu 0 tor

estimate
iAVentory planning and
Ocacionallu a cylin der wil be Cotrolpurpose, Crap hemico eompu

defective lhe qualdy control com pany uamk io kn00 G

nonager oter tha{the overoll |aleu aliauid chemical
Aom{ng

proportion dyedive c4linder 0renorma ll dicdribrled.

is ad an acreploble Iow evel. Sales ho 'o sgmp

He
ngnficonce
|1he no
ao ha
the

o
(PA
war to tert at 5% level op oco)
hol

bin om ial
boxey
diribution
o
|a4
34-a5S
2s s-87
|31 -38 S
doy

131lnd
le

|He cutaineda rordom comple

43
L00 boxs ond tound i00 cylinderbs.s- 40
[Link] rerorded he no . Q7
|bcret cortoining deerive|4iS- 3 l,
indere 0g,
d Ho Lor 4-4G

212
|No

in boxe oberved 4 46
39 035Y 49 584
0.1135 21, 15
12 20 2
3 4100552 42
4 0.23 22iai7451
5
0.0op36

P ao00
 (o-eE
O0107 ExampleS
Aet sahca
qu
Gurvey wau
done
do doterm ine
ction
on htel

A second
wil| return.
|whelher Ahey
all rapondet
|2ution was ark ed were
|ho 1hal hey
indcoted
fo reurn These
n likely
aked do indi ce
1quert wpre
Hhe Prtmor feDn forheir
|eponsereorm a hypelherh
|l 5% level gnyiconce
Holel
Golden Palm Palo
Feedloack Palm RoyalePincesd lo
Price 7 37 6

Locaticr 60
Room 5 31
I3 8
Olhes
Total 33

|o, Holl and feed back ore

independerl.
H Aotel and feedbock 0re n
inde pende nt

(o-e)?

Chquare Jeloriocepelenc
Concider a coigene talble consitird
e rejet b
a variables lax2 toble) >c,ocs
variable has R col egorees X là.592
e while oathe e cond vGriable har
Ccategories
|e wchtoeel lbe bypolhecf
abl hp associadion hese f0-E)°/
Huoo variobler. 2331529 2307 4
1f
|This hypolbess can be claled as 823 5

Hoj Ro ws and column are 31 a3 C411 1540I

independprd (Have no acoeiotion) 09 |2533 53 |410o
HRowr and olumng are dependent 130 588a O 59 3
(Have qssDciaion) %R765 8.198
The e calisliçs is as, 9en 13|.5882
5 |5 0G 0-O4O5
E 13|10.q|2 |0. 381 4
O03o7
where v (r-)l-) 3|5.6
1313G471

810.2353
-CR3t
|e feect. Ho
|27,10
 50595 L5.991
al40 7 12. 590.
Ne redt Ho.
We fail to
fropor ionr
reje ct He
workesr rated
The hole! and
feedback have
an aSpciio satisfa dory are equal

hsgdatye
Sometimes We hoogasaidy
iteresled Tn
re
hypalheiang dhe equalkly
two ropordionc bwn
variabler which are
he categorised.
ptocedure remain same a
tast for
|independente excepl hypehec alement

Example E
Consider
peformance workers
based on level
educaion 3jven
as follous,
Educational level.
Ao hghHgh shomplek
<SchoolN conmpl high schTO
Rding Satyaty
Not cai
Q5
Total 80
A 5Y level o iqni
conce
determine 1here is homcqenerty
in Aerms o [Link]
acrasS
|levels educ ation
Ho, Proporti ons Df workers roted
Sat nsfa ctory ore eaqual acracr

H. , digerent
PrepDrti
fatioory
educotion
ons
worke2 rated
are not equl
lereks

acrass diHerent education

lerls

(o- E)

v:(r-0cc-)
e rect Ho ,

e-E)cl
G3 O.75
G5 7 0 3511

O8333
6D545
X5 D595