UNI

T–4
CONCEPTOFRANDOM VARI
ABLE

Random v ar
iabl
e:
1)Ar andom vari
abl
eisaf
unct
ionX(
w)wi
thdomai
n‘S’andr
ange(
-∞,
+∞ )such
thatfor

ev
eryr
eal
number‘
a’t
heev
ent[
w:X(
w)≤a]εB

2)LetSbethesamplespaceassociat
edwit
hagivenr
andom exper
iment
.Real
val
uedf
unct
iondef
inedonSandt aki
ngval
uesi
nRiscal
ledaonedimensional
r
andom v
ari
abl
e.

*Random v
ari
abl
esar
edenot
edbycapi
tal
let
ter
sX,
Y,Z….

*Val
uesar
erandom v
ari
abl
esar
edenot
edbysmal
ll
ett
ersx,
y,
z….
.

Pr
oper
ti
esofar
andom v
ari
abl
e:

I
fXi sarandom v
ari
abl
eanda,bar
eanyt
woconst
ant
s,aX+bi
sal
soa
r
andom v
ari
able
I
fX i
sar
andom v
ari
abl
e,X2i
sal
soar
andom v
ari
abl
e.
I
fXi
sar
andom v
ari
abl
e,
1\Xi
sal
soar
andom v
ari
abl
e.
IfXandYar
etworandom v
ari
abl
esdef
inedovert
hesampl
espace,X+Y,X-Y,
aX,bY,
aX+bYarealsor
andom v
ari
abl
es.Whereaandbar
econstant
s.
Ty
pesofr
andom v
ari
abl
es:

Ar
andom v
ari
abl
eisdi
vi
dedi
ntot
wot
ypes

i
)Di scr
eterandom vari
able
Av ari
abl
ewhi chcanassumeonl yacountabl
enumberofr
ealv
aluesand
forwhi chthev al
uewhi cht hev
ariabl
etakesdependsonchanceiscall
eda
discret
erandom var
iable.
Or
Ar ealval
uedf uncti
ondefi
nedonadi scret
esampl
espaceiscall
eda
discret
erandom var
iable.

Exampl
e:
NumberofHeadswhent
hreeunbi
asedcoi
nsar
etossed,
Tot
aloff
aceswhent
wodi
cear
ethr
own

KAKATI
YASERI
ES Page1
 Numberofsui
cidesi
naday
.
Numberofpr
int
ingmi
stakesi
napage.
Mar
ksofast
udent
Pr
obabi
li
tymassf
unct
ion:

I
fX isadi scr
eterandom var
iablewi t
hdisti
nctval
uesX1,X2,X3.
..
..
..
..
,thent
he
f
unct
ionp(
X)definedasP(x)=P( X=xi)=pi.Foral
lxi

i
sknownaspr
obabi
l
itymassf
unct
ionofr
andom v
ari
abl
eX.

Pr
oper
ti
esofpr
obabi
li
tymassf
unct
ion:

Thepmf(
probabi
l
itymassf
unct
ion)p(
X)sat
isf
iesf
oll
owi
ngpr
oper
ti
es

i
)p(
X)≥0i
.epr
obabi
l
ityofanyxii
snonnegat
ive

i
i
)Ʃp(
X)=1i
.et
otal
probabi
l
ityofaR.
Visequal
to1

i
i)Cont
inuousr
andom v
ari
abl
e

Arandom v
ariabl
eXi ssai
dtobecont
inuousi
fitcant
akeal
lpossi
ble
v
aluesbet
weencert
ainli
mit
s.or

Ar ealval
uedfunct
iondef
inedoncont
inuoussampl
espacei
sknownas
cont
inuousrandom v
ari
abl
e.

Exampl
es:

Hei
ghtofaper
son.
Ti
mei
nter
val
bet
weent
wosuccessi
vear
ri
val
s.
Di
stancebet
weent
woobj
ect
s.
Pr
iceofai
tem.
Av
eragemar
ksofast
udent
.
Pr
obabi
li
tydensi
tyf
unct
ion

Thepr
obabi
l
itydensi
tyf
unct
ionofar
andom v
ari
abl
exi
sdef
inedasf
(x)

Pr
oper
ti
esofpdf(
Probabi
l
itydensi
tyf
unct
ion)

i
)f(
x)≥0i
.epr
obabi
l
ityofanyi
nter
val
ofXi
snonnegat
ive

i
i
)ʃf
(x)dx=1i
.et
otal
probabi
l
ityofaR.
Visequal
to1.
KAKATI
YASERI
ES Page2
Not
e:f
oracont
inuousr
andom v
ari
abl
ethepr
obabi
l
ityatapar
ti
cul
arpoi
nti
s
al
way
szer
o.

Di
str
ibut
ionFunct
ion.

I
fxi
sar
andom v
ari
abl
ewi
tht
hepr
obabi
l
ityf
unct
ionf
(x)
,thent
hef
unct
ion

FX(
x)=P[X≤x]i
scal
l
edt hedi
str
ibut
ionf
unct
ionar
esomet
imescumul
ati
ve
di
str
ibuti
onfunct
ionoft
herandom v
ari
ableX.

x)=P[X≤x]=∑x
FX( ∞p
- (
x)f
oradi
scr
eter
andom v
ari
abl
e
x
=-∞ ʃf(
x)dxf
oracont
inuousr
andom v
ari
abl
e

Pr
oper
ti
esofdi
str
ibut
ionf
unct
ion

I
faandbar etwoconstantv
aluessucht
hata<bandF(
x)i
sthedi
str
ibut
ion
f
unct
ion.ThenP[a≤X≤b]=F[
b]-
F[a]
.
I
fF(
x)i
sthedi
str
ibut
ionf
unct
iont
hen0≤F[
x]≤1.
I
fx<yt
henF[
x]<F[
y].
F[
-∞]=0.andF[
+∞]=1.
Theoret
icalDistri
but
ions:
Bernoul
liExperiment
:Ar andom exper
imentwhoseoutcomesareoft wotypes
successandf ai
lur
eofoccurri
ngwithprobabi
l
iti
espandqrespect
ivel
yiscal
l
eda
Bernoul
liexperi
mentoraBernoul
l
itri
al

Bernoul
livar
iable:Arandom v
ari
ableXisdef
inedsuchthati
ttakesv al
ue1when
successoccursand0i ffai
l
ureoccursi
naBernoul
liexper
iment
,t henXi ssai
dto
beaBer noul
l
i v
ar i
abl
e

Ber
noul
lidi
str
ibut
ion:

Arandom v
ari
abl
eXissai
dtohav
eBer
noul
l
idi
str
ibut
ionwi
thpar
amet
erpi
f
i
tspr
obabil
it
ymassfunct
ioni
sgi
venby

x)=pxq1-x ;
p( if X=0,
1 and‘
0’ot
her
wise.

Pr
oper
ti
esofBer
noul
lidi
str
ibut
ion

Theonl
ypar
amet
erofBer
noul
l
idi
str
ibut
ioni
sp
MeanofBer
noul
l
idi
str
ibut
ioni
sp
Var
ianceofBer
noul
l
idi
str
ibut
ioni
spq

KAKATI
YASERI
ES Page3
 MeanofBer
noul
l
idi
str
ibut
ioni
sgr
eat
ert
hanv
ari
ance
Momentgener
ati
ngf
unct
ionofBer
noul
l
idi
str
ibut
ioni
s(q+pet)
Exampl
esf
orBer
noul
l
itr
ial
aret
ossi
ngacoi
n,t
est
ingel
ect
ri
cal
bul
bet
c.
Bi
nomi
alDi
str
ibut
ion:

Bi
nomi
alDi
str
ibut
ionwasi
nvent
edbyJamesBer
noul
l
iin1713
Arandom var
iableXissaidt
ohav eBinomi
alDist
ri
but
ioni
fitassumesonl
y
nonnegat
ivevaluesandit
sprobabi
li
tymassfuncti
oni
sgiv
enby

x)=nCxpxqn-x ;
p( if X=0,
1,
2..
..
..
..
n and‘
0’ot
her
wise

Bi
nomi
aldi
str
ibut
ioni
sgener
ali
zat
ionofBer
noul
l
idi
str
ibut
ion
Thet
woi
ndependentpar
amet
ersofBi
nomi
alDi
str
ibut
ionar
e n andp
Phy
sicalcondi
ti
onsf
orbi
nomi
aldi
str
ibut
ion:

Eachtr
ial
hast
woexhaust
iveanddi
sjoi
ntout
comes,
knownassuccessand
fai
l
ure
Thenumberoft
ri
alsni
sfi
nit
e
Thet
rai
l
sar
eindependentofeachot
her
Thepr
obabi
l
ityofsuccesspconst
antf
oreacht
ri
al
Pr
oper
ti
esofBi
nomi
alDi
str
ibut
ion:

Thepar
amet
ersofBi
nomi
alDi
str
ibut
ionar
enandp
I
fn=1t
heBi
nomi
alDi
str
ibut
iont
endst
oBer
noul
l
idi
str
ibut
ion
MeanofBi
nomi
alDi
str
ibut
ioni
snp
Var
ianceofBi
nomi
alDi
str
ibut
ioni
snpq
Forabi
nomi
alv
iti
atemeani
sal
way
sgr
eat
ert
hanv
ari
ance
q+pet)
n
Themomentgener
ati
ngf
unct
ionofBi
nomi
alDi
str
ibut
ioni
s(
q+peit)
n
Thechar
act
eri
sti
cfunct
ionofBi
nomi
alDi
str
ibut
ioni
s(
Binomi
alDi
str
ibut
ionisposi
ti
vel
yskewedi
fq>p,
negat
ivel
yskewedi
fq<p
andsymmetri
califp=q=1/
2.
Bi
nomi
alDi
str
ibut
ioncur
vei
slept
okur
ti
c,f
orl
argeni
ttendst
omesokur
ti
c.
Bi
nomi
alDi
str
ibut
ionsat
isf
iesaddi
ti
vepr
oper
tyi
fp1=p2=pi
.el
etX1 andX2

KAKATI
YASERI
ES Page4
 aret woi ndependentbi
onomialv
arit
eswit
hpar amet
ersn1,
p1 andn2,p2
respect
ivelythenX1+X2i
salsoabionami
alvar
iat
ewithpar
ametersn1+n2,p
i
fp1=p2=p.
ModeofBinomialDi
str
ibut
ioni
)if(n+1)pisnoninteger
,thei
ntegralpar
t
of(n+1)pi smode.I i
)if(n+1)pisaninteger
,bi
nomi aldi
str
ibuti
onis
bi
modaldi
stri
buti
on,
twomodev al
uesar
e(n+1) pand[(n+1)p]
-1.
I
fxisbi
nomialdi
str
ibut
ionwit
hparamet
ersnandp,t
henY=n-
xisal
soa
Bi
nomi
alDi
stri
but
ionwit
hparamet
ers(n,
q).
Poi
sonDi
str
ibut
ion

I
twasdi
scov
eredbySi
monDenni
sPoi
ssoni
n1937
Poi
sondi
str
ibut
ioni
sal
i
mit
ingf
orm ofBi
nomi
alDi
str
ibut
ion
A bi
nomialrandom v
ari
abl
eX t
ends t
o poi
son di
str
ibut
ion undert
he
f
oll
owi
ngcondi
ti
ons

i
)Ifnumberoft
rai
l
sni
slar
gei
.en→∞

i
i
)Const
antpr
obabi
l
ityofsuccesspi
sver
ysmal
li
.ep→0

i
i
i)np=λi
sfi
nit
e

Def
ini
ti
on:Arandom var
iabl
eXissaidt
of ol
l
owpoisondi
str
ibut
ion,i
fitassumes
onl
ynonnegati
veval
uesanditspr
obabi
l
itymassf
uncti
oni
sgivenby

x)=e-λλx/x!
p( ;i
fX=0,
1,
2,
3,
..
..
..
..
.∞ and0;
other
wise.

Exampl
esf
orPoi
ssonv
ari
abl
e:
numberofdeat
hsi
naci
tyy
out
osui
cide
numberofdef
ect
ivei
temsi
naboxof100i
tems
numberofpl
aneacci
dent
sperweek
numberofpr
int
ingmi
stakei
nawel
lpubl
i
shedt
extbook
Pr
oper
ti
esofpoi
sondi
str
ibut
ion

I
thasonl
yonepar
amet
erλ
Themeanofpoi
sondi
str
ibut
ioni
sλ
Thev
ari
anceofpoi
ssondi
str
ibut
ioni
sλ
Themeanandv
ari
anceofpoi
sondi
str
ibut
ionar
eequal

KAKATI
YASERI
ES Page5
 t
Themomentgener
ati
ngf
unct
ionofpoi
sondi
str
ibut
ioni
sexp{λ(
e-1)
}
Thechar
act
eri
sti
cfunct
ionofpoi
sondi
str
ibut
ioni eit-
sexp{λ( 1)}
Themeasur
eofskewnessβ1=1/
λi.
epoi
sondi
str
ibut
ioni
sposi
ti
vel
yskewed.
Themeasur
eofkur
tosi
sβ2=3+1/
λ.
Mode:I
fλisannoninteger
,thei
ntegr
alpar
tofλismodeifλisanintegert
hen
poi
ssondi
str
ibut
ionhastwomodev al
uesoneisλandanotheronei
sλ- 1
XandYar
etwoIndependentpoi
sonv
ari
ateswi t
hparamet
ersλ1 andλ2, t
hen
X+Yisapoi
sonvar
iat
ewi thpar
amet
erλ1+λ2,butX-
Yisnotapoissonvari
ate.
Forl
arge‘
n’poi
sondi
str
ibut
iont
endst
onor
mal
dist
ri
but
ion.
Nor
maldi
str
ibut
ion

Aconti
nuousrandom v ar
iableXi ssai
dtohavenor
maldi
str
ibut
ionwi
th
2
par
amet
ersmeanμandv ar
ianceσ , i
fit
spdfisgi
venby

1 1 2
f
(x)= exp{
-( x-
μ/σ)}:-
∞ <X<+∞ ,
-∞ <μ<+∞,σ>0
σ√2∏ 2

Thepar
amet
ersofnor
mal
dist
ri
but
ionar
eμandσ.
Nor
maldi
str
ibut
ionf
ir
stdi
scov
eredbyDe-
moi
vi
rei
n1733andal
soknown
t
oLapl
acei
n1774
LaterI
twasderi
vedbykar
kF.
Gaussi
n1809andusedi
tfort
hest
udyof
err
orsi
nAstr
onomy.
Si
nce thi
s di
str
ibut
ion was der
ived byGauss i
tis cal
l
ed as Gaussi
an
di
str
ibuti
on
Themostcommonl
yuseddi
str
ibut
ioni
snor
mal
dist
ri
but
ion
I
nnor
mal
dist
ri
but
iont
her
angeofXi
s-∞t
o+∞
X-
μ
Zi
sknownasst
andar
dnor
mal
vir
iat
eandZ= .
σ
Themeanandv
ari
anceofst
andar
dnor
mal
var
iat
ear
e0and1
Char
act
eri
sti
csofnor
maldi
str
ibut
ion

Thenor
maldi
str
ibut
ioncur
vei
sbel
lshapedandi
ssy
mmet
ri
calaboutt
hel
i
ne
X=μ
Themeanmedi
anandmodeofnor
mal
dist
ri
but
ionofcoi
nci
de

KAKATI
YASERI
ES Page6
 Thear
eaundert
henor
mal
cur
vewi
thi
nit
srange-
∞to+∞ i
sal
way
sUni
ty
Thepoi
ntsofi
nfl
ect
ionofnor
mal
cur
vear
eμ±σ
Thenor
mal
cur
vehasuni
modal
Al
loddor
dermoment
sofNor
mal
Dist
ri
but
ionar
ezer
o
Allev
enor
dermoment
sofNor
malDi
str
ibut
ionar
eμ2n=1.
3.5.
..
. .
.(
2n-
1)
2n
σ.
Themeasur
eofskewnessβ1=0
Themeasur
eofkur
tosi
sβ2=3
2
Quar
ti
ledev
iat
ionofNor
mal
Dist
ri
but
ioni
sQ.
D= σ
3
4
Themeandev
iat
ionaboutmeani
sM.
D= σ
5
Normaldist
ribut
ion has maxi
mum pr
obabi
l
ity atX =μ,wher
e maxi
mum
pr
obabi
li
typ(x)=1/σ√2∏.
Themomentgener
ati
ngf
unct
ionofNor
malDi
str
ibut
ioni
s exp{μt+
σ2t
2
}
Thechar
act
eri
sti
cfunct
ionofnor
mal
dist
ri
but
ioni μt-σ2t
sexp{i 2
}
Normaldist
ri
bution sati
sfi
esaddit
iveproper
ty.I
fX1,X2..
..
..
..
.Xn ar
en
i
ndependentrandom v ar
iabl
esthenX1 +X 2 +.
..
..
..
+Xn i
salsoanormal
var
iabl
e.
Ar
eapr
oper
tyofNor
mal
Dist
ri
but
ion
i
) P[μ-
σ≤X≤μ+σ]=0.
6826=P[-
1≤Z≤+1]
.

i
i
) P[μ-
2σ≤X≤μ+2σ]=0.
9544=P[-
2≤Z≤+2]
.

i
i
i) P[μ-
3σ≤X≤μ+3σ]=0.
9973=P[-
3≤Z≤+3]
.

I
mpor
tanceofNor
malDi
str
ibut
ion:

Mostoft
hedi scr
etedist
ri
butionssuchasbi
nomi
al,poi
sson,et
ctendt
o
nor
maldi
str
ibuti
onas‘n’i
ncreases.
Almostal
lsampli
ngdistr
ibut
ionsl
i
ket ,f,Chisquar
ear
efort
hei
rlar
ge
degr
eesoffr
eedom t
endstonormal
dist
ribut
ion.
Manyv
ari
abl
es whi
ch ar
e notnor
mal
l
ydi
str
ibut
ed can be nor
mal
i
sed
KAKATI
YASERI
ES Page7
 t
hroughsui
tabl
etr
ansf
ormat
ions
Theent
ir
elar
gesampl
etheor
yisbasedonnor
mal
dist
ri
but
ionpr
oper
ti
es
Smal
lsampl
etheor
yisal
sobasedonnor
mal
dist
ri
but
ionpr
oper
ti
es
Test
ingofhy
pot
hesi
s

Theconceptt
est
ingofhy
pot
hesi
swasi
nit
iat
edbyNey
mannandpear
son.
Stat
ist
icalhy
pothesi
s: stat
ist
icalhy
pot
hesi
sissomestat
ementorasser
ti
on
aboutapopulat
ionwhichwewantt over
if
yonthebasi
sofi
nfor
mat
ionav
ail
abl
e
fr
om asample.

Simpl
e and Composit
e Hy pothesis :Ifa stat
ist
icalhypothesi
sspeci
fi
esthe
popul
ati
oncomplet
elythenitistermedassi mplestati
sti
calhypot
hesi
sot
herwi
se
i
tiscal
ledacomposit
est at
ist
icalhypothesi
s.

Exampl
es:

H:
μ=40,
H:μ=50andσ=10ar
esi
mpl
ehy
pot
hesi
s.

H:
μ<40orH:
μ>40orH:
μ=50,
σ>10et
car
eComposi
tehy
pot
hesi
s.

Nul
lhypothesis:Indef
ini
test
atement
s aboutt
hepopul
ati
onpar
amet
eri
sknown
asnul
lhypothesisusuall
yor

Accordi
ngtoprofessorRAfi
sher
,nul
lhy
pothesisisthehypot
hesi
swhi
chi
stest
ed
forpossi
bler
eject
ionundert
heassumpt
ionthatiti
strue

Nul
lHy
pot
hesi
sisdenot
edbyH0
Alt
ernat
ivehypot
hesis:A st
atementwhi
chi
sal
ter
nat
ivet
onul
lhy
pot
hesi
sis
knownasalt
ernat
ivehypot
hesi
s.

Al
ter
nat
iveHy
pot
hesi
sisdenot
edbyH1
Err
orsintest
ingofhypot
hesis:
ThereareTwotypesoferr
orsintest
ingofhy
pot
hesi
s.

TypeIer ror
:Reject
ionof H0 whenH0 i
strueinknownast ypeIer
ror
.The
probabi
l
ityofTy
peIer r
ori
sdenot
edbyα.TypeIer
rori
salsoknownaspr
oducer
's
ri
sk

TypeIIerr
or:Accepti
ngofH0 whenH0 i
sFal
seisknownastypeIIerr
or.The
pr
obabil
it
yoft ype I
Ierr
ori
s denot
ed byβ .
Type I
Ierr
oris al
so known as
Consumersri
sk.

KAKATI
YASERI
ES Page8
 Rej
ectH0 AcceptH0
H0Tr
ue Wr
ong-
TypeIer
ror Cor
rect

H0Fal
se Cor
rect Wr
ong-
TypeI
Ier
ror

Cri
ti
calr
egi
on:Theregi
oni
nthesamplespaceSwhi
chamount
stor
eject
ionofH0
i
stermedascri
ti
cal
regi
onorr
eject
ionr
egi
on..

I
tisdenot
edbyW
I
tssi
zei
sα
Levelofsi gni
fi
cance:Thepr
obabi
l
ityofTy
peIer
rorα i
sknownasl
evelof
si
gnif
icance.

Poweroft
heTest:1-
βisknownaspoweroft
het
est

Onetail
edt woandtwotail
edt est
At estofanyst
ati
sticalhypot
hesi
swher
etheal
ter
nat
ivehy
pothesi
sisone
tai
l
ediscal l
edinonetai
ledtest.Inonet
ail
edt
estt
her
earetwotypesri
ghtt
ail
ed
andlef
ttail
ed.

Exampl
eH1:
μ<μ0(
lef
tti
l
ed)

H1:
μ>μ0(
ri
ghtt
il
ed)

At estofstat
ist
icalhypot
hesiswheret
heal
ter
nat
ivehy
pot
hesi
sist
wot
ail
edi
s
knownast wotail
edtest.H1:μ1≠μ2

Cr
it
icalv
alues:

Thevalueofat eststati
sti
cwhi chsepar
atesthecri
ti
calr
egionandthe
acceptanceregi
oniscal l
edt hecrit
icalval
ue.cri
ti
calval
ueisalso knownas
si
gnifi
cantval
ueortabl
ev al
ue.

Cr
it
ical
val
uedependsupon

i
)thel
evel
ofsi
gni
fi
canceusedi
nthet
estand

i
i
)theal
ter
nat
ivehy
pot
hesi
swhet
heri
tist
wot
ail
edoronet
ail
ed.

Forast
andar
dnor
mal
var
iat
eZt
hecr
it
ical
val
uesar
egi
venbel
ow

KAKATI
YASERI
ES Page9
ZSt ati
sti
c-crit
ical At1%l
.o.
s At5%l
.o.
s At10%l
.o.
s
value
Twot ail
edt est 2.
58 1.96 1.645
Righttail
edt est 2.
33 1.645 1.28
Lefttai
ledtest -
2.33 -
1.645 -1.28

Pr
ocedur
efort
est
ingofhy
pot
hesi
s:

Thef
oll
owi
ngar
evar
iousst
epsi
ntest
ingofhy
pot
hesi
s

Fi
rstset
upnul
lhy
pot
hesi
sH0
Setuptheal
ter
nat
ivehypot
hesi
sH1.thi
swil
lenabl
eust
odeci
dewhet
her
wehavetouseasi
nglet
ail
edtestort
wotai
l
edtest
Choosetheappropr
iat
elev
elofsigni
fi
canceαdependi
ngont
her
eli
abi
l
ityof
theest
imatesandall
owabl
erisk.
Forl
argesampl
escomput
ethet
estst
ati
sti
cZ
I
fthecal
cul
atedv
alueofZi
slesst
hant
abl
eval
ueofZweacceptH0
Lar
gesampl
etest
s

Forlargev
aluesofnweapplythenor
maltestwhichi
sbasedupont
he
fundament
alareapr
oper
tyoft
henor
maldi
str
ibut
ion,
P[μ-
3σ≤X≤μ+3σ]=
0.9973.
Pr
act
ical
l
yasampl
emayber
egar
dedasl
argei
fngr
eat
ert
han30
t
-E(
t)
Testst
ati
sti
cforl
argesampl
esi
sZandZ= fol
l
owsN(
0,1)
S.E(
t)
Thef
oll
owi
ngar
evar
iousl
argesampl
etest
s:

i
)Testofsi
gni
fi
cancef
orsi
ngl
epr
opor
ti
on

i
tisusedt
otestthesi
gni
fi
cancedi
ff
erencebet
weensampl
epr
opor
ti
onpand
i
t’
sassumedval
ue

Her
ethet
estst
ati
sti
cisZ=(
p-P)
/PQ/
n̴ N(
0,1)
.
Inthi
stestt
hati
sthev
ari
abl
exi
sassumedasabi
nomi
alv
ari
abl
ewi
tht
he
par
ametersnandp.
i
i
)Testofsi
gni
fi
cancef
ordi
ff
erenceofpr
opor
ti
ons

Her
ewet
estt
hesi
gni
fi
cantdi
ff
erencebet
weent
wopopul
ati
onpr
opor
ti
ons

KAKATI
YASERI
ES Page10
 Thet
estst
ati
sti
cZ=(
P1-
P2)
1
/PQ[ + ( )( )
1
n1 n2
]̴ N(
0,1)
.

i
i
i)Testofsi
gni
fi
cancef
orsi
ngl
emean

Herewet estthesi
gni
fi
cantdi
ff
erencebet
ween sampl
emean and i
ts
hypot
het
icalv
alue
σ
Thet
estst
ati
sti
cZ=(
X̅-μ0)
/ ̴ N(
0,1)
.
n
i
v)Testofsi
gni
fi
canceofdi
ff
erenceofmeans

Wher
ewet
estt
heequal
i
tyoft
wopopul
ati
onmeans

Thet
estst
ati
sti
cZ=(
X̅1-X̅2)
/σ[ (n11)+(n12)] ̴ N(
0,1)
.

v
)Si
gni
fi
canceoft
hedi
ff
erenceofst
andar
ddev
iat
ion

*Her
ewet
estt
heequal
i
tyoft
wopopul
ati
onst
andar
ddev
iat
ions

Thet
estst
ati
sti
cZ=(
s1-s2)
/σ[ (21n1)+(21n2)]
̴ N(
0,1)
.

Smal
lsamples:
I
fasampleofsi
zeni
slesst
hanorequal
to30i
sconsi
der
edasasmal
lsampl
e.

Thef
oll
owingarest
ati
sti
csusedi
nsmal
lsampl
i
ngt
heor
y
Chisquar
estat
ist
ic
tst
ati
sti
c
Fst
ati
sti
c
e‫א‬2st
Chisquar at
ist
ic:t
hesquar
eofastandar
dnor
malv
ari
atei
sknownasaChi
squar
estati
sti
cwithonedegr
eeoffr
eedom

X-
μ2
Chi e‫א‬2st
squar ati
stc=Z2=( )
i isa‫א‬2 v
ari
atewi
th1degr
eeoff
reedom
σ

X-
μ2
Chi e‫א‬2st
squar ati
stc=ƩZ2=Ʃ( )
i isa‫א‬2 v
ari
atewi
thndegr
eeoff
reedom
σ

Appl
icat
ionsofChisquar
est
ati
sti
c

Tot
estt
hesi
gni
fi
canceofhy
pot
het
ical
val
ueoft
hepopul
ati
onv
ari
ance

KAKATI
YASERI
ES Page11
 Tot
estt
hegoodnessoff
it
Tot
estt
hei
ndependenceofat
tri
but
es
To testthe homogenei
ty ofi
ndependentest
imat
es oft
he popul
ati
on
var
iance
To testthe homogenei
ty ofi
ndependentest
imat
es oft
he popul
ati
on
cor
relat
ioncoef
fi
cient
Goodnessoff
itt
est

Thi
stestwasgi
venbyKar
lPear
son
I
tisusedt
otestt
hegoodnessoff
it
ti
ngofdi
str
ibut
ionsorcur
vef
it
ti
ng.

oi-
ei2
Thet
estst
ati
sti
cChi e‫א‬2=Ʃ(
squar )
ei
Thedegr
eesoff
reedom f
ort
hist
esti
sn-
1
Testofi
ndependenceofat
tri
but
es

Thi
stesti
susedt
otestt
hei
ndependenceoft
woat
tri
but
es.

oi-
ei2
Thet
estst
ati
sti
cChi e‫א‬2=Ʃ(
squar )
ei
Forarxscont
ingencyt
abl
edegr
eesoff
reedom i
s(r
-1)
(s-
1)
Degr
eesoff reedom:t
henumberofi ndependentv
ari
ant
swhi
chmakeupt
he
st
ati
sti
cisknownasthedegr
eesoff
reedom

tst
ati
sti
c

thetst
ati
sti
cwasi nt
roducedbyWSgussetunderhi
spennamest
udent
,
andi
twasdev el
opedbyprofessorRAFi
sher
Theassumpti
onintstati
sti
ci sthepopulat
ionfr
om whi
chsampl
esar
e
drawni
snormal,
i.
etheparentpopul
ati
onisnormal
Thetst
ati
sti
cisal
soknownasst
udent
’stst
ati
sti
c
Thetst
ati
sti
c t=(
X̅-μ)/s/nf
oll
owstwi
thn-
1d.
f.
WhereX̅ samplemeanands2 issampl
emeansquar
e,i
tisaunbi
ased
est
imat
orofpopul
ati
onmeansquar
e

Thedegr
eesoff
reedom f
oratst
ati
sti
cisn-
1

KAKATI
YASERI
ES Page12
 Thetdi
str
ibut
ioni
ssy
mmet
ri
cal
dist
ri
but
ion

Appl
icat
ionsoftdi
str
ibut
ion

Totesti
fthesampl
emeansi
gni
fi
cant
ly di
ff
erf
rom t
hehy
pot
het
icalv
alue
oft
hepopulat
ionmean
Tot
estt
hesi
gni
fi
canceoft
hedi
ff
erencebet
weent
wosampl
emeans
Totestt
hesignif
icanceofanobser
vedsampl
ecor
rel
ati
onCoef
fi
cientand
sampler
egr
essioncoeffi
cient
Tot
estt
hesi
gni
fi
canceofobser
vedpar
ti
alcor
rel
ati
oncoef
fi
cient
Pai
redTt
esti
susedt
otestt
heequal
i
tyoft
wodependentsampl
emeans

Fst
ati
sti
c

I
fXandYar etwoi ndependentchi-
squarevar
iant
swi
thv
1andv
2degr
eesof
f
reedom r
espect
ivel
ythenFst at
ist
ici
sgivenby

X/
v1
F= f
oll
owsFwi
thv
1,v
2degr
eesoff
reedom
y/
v2

Therati
ooftwoindependentchi
-squar
evar
it
esdi
vi
dedbyt
her
espect
ivedegr
ees
off
reedom i
sknownasFst ati
sti
c

Appl
icat
ionsofFdi
str
ibut
ion

Ft
estusedt
otestt
heequal
i
tyoft
wopopul
ati
onv
ari
ances
Ft estisusedt
otestt
hesi
gni
fi
canceofobser
vedmul
ti
plecor
rel
ati
on
coeffi
cient
Ft
esti
susedt
otestt
hel
i
near
it
yofr
egr
essi
on
Ft
estusedf
ort
est
ingt
hesi
gni
fi
canceofobser
vedsampl
ecor
rel
ati
onr
ati
o
Ft
esti
susedf
ort
est
ingofequal
i
tyofsev
eral
means
Theor
yOfEst
imat
ion.

Whi l
einferr
ingstat
isti
call
yaboutapopul at
ionparamet eronthebasisofa
random sampledr awnf r
om t hepopulati
on,wef acet wo diff
erentty
pesof
probl
ems.I nthef i
rstsit
uation,thepopul
ationunderdi scussioniscompl et
ely
unknownt ousandwewoul dl i
ketoguessaboutt hepopulationparameter(s)
from ourknowl edge aboutt he sample observ
ations.Thisi s known as
Esti
mationofpopul at
ionparameters.

I
nthesecondsi
tuat
ion,somei
nfor
mat
ionaboutt
hepopul
ati
oni
sal
ready
KAKATI
YASERI
ES Page13
avai
labl
eandwewouldli
ketoveri
fyhowfart
hatinfor
mationisv
ali
donthebasi
s
oftherandom sample drawnf rom t
hatpopulati
on.Thissecondaspecti
s
knownastest
sofsigni
fi
cance.

Poi
ntEst
imat
ion:
Letusconsiderapopulati
oncharacter
isedbyanunknownpopul at
ionpar amet er
θ where θ could be popul
ati
on mean orpopul ati
on vari
ance ofa nor mal
popul
ati
on.Inordertoesti
matetheparamet er,wedrawarandom sampl eofsi ze
nfrom thepopulat
ionandletusdenot ethesampl eobser
v at
ionsby,x1 , x2 ,x3
,………. .xn.We areinsearchofast ati
sti
cT,whi chisaf uncti
onofthesampl e
obser
vationsx1,x2,x3,xn , thatcanesti
mat etheparameter.
Tisknownt obeanestimatoroftheparamet erθ

Tisdescri
bedas,tobemoreprecise,apointest
imatorofθasTr epr
esentsθby
asingl
evalueorpointandt
hev al
ueofT, asobtai
nedf r
om thesample,
isknown
as pointesti
mate.The pointest i
matorofpopul ation mean,popul at
ion
var
ianceandpopulati
onproporti
onar et
hecor r
espondingsamplest
atist
ics.
Thecr
it
erionf
orgoodest imat
orar
e
a) Unbi
( asedness
b) Con
( sistency
c) Ef
( f
iciency
d) Suf
( ficiency
a) Unbi
( asedness:
Astat
ist
icTisknowntobeanunbi asedest
imat
oroftheparameterθif
theexpect
ati
onofTi sθ.ThusTisunbi
asedofθifE(T)=θ.
I
fabovedoesnotholdt
henTisknownt obeabiasedesti
matorofθ.The
biasi
sknown t obeposi
ti
vei
fE(T)–θ>0andnegat i
ve ifE(T)–θ<0.

Astat
ist
icTisknowntobeamini
mum v
ari
anceunbi
asedesti
mator(
MVUE)ofθ
i
f(i)Tisunbiasedforθ and(i
i
)Thastheminimum vari
anceamongal
lthe
unbi
asedest
imator
sofθ.

Forapar
ameterθ,ther
eexist
sagoodnumberofunbi
asedstati
sti
csandthati
s
whyunbi
asednessisconsi
der
edalongwi
thmi
nimum v
ari
ance.Thesampl
emean
i
sanMVUEf orpopul
ati
onmean.Thesampl
est
andar
ddevi
ati
on

Ex
ampl
es:
Sampl
emeani
sunbi
asedest
imat
orofpopul
ati
onmean
Sampl
ev ar
iancei
snotaunbiasedesti
matorofpopul
ati
onvari
ance.Sampl
e
meansquareisaunbi
asedesti
matorofpopul
ati
onvar
iance.
I
fTisunbi
asedest
imat henT2i
orofθt sabiasedest
imatorofθ2.
Thesampleproport
ionp isanunbiased est
imatorforthepopul
ati
on

KAKATI
YASERI
ES Page14
 propor
ti
onP.
(
b) Consi
stency
Astati
sticTisknownt obeconsi
stentest
imatoroftheparamet
erθ,ift
he
diff
erencebet weenTandθcanbemadesmal lerandsmal l
erbytakingthe
sampl esi
zenl argerandlar
ger.or
Tisconsistentest i
matorofθifP[IT-θI<ɛ]>1- ᶯwher eɛandᶯar ever
ysmal l
realnumber s.i
.e.Tisconsist
entest
imatorofθifTconvergestoθinprobabil
i
ty.
or
Tisconsi st
entforθifE(T)→θandV( → 0asn→ ∞
T)
Thesamplemean,sampleSD andsampl epr
opor
tionareal
lconsi
stent
est
imat
orsf
ort
hecorrespondi
ngpopul
ati
onpar
ameters.
Al
lconsi
stentest
imat
orsneednotbeunbi
asedest
imat
ors.
SampleStandar
ddeviat
ionisconsist
entofpopulat
ionstandar
ddev
iat
ion
butnotunbi
asedest
imatorofpopul
ati
onstandar
ddev i
ati
on.
(
c)Ef
fi
ciency
A stati
sti
cT isknownt obeanef f i
cientesti
matorofθ ifT hast he
mini
mum st andar
d er r
or(orVariance)amongal ltheesti
matorsofθwhent he
samplesizeiskeptfi
xed.Likeunibiasedestimators,morethanoneconsi
stent
est
imatorexistsforθ.Tochooset hebestamong t hem,weconsiderthat
est
imatorwhichisbot
hconsistentandeffi
cient.

LetT1 andT2 aretwoesti

mator
sofθ,ThenT1 i
ssai
dtobemosteffi
cient
est
imatort
hatT2,ifV(T1)<V(T2)andt
heef
fi
ciencyofT1ov
erT2 i
sdenot
edbyE
andgi
venbyE=V( T1)/
V(T2)
Ei
sal
way
slesst
hanorequal
to1.
Thesamplemean i
sbot
hconsi
stentandef
fi
cientest
imat
orf
ort
he
popul
ati
onmean.
(d)Suf
fi
ciency
.
Astati
sti
cTisknowntobeasuf
fici
entesti
mat orofθi
fTcontai
nsal
lthe
i
nformat
ion aboutθ.Howev
er,t
hesuf
fi
cientstat
ist
icsdonotexi
stsforal
lthe
paramet
ers.Or
Let x1 ,x2 ,x3 ,………. .xn bear andom sampl et akenfrom apopul ati
onwith
parameterθ.T(1x , x2 ,x3 ,… …… .
.xn)bestati
sti
cusedt oestimateanunknown
parameterθ, t
henT(x1 ,x 2 ,x3 ,……… .
.xn)i
ssai dtobesuf fi
cientstat
ist
icofθ,i
f
theconditi
onalprobabili
tyofT( x1 ,x2 ,
x3,… ……..xn)giv
enθi sindependentofθ.
The sampl
e mean i
s a suf
fi
cientest
imat
orf
ort
he cor
respondi
ng
popul
ati
onmean.
Ʃxi
ssuf
fi
cientf
orpi
nber
noul
idi
str
ibut
ion
ForaNor
mal
dist
ri
but
ion,
Ʃxi
ssuf
fi
ci orμandƩX2i
entf ssuf
fi
ci orσ2
entf
Fact
ori
zat
ionTheor
em i
susef
ult
ofi
ndsuf
fi
cientst
ati
sti
cs.I
twasgi
ven
KAKATI
YASERI
ES Page15
 byFisher-
Neyman
Int
er valEsti
mat i
on
I
nsteadofestimatingaparameterθbyasi nglev al
ue,wemayconsi deran
i
ntervalofv al
ueswhi chi ssupposedt o contai
nt hepar ameterθ.Ani nter
val
esti
mat ei
sal waysexpressedbyapai rofunequalr ealv al
uesandt heunknown
paramet erθl i
esbetweent heset
wov alues.Hence,ani ntervalest
imati
onmaybe
definedasspeci fyi
ngt wov al
uesthatcontainstheunknownpar ameterθonthe
basisofar andom sampl edrawnfrom t
hepopul at
ioninallprobabil
it
y.
Onthebas
isofar andom sampl
edrawnf r
om thepopul
ati
onc
haract
eri
sedbyan
unk
nownparamet
erθ,letusfi
ndtwostat
ist
icsT1andT2sucht hatP(T1<θ)=
α1
P(T2>θ)=α2,

foranyt
wosmall
posi
ti
vequanti
ti
esα1andα2.
Combini
ngt
hesetwocondi
ti
ons,wemaywrit
e
P(T1≤θ≤T2)=1–αwher eα=α1+α2

Itimpliesthattheprobabi
li
tythattheunknownpar amet
erθliesbet
ween
thetwost at
isti
cT1andT2i s(1–α).Thei nter
val[T1,T2 ],T1 <T2,i s
knownas100( 1– α)% conf i
dencelimit
s toθ.T1 i sknownast hel
ower
conf
idenceli
mi t(LCL)andT2isknownasupperconf idenceli
mit(
UCL)toθ.

(1 – α)i st er
med as conf idence coeffi
cientcor responding tot he
confidencei nterv
al[T1 ,T2 ]. Thet erm“confi
dencei nt erval ”hasitsori
ginin
thefactt hatifweselectα=0. 05,t
henwef eelconfi
dentt hatt heinter
val[T1 ,T2
],woul dcont aintheparameterθin(1–α)%or( 1–0. 05) %or95percentof
casesandt heamountofconf i
dencei s95percent.Thisf urthermeanst hatif
repeatedsampl esofaf i
xedsizearetakenf r
om thepopul at ionwi ththeunknown
paramet erθ,t henin95percentoft hecases,thei nt
er v al[ T1 ,T2 ]woul d
containθandi ntheremaining5per centofthecases, i
twoul df ai
ltocontai
nθ.

Pr
act
icebi
ts:
1) Par
amet
ersar
ethoseconst
ant
swhi
choccuri
n:
a)sample b)probabi
li
tydensit
yfuncti
on
c)aformula d)noneoftheabov e
2)Esti
mat i
onandest i
mat orare:
a)synonyms b)di ff
erentc)r el
atedtopopulati
on d)noneoftheabov
e
3)Anest i
mationisconsideredtobet hebestifi
tsdist
ri
buti
onis:
a)conti
nuous b)di
scret
e
c)concent
ratedaboutt hetrueparameterval
ue d)Nor
mal
4)Forabernouli distr
ibuti
on
a)mean<var b)mean>v ar c)mean=v ar d)mean<=v ar

KAKATI
YASERI
ES Page16
5) Theoutcomesofanex per imentcl assi fiedassuccess A&f ailureA’wi llfoll
ow a
Ber noul lidi stribut i
oni ff:
a)P( A) =1/ 2 b)P( A) =0 c)P( A) =1d)P( A)r emai nconst antinal lt
ri
als
6)Themeanandv ar ianceofBi nomi al distancear e8&4t henP( x=1)=?
a)1/212 b) 1/24 c) 1/26 d) 1/28
7)Manuf act ur erpr oducesswi tchesandex periencet hat2%swi t
chesar edef ectiv
e. t
he
pr obabi lityt hati naboxofsoswi tches, therear eatmostt wodef ectiv
eis
a)2. 5e-1 b)e-1 c)2e-1 d)none
8)Xf oll
owsb( 3,1/ 2)&Yf ollowsb( 5,1/ 2)t henP( x+Y=3)i s
a)7/ 16 b) 7/32 c) 11/ 16 d)none
9)LetXf ol l
owsP( 1)&Yf ol l
owsP( 2)t henP( x+y <3)i s
-
3 -
3 -
3
a)e b)3e c)4e d)8. 5e-3
10)IfXf ollowsB( n, p)t henY=n- xis.
a)B( n,1) b)B( n,x) c)B( n,p) d)B( n,q)
11)A di stribut ion,I nwhi chmeani sequal tov ariance.
a)B. D b)P. D c)N. B. D d)H. G.D
12)Themgfofber noul idistancei s
a)( q+pet) b) ( q+pet) 2
c) ( q+pet) n
d)None
13)Fornor mal distribut i
on, poi nt sofi nflect i
onar e
a)μ±σ b)±σ c) μ d)μ±6σ
14)Inbi nomi al distancei fn=1t henr educest o
a)Ber noul i b)NBD c)PD d)nor mal
15)IfXf ollowsN( 8, 64) ,thest andar dnor mal variateZi s
X-8 X- 64 X- 8 X-64
a) b) c) d)
8 8 64 64
16)Nor mal dist ribut i
onwasi nv ent edby
a)Lapl ace b)De- mov i
re c)Gauss d)al ltheabov e
17)Theout comesoft ossi ngacoi n3t imesar eav ariableoft ype
a)Cont inuous.B)di scr ete.C)notanotb.D)bot haandb
18)Thehei ghtofaper soni s
a)Cont i
nuousr andom v ar i
abl e b)di scr eter andom v ar i
able c)bot h d)none
19)Fort wov ariabl esXandY cov (ax+b, cy +d)=. .....
...
.a,c>0
a)Cov (x,y) b) abcdCov (x ,
y) C)acCov (x ,
y) d)acCov (
x,y)+bd
20)Thet ypeofest imat esar e
a)Poi ntest imat e b)int erval estimat es
c)est i
mat i
onofconf idencer egi on d)al l theabov e
21)Ifanest i
mat orTofpopul ationpar amet erconv ergest o θi nPr obabi l
it
yasnt endsto
infinity, i
ssai dt obe:
a)Suf ficient b)ef ficient c)consi stent d)unbi ased
22) Theest imat or‘ sampl emean’ ofpopul ationmeani s:
a)anunbi asedest imat or b)aconsi stentest i
mat or
c)bot h( a)and( b) d)nei ther( a)nor( b)
23)Iftheexpect edv alueofanest imat ori snotequal toi t
spar amet erθi tissaidt obea:

KAKATI
YASERI
ES Page17
 a)Unbi asedest imat or b)biasedest imator
c)consi stentest imat or d)noneoft heabov e
24)Bi
asofanest imat orcanbe:
a)Posi ti
v e b)negat ive c)ei therposi t
iveornegat ive d)al wayszer
o
25)Ahy pothesi smaybecl assi fiedas:
a)Si mpl e b)composi te c) nul
l d)al lt
heabov e
26)Thehy pot hesi sUndert esti s:
a)Si mpl ehy pot hesi s b)alternat ivehy pothesis
c)nul lhy pot hesi s d)noneoft heabov e
27)Whet herat esti sonet ai l
edort wot ai leddependson
a)Al ternat ivehy pot hesis b)composi t
ehy pothesis
c)nul lhy pot hesi s d)si mpl ehy pot hesi s
28)A wr ongdeci sionaboutH0l eadst o:
a)Oneki ndofer ror b)t woki ndsofer ror
c)thr eeki ndsofer ror d)fourki ndsofer ror
29)Powerofat esti srel atedt o
a)Ty peIer ror b)typeI Ierror
c)t ypesIandI Ier r
or sbot h d)noneoft heabov e
30)Levelofsi gni fi
cancei sthepr obabi li
tyof :
a)Ty peIer ror b)typeI Ierror
c)t ypesIandI Ier r
or sbot h d)noneoft heabov e
31)Areaoft hecr iticalr egi ondependson:
a)Si zeoft ypeIer ror b)sizeoft y peIerror
c)v alueoft hest at isti
c d)numberofobser vations
32)Sizeofcr iti
cal regioni sknownas:
a)Poweroft het est b)si zeoft ypeIer ror
c)crit
ical valueoft het estst atisti
cs d)si zeoft het est
33)Degreesoff reedom i sr elatedt o:
a)Noofobser v at i
onsi nat est b)noofi ndependentobser vati
ons
c)bot h d)noneoft heabov e
34)Accept i
ngH0whenH0i sf alsei sknownas
a)Ty peIer ror b)typeI Ierror
c)t ypesIandI Ier r
or sbot h d)noneoft heabov e
35)Zv al
uef orat wot ai ledt estat5%L. O.S
a)1.96 b)2. 52 c)1. 68 d)3
36)Squareofst andar dnor mal var i
atei sknownas
a)tst atistic b)Fst at istic c)‫א‬2st atistic d)None
37)Whet herat esti sonesi dedort wosi dedi sdeci dedby
a)Al ternat ivehy pot hesis b)nul lhy pot hesi
s
c)si mpl ehy pothesi s d)Composi tehypot hesis
38)Awr ongDeci sionaboutH0l eadst o
a)Oneki ndofer ror b)t woki ndsofer ror
c)thr eeki ndsofer ror d)none
39)Powerofat estr elat edt o
a) Ty peIer r orb)Ty peI Ier ror c)Ty peIandI Ierrord)none

KAKATI
YASERI
ES Page18
 40)Thesi z eofRej ectionr egioni s
a) α b)β c)1- β d)1- α
41)Fact ori
sat i
ont heorem f orsuf f
iciencyi sknownas:
a)Rao- blackwel ltheor em b)crammerRaoTheor em
c)FisherNey manst heor em d)none
42)sampl erv ariancei sa…………est imat orofpopul ationvari
ance
a)unbi asedb)consi stentc)bot hd)None
43)ifTi sunbi asedest imat orofθt henT2i s……….Est imatorofθ2
a)Biasedb)unbi asedc)consi stentd)none
44)Fact ori
z ati
ont heorem pr ovides………. .
stati
stics
a)Suff i
cient b)ef ficient c)consi st
entd)unbi ased
2
45)InNor mal distributionsuf fi
cientst ati
sticforσ i s
a)ƩX b)ƩX2 c)s2 d) x̅
46)Reject i
ngHowhenHoi st ruei s
a)TypeIer ror b)Ty peI Ierr
or c)TypeIandI Ierr
ord)none
47) Accept i
ngH0whenH1i st ruei s
a)Ty peIer ror b)Ty peI Ierr
or c)TypeIandI Ierr
ord)none
48)pmfp( x)sat isfies
a)nonnegat i
vef oranyX b)Tot alofP( x)is1 c)bot h d)none
49)Poi ssondi st r
ibutionhas
a)+v eskewness b)–v eskewness c)noskewness d)anyofabov e
50)binomi al distri
but i
onhas
a)+v eskewness b)–v eskewness
c)noskewness d)itdependsonpandq
Ans:
1-b 2-b 3-c 4-b 5-d 6-a 7-a 8-b 9-d 10-
d
11-b 12-a 13-a 14-a 15-a 16-d 17-b 18-a 19-c 20-
d
21-c 22-c 23-b 24-c 25-d 26-c 27-a 28-b 29-b 30-
a
31-a 32-b 33-b 34-b 35-a 36-c 37-a 38-b 39-b 40-
a
41-c 42-b 43-a 44-a 45-b 46-a 47-b 48-c 49-a 50-
d

Pr
obl
ems:

Pr
obabi
li
tyf
unct
ion:

1)PmfofXi
sgivenbyf
oll
owi
ngt
abl
e.
X 1 2 3 4 5

P(
x) k k/
2 k/
3 k/
4 k/
5

i
)fi
ndK i
i
)fi
ndP(
X≤4)i
i
i)f
indp(1<x<4)

Answer:i
)wehav
etoat
alpr
obabi
l
ityi
suni
ty

KAKATI
YASERI
ES Page19
 ƩP(X)=1 → k+k/ 2+k/3+k/4+k/
5=1
137k/60=1 → k=60/ 137
i
i)P(
X≤4)=1-P( X=5)=1-12/137=125/ 137
i
ii
)p(1<x<4)=P( X=2)+P( X=3)=30/ 137+20/137=50/137.
2
2)Thepdfofacont i
nuousr andom vari
ableisgiv
enbyf(
x)=kx 0<X<1f
indk?
Answer:Wehav etotalprobabil
i
tyisunity
1 2
x∫
f(
x)dx=1 → 0∫kxdx=1
3 1
[
kx/ 3]0 =1 → K[( 1/3)
-(0/3)]=1→k/3=1→ k=3.

Bi
nomi
aldi
str
ibut
ion:
1)Fi
ndt hePr obabi li
tyofgett
ingof3headswhen6coi nsaretossed
si
multaneousl y?
Answer:Let ,Xdenot esnumberofheadswhen6coi nsar etossed
NowXi sbinomi alVar
iatewi thn=6andp=0. 5. andwehav etof
indp(
X=3)
n x n-x
p(x)= Cxpq ;X=0, 1,2..
...
...
n
6!
p(X=3)=6C3( 0. 3
5)(0.
5)6-3
= (0.
015625)=0. 3125.
3!3!
2)Si
xcoi nsar etossedonce.Fi ndt hepr obabi
li
tyofdrawing
i)atmost3heads
ii
)atl east3Heads
ii
i)noheads
iv)atleastonehead
Ans::Let ,
Xdenot esnumberofheadswhen6coi nsaretossed
NowXi sbinomi alVar
iatewi thn=6andp=0. 5.
i)pr obabi l
i
t i
esofatmost3heads=p( X≤3).
p(
x)=nCxpxqn-x ;X=0, 1,
2..
..
..
..
n
p(
X≤3)=P( X=0) +P(x=1)+P(X=2)+P( X=3).
6 0 6-
0 6 1 6-
1 6 2 6-
2 6 3 6-
3
= C0( ½)( 1/
2) + C1( ½)( 1/2) + C2(½)(1/
2) + C3(
½) (1/
2)
6 6
=( 1/2) {C0+6C1+6C2+6C3}=0. 65625

i
i
)pr
obabi
l
iti
esofatl
east3heads=p(
X≥3)=P(
X=3)
+P(
x=4)
+P(
X=5)
+P(
X=6)
=6C3(
3 6-
3 6 4 6-
4 6 5 6-
5 6 6 6-
6
½) (1/
2) + C4(
½) (1/
2) + C5(
½) (1/
2) + C6(
½) (1/
2)
6 6
=(
1/2){C3+6C4+6C5+6C6}=0. 65625

X=0)=6C0(
0 6-
0 6
i
i
i)Pr
obabi
l
ityofnoheads=P( ½) (1/
2) =(
1/2)=0.
0156

i
v)Pr
obabi
l
ityofatl
eastonehead=1-
P(X=0)=1-
0.0156=0.
984.

3)Themeanandv ari
anceofbi
nomial
dist
ri
buti
onar
e4and3r
espect
ivel
yfi
ndi
ts
parameter
s
Ans:Inusualnot
ati
onswearegiv
ennp=4, npq=3

KAKATI
YASERI
ES Page20
 Vai
ance 3
Wehaveq= =
mean 4
Nowp=1- q=1-(3/4)=1/
4
Alsowehavemean=4i .
enp=4→ n( 1/4)=4→ n=16
Theparamet
ersofbinomial
dist
ri
but
ionaren=16andp=½

4)Forabinomialvari
atemeanis10andst
andar
ddev i
ati
onis5.Fi
ndp
Ans:Wear egivenmean=10andSD=5i .
evari
ance=25
Since,f
orabinomi al
var
iat
emeanisal
waysgreatert
hanvar
iance,
thegi
ven
dataisincor
rect.

5)Forabinomi alvari
at emean10andv ari
ance5.Findit
smode.
Ans:Mode:casei )if(n+1)p=m i sainteger
,m andm- 1aremodeofbi nomial
v ari
able.
Caseii
)if(n+1)p=m i snotai nt
egerandm=i+f, t
hentheintegralpartofm i
s
mode
Wear egi v
enmean=10andv ar
iance=5
Vaiance
Wehav eq= =5/10=½ , si
nceq=½, palso½
mean
Alsowehav emean=10→ np=10→ n/ 2=10→ n=20.
Now( n+1)p=21* 1/
2=10. 5isnotaintegerandtheintegralpart10ismode.

Poi
ssonDi
str
ibut
ion:
1)I
f3%ofitemsproducedbyamachi near
edefect
ive.Fi
ndt
hepr
obabi
l
ityt
hati
na
sampl
eof100items,exactly5i
temsaredef
ecti
ves.
Ans:Wearegiv
enn=100,p=0.
03andλ=np=100*.
03=3
WehavetofindP(X=5)
Forpoi
ssonv ar
iat
ewehav e p(
x)=e-λλx/x!;i
fX=0,
1,
2,
3,
..
..
..
..
.∞.
-3 5
P(
X=5)=e 3/5!=( 0.
04979*243)
/40=. 10

2)ForaPoi ssonvari
abl
eX, P(
X=1)=P(X=2)
.FindmeanandSDofX
-
λ x
Ans:Forpoi ssonvar
iat
ewehav e p(x)=e λ/x !;i
fX=0,1,
2,
3,
..
..
..
..
.∞.
Wear egivenP(X=1)=P(X=2)
e-λλ1/1!= e-λλ2/2!→ 2*e-λλ1=e-λλ2
→2λ=λ2→ λ=2
Wehav emeanofpoissonvar
iat
eisλ, mean=2
AndSDofpoi ssonvari
ateis λ= 2.

KAKATI
YASERI
ES Page21
3)I
f3%ofi t
emspr oducedbyamachi nearedefecti
ve.Fi
ndtheprobabi
li
tyt
hatina
sampleof100items, i
)ex actl
y2itemsar edefecti
vesandii
)noitem i
sdefect
ive
Ans:Wear egi
venn=100, p=0. 02andλ=np=100* .02=2
Wehavetof i
ndi)P(X=2)i i
)P( X=0)
Forpoi
ssonvariatewehav e p(x)=e-λλx/x!
;i fX=0, 1,
2,
3,
..
..
..
..
.∞.
-2 2 -2
P(
X=2)=e 2/2!=2e
X=0)=e-220/0!=e-2.
P(

4)Abookcont ai
ns100mi spri
ntsdistr
ibut
edr andomlythroughouti t100pages.Whatis
theprobabil
it
yt hatapageobser v edatrandom contains2orl essmi st
akes?
Ans:Wear egi
ven, Abookcont ains100mi sprint
sdistr
ibutedrandoml ythroughouti
t100
pagesi.eaveragemi st
akesperpagei s1.Λ=1
Wehav etofindP( X≤2)
-
1 0 -
1 1 -1 2
P(X≤2)=P(X=0) +P(X=1)+P(X=2)=e 1/0!+e 1/1!+e 1/2!
-
1 5
=e {1+1+( 1/2)}=
2e

NormalDi
str
ibut
ion:
AreaPr
opert
yofnormal
distr
ibut
ion.
i
) P[μ- σ≤X≤μ+σ]=0. 6826=P[- 1≤Z≤+1].
i
i) P[μ-2σ≤X≤μ+2σ]=0. 9544=P[- 2≤Z≤+2].
i
ii
) P[μ-3σ≤X≤μ+3σ]=0. 9973=P[-3≤Z≤+3] .

 P[0≤Z≤+1]= P[- 1≤Z≤0]

.=0.
6826/2=0.3413
 P[0≤Z≤+2]= P[- 2≤Z≤0]
.=0.9544/
2=0.4772
 P[0≤Z≤+3]= P[- 3≤Z≤0]
.=0.9973/2=0.
49865
 P[
Z≥0]=P[
Z≤0]=0.
5

1)l
etXisanor malvariablewi thmean70andv ari
ance4.Findthepr
obabi
l
iti
es
i)X≥68i i
)X≥72i ii
)X≥74i v)P(68≤X≤74)
Ans:Wear egivenmeanμ=70andVar ianceσ2=4→ σ=2
i)P(X≥68)=P( Z≥(68- 70)/2)=P( Z≥-1)=P[ -1≤Z≤0]+P[Z≥0]
=0.
3413+0. 5=0. 8413.
ii
)P(X≥72)=P( Z≥( 72-70) /2)=P(Z≥1)=P[ Z≥0]-
P[0≤Z≤1]
=0.5-.
3413=0. 1587
ii
i)P(X≥74)=P( Z≥( 74-70) /
2)=P( Z≥2)=P[ Z≥0]-P[0≤Z≤2]
=0.5-
.4772=0. 0228
iv)P(68≤X≤74)=P{ (68- 70)/2≤X≤( 74-70) /
2}=P[-1≤Z≤2]
=P[- 1≤Z≤0]+P[0≤Z≤+2]=0. 3413+0. 4772=0.8185.

KAKATI
YASERI
ES Page22