Basi

cSt
ati
sti
cs
1.1. Def initionsandcl assi ficationofSt atistics
Definit
ions:
Stati
sticsi sdef i
neddi ff
er ent lybydi ff
erentaut hor sov erper i
odoft ime.I nt heol den
daysst atisticswasconf inedt oonl yst ateaf f
ai r
sbuti nmoder nday si tembr aces
al
mostev er yspher eofhumanact i
vity.Ther efor e,anumberofol ddef ini
t i
ons,whi ch
wereconf inedt onar r
owf i
eldofenqui ry,wer er epl acedbymor edef initions,whi char e
muchmor ecompr ehensi veandex haust i
ve.
Wecandef inei tint wosenses
 I nt hepl ur alsense:st atisticsar et her awdat at hemsel ves( Numer icalf act s),l i
ke
statisticsofbi rths,st atist i
csofdeat hs,st atisticsofst udent s,st atisticsofi mpor t
s
andexpor ts,etc.
 I nthesi ngul arsense:st at i
sticsisconcer nedwi thscientificmet hodsf orcol lecting,
organi zi ng,summar izing,pr esentingandanal y zingdat aaswel lasder i
vingv ali
d
concl usi onsandmaki ngr easonabl edeci sionsont hebasisoft hisanal ysi
s.
Classifi
cat ions:
Dependi ngonhow dat acanbeusedst ati
st i
csi ssomet i
mesdi vi
dedi nt ot womai n
areasorbr anches.
1.Descr iptiveSt ati
stics:
 i sconcer nedwi t
hsummar ycal culations, gr aphs, chartsandt abl es.
 Gener allychar act erizesordescr i
besasetofdat ael ement sbygr aphi cally
displ ay i
ngt hei nfor mat ionordescr i
bi ngi tscent r
alt endenci esandhow i tis
distr i
but ed.
 I ndescr iptivest at i
sticsourobj ectivei st odescr i
beagr oupofdat at hatwe
hav e‘ inhand’ i.
e.dat at hatar eaccessi blet ous.
 Wear enoti nterest edi not herdat at hatwear enotgat hered.
Exampl e:I fwecal culat et heav eragenumberofcomput erf r
om 2017t o2021as
t
henourwor
kbel
ongst
othe

domainofdescr i
pt i
vestatisti
cs.
Ifwesayt hatther ewasani ncreaseof724comput ersfrom 2017t o2021,t hen
againthi
sbel ongst othedomai nofdescr ipti
vest at
isti
cs.
2. I nfer
entialSt atisti
cs:consi sts ofgener al
izing from sampl es t o populations,
per f
orming est i
mat ions and hy pothesist est
s,det ermini
ng r elat
ionships among
variabl
es,andmaki ngpr edicti
ons.
Exampl e:Supposewewantt ohav eani deaaboutt heper centageofi l
li
teratesi nour
country.Wet akeasampl efrom thepopul at
ionandf indthepr oporti
onofi l
li
teratesinthe
sampl e.Thissampl epr oporti
onwi t
ht hehel pofpr obabi l
ityenablesust omakesome
i
nferencesaboutt hepopul ati
onpr oportion.Thisstudybel ongstoi nf
erentialstatist
ics.
1.2. St agesi nSt at isti
calI nvest igation
Stat i
sti
calI nvesti
gat i
on i ncludes f ormulati
ng t he pr oblem,and t hen col l
ect i
ng,

1
organizingandcl assi fy ing,pr esent i
ng,anal yzingandi nt erpret i
ngofst atisticaldat a.
Let’
sseeeachst agei ndet ail
I.For mul atingt hepr obl em:f irstresear chmustemanat ei ft herei sapr oblem.At
thi sst aget hei nv est igat ormustbesur et ounder standt hepr oblem andt hen
for mul at eitinst at i
sticalt erm.Cl arifyt heobj ectivesv er ycar efully.Ther efore, the
firstst agei nanyst atist ical investigat i
onshoul dbet o:
 Getacl earunder st andi ngoft hephy si
cal backgr oundt othesi tuat i
onunderst udy ;
 Cl arifyt heobj ectiv es;
 For mul at et heobj ect ivei nst atisti
cal t
erms
I.Pr
I opercol lectionofdat a:i nor dert odr aw v ali
dconcl usions,i tisi mpor tant‘good’
data.Dat aar egat her edwi t hai mt omeetpr edet ermi neobj ect i
ves.I not herwor ds,
thedat amustpr ov ideanswer st opr oblems.Thedat ai t
sel ff ormt hef oundat ionof
statistical anal ysesandhencet hedat amustbecar eful l
yandaccur at el
ycol l
ected.I n
section1. 6wewi llseet hemet hodsofdat acollect i
on.
I
I.Or
I gani zat i
onandcl assi ficat ionofdat a:int hi
sst aget hecol lect eddat aor gani zedi n
asy st emat icmanner .Thatmeanst hedat amustbepl acedi nr elationt oeachot her .
Thecl assi ficationorsor tingoutofdat ais, byi t
self ,
aki ndofor gani zat i
onofdat a.
V.Pr
I esent at i
onofdat a:Thepur poseofput tingt heor gani zeddat ai ngr aphs,char t s
andt abl esi st wo- fol d.Fi rst,i ti sav i
sualwayt ol ookatt hedat aandseewhat
happenedandmakei nter pret ations.Second,i ti susual l
yt hebestwayt oshowt he
datat oot her s.Readi ngl ot sofnumber si nt het extput speopl et osl eepanddoes
li
ttl
et oconv eyinfor mat ion.
V.Anal ysesofdat a:ist hepr ocessofl ooki ngatandsummar izingdat awi tht hei ntent
to ext ractusef uli nf ormat ionanddev elopconcl usions.Dat aanal ysisi scl osel y
rel
at edt odat ami ning,butdat ami ningt endst of ocusonl argerdat aset s,wi thless
emphasi sonmaki ngi nfer ence, andof t
enusesdat at hatwasor iginal l
ycol lectedf or
adi fferentpur pose.I nt hisst agedi fferentt ypesofi nfer ential statisticalmet hodswi l
l
appl
y.Fori
nst
ance,
hypot
hesi
stest
ingsuchas t
estofassoci
ati
on.
VI.I nterpretati
onofdat a:interpretati
onmeansdr awi ngv ali
dconcl usi
onsf r
om dat a
whi chf ormt hebasi sofdeci sionmaki ng.Cor r ectinterpret
ati
onr equir
esahi gh
degr eeofski llandexper i
ence.
Not ethat:Anal ysesandi nterpret
at i
onofdat aaret hetwosi desoft hesamecoi n.
1.3. Def i
nitionofSomeTer ms
Inthissect i
on, wewi l
ldefi
net hoset ermswhi chwi l
lbeusedmostf requentl
y.Thesear e:
Data:ar ethev alues( measur ement sorobser vati
ons)t hatthev ari
ablescanassume.OR
Factsorf i
guresf rom whicht heconcl usioncanbedr awn.
Raw Dat a:Dat asheet sar ewher et hedat aar eoriginall
yr ecorded.Origi
naldataar e
call
edr awdat a.Dat asheet sareof tenhanddr awn,butt heycanal sobepr int
outsfrom
databasepr ogramsl ikeMi crosoftExcel .
Popul ati
on:Thet otali
tyofal lsubjectswi thcer t
aincommonchar acteri
sti
csthatar e
beingst udiedinaspeci fi
edt i
meandpl ace.

2
Sampl e:Isa por t
ion ofa popul ation which i ssel ected usi ng somet echniqueof
sampling.Sampl emustber epresent ati
veofthepopul ati
onsot hatitmustbesel ected
byanyoft hedev elopedtechnique.
Sampl i
ng:I st hepr ocessofsel ect i
nguni ts( e.g.,peopl e,househol ds,or ganizations)
from apopul ationofi nter
estsot hatbyst udy i
ngt hesampl ewemayf airl
ygener al
ize
ourresultsbackt othepopul at
ionf r
om whi cht heywer echosen.Ther earet wot ypesof
samplingt echni quesnamel yr andom sampl i
ngt echniqueandnon- r
andom sampl ing
techni
que.
Random sampl ing technique orpr obabil
i
tysampl ing t echnique gives a non-zer o
chancef oral lel ementst obei ncludedi nt hesampl e.I not herwor ds,therei sno
personalbiasr egardingtheselection.
Non-random sampl i
ng technique i s mostly known as non- pr
obabi l
i
ty sampl ing
techni
quesandi nt hi
scasenotal lel ementsofapopul ationhav eaknownchanceof
i
nclusionori fsomeout comeshav eazer ochanceofbei ngsel ectedasasampl e.
Sampl esize:Thenumberofel ement sorobser v ationtobei ncludedint hesampl e.
Paramet er:Anymeasur ecomput edf rom thedat aofapopul ati
on.
Exampl e:Popul ati
onsmean andpopul ati
onst andarddev iati
on
St
ati
sti
c:Anymeasur
ecomput
edf
rom t
hesampl
e.
Exampl
e:sampl
emean ,
sampl
est
andar
ddev
iat
ion
Survey:Acol lect i
onofquant i
tat i
vei nf or mationaboutmember sofapopul ati
onwhenno
specialcontrol isexer cisedov eranyoft hefact or sinfl
uencingt hev ari
ableofi nterest.
Sampl esur vey:Asur v eyt hatincl udeonl yapor tionoft hepopul ati
on.
Census:Acol lectionofi nfor mat i
onaboutev erymemberofapopul ati
on
Sampl esurveyhast hef oll
owi ngadv ant agesov ercensus
 Sampl esur veysav est imeandcost
 Hasgr eataccur acy
 Av oidwast ageofmat er ial
Vari
able:Av ariablei sachar acter isticorat tributet hatcanassumedi fferentv alues.
Vari
ables whose v alues ar e det er mi ned by chance ar e cal l
ed r andom v ari
ables.
Vari
ablesar eof tenspeci fi
edaccor di ngt ot heirt ypeandi ntendeduseandhence
vari
ablecanbecl assi f
iedi nt ot wonamel yqual itati
veandquant it
ativ
ev ariables.
 Aquant i
tat i
v ev ariabl eisnat urallymeasur edasanumberf orwhi chmeani ngf ul
ari
thmet icoper ati
onsmakesense.Exampl es:Hei ght,age,cr opy i
eld,GPA,sal ary,
temperat ure, area, airpollutioni ndex( measur edi npart
spermi ll
i
on) ,
et c.
 Qual i
tativev ariable:Anyv ari
abl et hatisnotquant i
tativei squal i
tati
v e.Qual it
at i
ve
vari
ablest akeav aluet hati soneofsev eralpossi blecat egor i
es.Asnat urally
measur ed,qual it
at ivev ariableshav enonumer icalmeani ng.Exampl es:Hai rcol or,
gender,fieldofst udy ,mar ital status, poli
tical affi
li
ati
on, statusofdi seasei nfecti
on.
Quant i
tat
ivev ariablescanbecl assi f
iedasdi scr eteandcont i
nuousv ari
abl e.Di screte
vari
ablescanassumecer tainnumer icalv alues.Thati s,t herear egapsbet weent he
possiblevalues.Suchas0,1,2. ..Itmaybecount ablefi
ni t
eorcount ablei nfinit
e.For

3
exampl et henumberofst udent si nacl assr oom, numberofchi ldrenaf ami ly .
Continuousv ariablecant akeanyv aluewi thinaspeci fiedi nt ervalwi t
haf i
ni t
eenough
measur ingdev ice.Nogapsbet weenpossi blev alues.Theyar eobt ainedbymeasur i
ng.
Forexampl e,consi dert hehei ght soft wopeopl enomat terhowcl osei tiswecanf i
nd
anotherper sonwhosehei ghtf allssomewher ebet weent het wohei ghtsi sacont inuous
vari
abl e.
1.4Appl i
cat i
onsofst at i
stics
 Apar tf rom hel pingel icitani nt elligentassessmentf rom abodyoff i
gur esand
fact s, statisticsi si ndi spensabl et ool f
oranysci entifi
cenqui ry-r
ightf rom t hest age
ofpl anni ngenqui r
yt ot hest ageofconcl usi on.Itappl iesal mostal lsci ences:pur e
andappl ied,phy si
calnat ural,bi ol ogical,medi cal,agr i
cul turalandengi neer i
ng.I t
also f inds appl i
cat i
ons i n soci aland managementsci ences,i n commer ce,
busi nessandi ndust ry.
 I nal mostal lfi
el dsofhumanendeav or.
 Al mostal lhumanbei ngsi nt hei rdai lyl i
fear esubj ect edt oobt aini
ngnumer ical
fact s.
 Appl icabl e i n some pr ocess e. g. invent i
on of cer tain dr ugs,ext ent of
env ironment al pol l
ut ion.
 I ni ndust ri
esespeci allyinqual i
t ycont rolarea.
I
. UsesofSt atist i
cs
 St at i
sticspr esent sf acti nt hef or m ofnumer icaldat a
 I tcondensesandsummar izesamassofdat ai nt oaf ew pr esent ableand
pr ecisef igures.
 I tf acilit
at escompar i
sonofdat a
 I thel psi nfor mul atingandt est inghy pothesi s
 I thel psi npr edi ctingf ut ur et rend
 I thel psi nfor mul atingpol i
ces.
I
I. Limi tationsofSt atistics
Stati
sticswi thal litswi deappl i
cat i
oni nev eryspher eofhumanact ivityhasi tsown
li
mitation.Someoft hem ar egi venbel ow
 St at isticsi snotsui tabl et ot hest udyofqual i
tativephenomenon:Si ncest atisticsi s
basi cal lyasci enceanddeal swi thasetofnumer icaldat a,itisappl icabl et ot he
studyofonl yt hese subj ect sofenqui ry,whi ch can be expr essed i nt er msof
quant itative measur ement s.As a mat teroff act ,qual it
ative phenomenon l ike
honest y,pov erty,beaut y,int ell
igenceet c,cannotbeexpr essednumer i
cal lyandany
stat i
st icalanal y siscannotbedi rect lyappl iedont hesequal i
tati
vephenomenons.
Nev erthel ess,st atisticalt echni quesmaybeappl i
edi ndi r
ect lybyf i
rstr educi ngt he
qual itat i
veexpr essi onst oaccur at equant it
ativet erms.Forexampl e,thei ntell
igence
ofagr oupofst udent scanbest udi edont hebasi soft heirmar ksi napar ti
cul ar
exami nat i
on.
 St at istics does notst udy i ndi vidual s:St ati
st i
cs does notgi ve any speci f i
c

4
 impor tancet ot hei ndi viduali tems;i nf acti tdeal swi t
hanaggr egat eofobj ect s.
Individuali t
ems,whent heyar et akeni ndi vidual l
ydonotconst it
ut eanyst atistical
dataanddonotser veanypur posef oranyst at istical enqui ry.
 St atist icall awsar enotexact :Iti swel lknownt hatmat hemat icalandphy sical
sciencesar eexact .Butst atisticall awsar enotexactandst atisti
call awsar eonl y
appr oxi mat ions.St ati
st icalconcl usionsar enotuni versal lytrue.Theyar etrueonl y
onanav er age.
 St atist icst abl emaybemi sused:St ati
st icsmustbeusedonl ybyexper t
s;other wi se,
statist i
calmet hodsar et hemostdanger oust ool sont hehandsoft heinexper t.The
useofst at isticalt oolsbyt hei nexper i
encedandunt racedper sonsmi ghtl eadt o
wrongconcl usions.St at isticscanbeeasi lymi susedbyquot i
ngwr ongf igur esof
data.AsKi ngsay sapt ly‘ st atisticsar elikecl ayofwhi chonecanmakeaGodorDev i
l
asonepl eases. ’
 St atist icsi sonl y,oneoft hemet hodsofst udy ingapr oblem:St atisti
calmet hod
doesnotpr ovidecompl etesol utionoft hepr obl emsbecausepr oblemsar et obe
studiedt aki ngt hebackgr oundoft hecount riescul tur e,philosophyorr eli
gi oni nt o
consi der at ion.Thust hest atistical studyshoul dbesuppl ement edbyot herev i
dences.
1.5. Scal esofmeasur ement
Scalesofmeasur ementr ef ert oway si nwhi chv ariablesornumber sar edef i
nedand
categorized.Eachscal eofmeasur ementhascer tainpr oper tieswhi chi nt urndet ermi ne
the appr opr iateness f oruse ofcer t
ain st atisticalanal yses.The f ourscal es of
measur ementar enomi nal, or dinal ,interval, andr at io.
Nomi nalScal es:Lev elofmeasur ementwhi chcl assi fi
esdat ai nt
omut uallyexclusiv e, al
l
i
nclusivecat egor i
esi nwhi chnoor derorr anki ngcanbei mposedont hedat a.
 Noar ithmet icandr elat ional oper ationcanbeappl ied.
 Thusonl ygi vesnamesorl abel st ov ar iouscat egor ies.
Exampl es:
 Pol i
ticalpar typr ef erence( Republ ican, Democr at,orOt her ,
)
 Sex( Mal eorFemal e.)
 Mar italstatus( mar r i
ed, singl e, wi dow, div
or ce)
 Count rycode
2.Or dinalScal es:Lev elofmeasur ementwhi chcl assi fi
esdat aintocat egori
est hatcan
ber anked, howev erdiff er encesbet weent her anksdonotexi st .
 Ar ithmet icoper ati
onsar enotappl icabl ebutr elational oper ationsar eappl icabl e.
 Or deringi st hesol epr oper t yofor dinal scal e.
Exampl es:
 Let tergr ades( A, B, C, D, F).
 Rat i
ngscal es( Excel lent ,Verygood, Good, Fai r,poor ).
 Mi l
itaryst atus.
3.Interv alScal es:Lev elofmeasur ementwhi chcl assi fiesdat at hatcanber ankedand
diff
erences ar e meani ngf ul.Howev er,t her ei s no meani ngfulzer o,so r atios ar e

5
meaningl
ess.
 All arithmeti
coper at
ionsexceptdi
v i
sionareapplicabl
e.
 Rel ationaloper
ationsarealsopossible.
Examples:
 I Q, TemperatureinF0.
4.Ratio Scal es:Levelofmeasur ementwhi ch classi
fiesdatathatcan beranked,
di
ffer
encesar emeaningful,andthereisat ruezero.Tr uerat
iosexi
stbetweenthe
di
ffer
entuni t
sofmeasur e.
 Allar i
thmeticandr el
ati
onaloperat
ionsar eappli
cable.
Examples:
 Wei ght
 Hei ght
 Numberofst udents
 Age

Quest
ions

1.Whati
stheappl
i
cat
ionofst
ati
sti
csf
ory
ourf
iel
dofst
udy
?
2.Expl
aint
hedi
ff
erencesbet
weenasampl
eandapopul
ati
on.
3.Whyar
esampl
esusedi
nst
ati
sti
cs?

Descr
ipt
iveorI
nferent
ial
?
4.By2040atleast3.5bi
ll
ionpeopl
ewi
l
lrunshor
tofwat
er(
Wor
ldFut
ureSoci
ety
).
5. Ni
neoutoft
enon-
the-
jobf
atal
i
tiesar
emen(
Sour USATODAYWeekend)
ce: .
Chapt
erTwo
Met
hodsofdat
acol
lect
ionandPr
esent
ati
on
Oncei
tisdeci
dewhatt
ypeofst
udyi
stobemade,i
tbecomesnecessar
ytocol
l
ect
ed
i
nfor
mat
ionaboutt
heconcer
nedst
udy
,most
lyi
nthef
orm ofdat
a.I
nor
dert
ogener
ate
v
ali
dconcl
usi
onf
rom adat
a,i
nfor
mat
ionhast
obecol
l
ect
edi
nasy
stemat
icmanner
.
What
evert
hequal
i
tyofsampl
i
ngandanal
ysi
smet
hod,
ahaphazar
dlycol
l
ect
eddat
aset
i
slessl
i
kel
ytopr
oducev
aluabl
eandgener
ali
zabl
einf
ormat
ion.
Ty
pesofDat
a:
Ther
ear
etwot
ypes(
sour
ces)f
ort
hecol
l
ect
ionofdat
a.
(
1). Pr
imar
ydat
a
 Thepr
imar
ydat
aar
ethef
ir
sthandi
nfor
mat
ioncol
l
ect
ed,
compi
l
edandpubl
i
shedby
or
gani
zat
ionf
orsomepur
pose.Theyar
emostor
igi
naldat
ainchar
act
erandhav
e
notunder
goneanysor
tofst
ati
sti
cal
treat
ment
.

6
 Ref
ert
othoset
hatar
ecol
l
ect
edbyconduct
ingsur
veyt
omeett
hespeci
fi
cpr
obl
em
needsathand.
Exampl
e:Popul
ati
oncensusr
epor
tsar
epr
imar
ydat
abecauset
hesear
ecol
l
ect
ed,
compl
i
edandpubl
i
shedbyt
hepopul
ati
oncensusor
gani
zat
ion.
(
2). Secondar
ydat
a-
 Thesecondar
ydat
aar
ethesecondhandi
nfor
mat
ionwhi
char
eal
readycol
l
ect
edby
someone(
organi
zat
ion)f
orsomepur
poseandar
eav
ail
abl
efort
hepr
esentst
udy
.
Thesecondar
ydat
aar
enotpur
einchar
act
erandhav
eunder
gonesomet
reat
ment
atl
eastonce.
 dat
atakenf
rom al
readyav
ail
abl
epubl
i
shedorunpubl
i
shedsour
ce.
2.
1Met
hodsofcol
lect
ion
Ther
ear
ethr
eemaj
ormet
hodsofdat
acol
l
ect
ion
1.Sel
f-
admi
nist
eredquest
ionnai
re
2.Di
recti
nvest
igat
ion-
measur
ement(
obser
vat
ion)oft
hesubj
ectand
i
nter
viewi
ng(
face-
to-
face,
tel
ephone,
---)
3.Useofdocument
arysour
ce
1.Sel
f-
admi
nist
eredquest
ionnai
re
Quest
ionnai
rei
sthemai
ndat
acol
l
ect
ioni
nst
rumenti
nfor
malsampl
esur
vey
.Bef
ore
exami
ningt
hest
epsi
ndesi
gni
ngaquest
ionnai
reweneedt
orev
iewt
he

t
ypesofquest
ionsusedi
nquest
ionnai
res. Dependi
ngont
heamountoff
reedom
gi
vent
orespondenti
nof
fer
ingr
esponses,
ther
ear
etwobasi
cty
pesofquest
ionst
hat
canbeusedi
nquest
ionnai
res:open-
endedquest
ionsandcl
osedendedquest
ions.
Thet
ypeofquest
ionsf
orusewi
l
lbedet
ermi
nedbyt
hef
orm ofr
esponseswant
ed,
the
nat
ureoft
her
espondent
sandt
hei
rabi
l
ityt
oanswert
hequest
ions.
Open-
endedquest
ions:-al
l
owst
her
espondentt
oansweri
tfr
eel
yinhi
sorherown
wor
ds
Exampl
e:whatdoy
out
hinkar
ether
easonsf
orahi
ghdr
op-
outr
ateofv
il
lageheal
th
commi
tt
eemember
s?
Cl
osed–endedquest
ions:
-
Pr
edet
ermi
nedl
i
stofal
ter
nat
eresponsesi
spr
esent
edt
other
espondentf
orchecki
ng
t
heappr
opr
iat
eone(
s).I
timpl
i
est
hatt
her
espondent
’sanswer
sar
erest
ri
ctedi
nsome

7
 wayt
oal
i
mit
edr
angeofal
ter
nat
ives.
Di
recti
nvest
igat
ion
I
. Measur
ementor
/andobser
vat
ion
 dat
acanbeobt
ainedt
hroughdi
rectobser
vat
ionormeasur
ement
 pr
ovi
desaccur
atei
nfor
mat
ionbuti
tisexpensi
veandi
nconv
eni
ent
eg:
Landar
eameasur
ement
,Ani
mal
wei
ghtgai
n,Phy
sical
exami
nat
ion,di
rect
obser
vat
ionofwor
k.
I
I. I
nter
view
a)Face-
to-
Facei
nter
view
b)Tel
ephoneI
nter
view
3.Theuseofdocument
arysour
ce
 Ext
ract
ingi
nfor
mat
ionf
rom exi
sti
ngr
esour
ces.
 I
smuchl
essexpensi
vet
hananyot
hert
wosour
ces
 I
tisdi
ff
icul
ttogett
hei
nfor
mat
ionneededwhenr
ecor
dsar
ecompi
l
edi
n
unst
andar
dizedmanner
.
Exampl
e:-Hospi
tal
recor
ds,
prof
essi
onal
inst
it
utes,
Off
ici
alst
ati
sti
cs,
---
2.
2.Met
hodsofDat
aPr
esent
ati
on
Thi
stopi
cint
roducest
abul
arandgr
aphi
calmet
hodscommonl
yusedt
osummar
ize
bot
hqual
i
tat
iveandquant
it
ati
vedat
a.Tabul
arandgr
aphi
calsummar
iesofdat
acanbe
obt
ainedi
nannual
repor
ts,
newspaperar
ti
clesandr
esear
chst
udi
es.
Ev
ery
onei
sexposedt
otheset
ypesofpr
esent
ati
ons,soi
tisi
mpor
tantt
ounder
stand
howt
heyar
epr
epar
edandhowt
heywi
l
lbei
nter
pret
ed.
2.
2.2.Cl
assi
fi
cat
ionofDat
a
Thepr
ocessofar
rangi
ngdat
aint
ohomogenousgr
ouporcl
assesaccor
dingt
osome
commonchar
act
eri
sti
cspr
esenti
nthedat
aiscal
l
edcl
assi
fi
cat
ion.
ForExampl
e:Thepr
ocessofsor
ti
ngl
ett
ersi
napostof
fi
ce,
thel
ett
ersar
ecl
assi
fi
ed
accor
dingt
other
egi
onsandf
urt
herar
rangedaccor
dingt
ozones,
cit
ies,
etc.
2.
2.4Fr
equencydi
str
ibut
ion
Af
requencydi
str
ibut
ioni
stheor
gani
zat
ionofr
owdat
aint
abl
efor
m,usi
ngcl
assesand
f
requenci
es.Ther
ear
ethr
eebasi
cty
pesoff
requencydi
str
ibut
ions,
andt
her
e
ar
especi
fi
cpr
ocedur
esf
orconst
ruct
ingeacht
ype.Thet
hreet
ypesar
ecat
egor
ical
,
ungr
oupedandgr
oupedf
requencydi
str
ibut
ions.

8
Ther
easonsf
orconst
ruct
ingaf
requencydi
str
ibut
ionar
easf
oll
ows
 Toor
gani
zet
hedat
ainameani
ngf
ul,
int
ell
i
gibl
eway
.
 Toenabl
ether
eadert
odet
ermi
net
henat
ureorshapeoft
hedi
str
ibut
ion
 Tof
aci
l
itat
ecomput
ati
onal
procedur
esf
ormeasur
esofav
erageandspr
ead
 Toenabl
ether
esear
chert
odr
awchar
tsandgr
aphsf
ort
hepr
esent
ati
onofdat
a
 Toenabl
ether
eadert
omakecompar
isonsbet
weendi
ff
erentdat
aset
2.
2.4.
1.Cat
egor
icalFr
equencyDi
str
ibut
ion
Thecat
egor
icalf
requencydi
str
ibut
ioni
susedf
ordat
awhi
chcanbepl
acedi
nspeci
fi
c
cat
egor
iessuchasnomi
naloror
dinall
eveldat
a.Forexampl
e,dat
asuchasdat
asuch
aspol
i
ticalaf
fi
li
ati
on,r
eli
giousaf
fi
li
ati
on,ormaj
orf
iel
dofst
udywoul
dusecat
egor
ical
f
requencydi
str
ibut
ion.
The maj
orcomponent
s ofcat
egor
icalf
requency di
str
ibut
ion ar
e cl
ass,t
all
y and
f
requency
.
Exampl
e2.
1:Twent
y-f
ivear
myi
nduct
eeswer
egi
venabl
oodt
estt
odet
ermi
net
hei
r
bl
oodt
ype.Thedat
aseti
sgi
venasf
oll
ows:
A B B AB O
O O B AB B B
B O A O A O
O O AB AB A O
B A
Const
ructaf
requencydi
str
ibut
ionf
ort
hedat
a.
Sol
uti
on:
A B C D
Class Tall
y Frequency Percent
A //// 5 20
B ////// 7 28
O //////// 9 36
AB //// 4 16
2.
2.4.
2UngroupedFr
equencyDist
ri
bution
Whent
hedat
aar
enumer
icali
nter
est
edofcat
egor
ical
,ther
angeofdat
aissmal
land
each cl
ass i
s onl
y one uni
t,t
his di
str
ibut
ion i
s cal
l
ed an ungr
ouped f
requency
di
str
ibut
ion.
Themaj
orcomponent
soft
hist
ypeoff
requencydi
str
ibut
ionsar
ecl
ass,
tal
l
y,f
requency
and cumul
ati
vef
requency
.The st
eps ar
e al
mostsi
mil
arwi
tht
hatofcat
egor
ical
f
requencydi
str
ibut
ion.

9
Cumul
ati
vef
requenci
esar
eusedt
oshowhowmanyv
aluesar
eaccumul
atedupt
oand
i
ncl
udi
ngaspeci
fi
ccl
ass.
Exampl
e2.
2:Thef
oll
owi
ngdat
arepr
esentt
henumberofday
sofsi
ckl
eav
etakenby
eachof50wor
ker
sofacompanyov
ert
hel
ast6weeks.
2 0 0 5 8 3 4 1 0 0 7 1
7 1 5 4 0 4 0 1 8 9 7 0 1
7 2 5 5 4 3 3 0 0 2 5 1 3
0 2 4 5 0 5 7 5 1 1 0 2
A.Const
ructungroupedf
requencydi
str
ibut
ion
B.Howmanywor
ker
shadatl
east1dayofsi
ckl
eav
e?
C.Howmanywor
ker
shadbet
ween3and5day
sofsi
ckl
eav
e?
Sol
uti
on:
A.Si
ncet
hisdat
asetcont
ainsonl
yar
elat
ivel
ysmal
lnumberofdi
sti
nctordi
ff
erent
v
alues,i
tisconv
eni
entt
orepr
esenti
tinaf
requencyt
abl
ewhi
chpr
esent
seach
di
sti
nctv
alueal
ongwi
thi
tsf
requencyofoccur
rence.

B.Si
nce12oft
he50wor
ker
shadnoday
sofsi
ckl
eav
e,t
heansweri
s50-
12=38
C.Theansweri
sthesum oft
hef
requenci
esf
orv
alues3,
4and5t
hati
s4+5+8=17
2.
2.4.
3.Gr
oupedFr
equencyDi
str
ibut
ion
Someofbasi
cter
mst
hatar
emostf
requent
lyusedwhi
l
ewedealwi
thf
requency
di
str
ibut
ionar
ethef
oll
owi
ng:
 LowerCl
assLi
mit
sar
ethesmal
l
estnumbert
hatcanbel
ongt
othedi
ff
erent
cl
ass.
 UpperCl
assLi
mit
sar
ethel
argestnumbert
hatcanbel
ongt
othedi
ff
erent
cl
asses.

10
  Cl
assBoundar
iesar
ethenumberusedt
osepar
atecl
asses,
butwi
thoutt
hegaps
cr
eat
edbycl
assl
i
mit
s.
 Cl
assmi
dpoi
ntsar
ethemi
dpoi
ntsoft
hecl
asses.Eachcl
assmi
dpoi
ntcanbe
f
oundbyaddi
ngt
hel
owercl
assl
i
mitt
otheuppercl
assl
i
mitanddi
vi
dingt
he
sum by2.
 Cl
asswi
dthi
sthedi
ff
erencebet
weent
woconsecut
ivel
owercl
assl
i
mit
sort
wo
consecut
ivel
owercl
assboundar
ies.
Whent
her
angeoft
hedat
aisl
arge,t
hedat
amustbegr
oupedi
nwhi
cheachcl
asshas
mor
ethanoneuni
tinwi
dth.Whi
l
eweconst
ructt
hisf
requencydi
str
ibut
ion,wehav
eto
f
oll
owt
hef
oll
owi
ngst
eps.
1.I
dent
if
ythel
argestandl
owestv
alues
2.Fi
ndnumberofcl
assesbyusi
ngf
ormul
a;
St
ruge’
srul
e, wher
e t
henumberofcl
assi
sand i
sthe
numberofobser
vat
ions.

3.Fi
ndt
her
ange; or andFi
ndt
hecl
ass
wi
dthbydi
vi
dingt
her
angebyt
henumberofcl
asses

Not
ethat
:Roundt
heanswerupt
othenear
estwhol
enumberi
fther
eisa
r
emi
nder
.Fori
nst
ance, and
4.Sel
ectt
hest
art
ingpoi
ntast
hel
owestcl
assl
i
mit
.Thi
sisusual
l
ythel
owestscor
e
(
obser
vat
ion)
.Addt
hewi
dtht
othatscor
etogett
hel
owercl
assl
i
mitoft
henext
cl
ass.Keepaddi
ngunt
ily
ouachi
evet
henumberofdesi
redcl
ass cal
cul
atedi
n
st
ep3.
5.Fi
ndt
heuppercl
assl
i
mit
;subt
ractuni
tofmeasur
ement f
rom t
hel
owercl
ass
l
i
mitoft
hesecondcl
assi
nor
dert
ogett
heupperl
i
mitoft
hef
ir
stcl
ass.Then
addt
hewi
dtht
oeachuppercl
assl
i
mitt
ogetal
luppercl
assl
i
mit
s.
Uni
tofmeasur
ement
:Ist
henextexpect
edupcomi
ngv
alue.Fori
nst
ance,28,23,
52,
andt
hent
heuni
tofmeasur
ementi
sone.Becauset
akeonedat
um ar
bit
rar
il
y,
say23,t
hent
henextupcomi
ngv
aluewi
l
lbe24.Ther
efor
e, I
fthe

11
 dat
ais24.
12,30,21.
2thengi
vepr
ior
it
ytot
hedat
um wi
thmor
edeci
malpl
ace.
Take 24.
12 and guess t
he next possi
blev
alue.I
tis 24.
13.Ther
efor
e,
.
Not
ethat
: i
sthemaxi
mum v
alueofuni
tofmeasur
ementandi
sthev
alue
whenwedon’
thav
eacl
ueaboutt
hedat
a.
6.Fi
nd t
hecl
assboundar
ies. and

.I
n shor
t, and

Exampl
e2.
3:Consi
dert
hef
oll
owi
ngsetofdat
aandconst
ructt
hef
requencydi
str
ibut
ion.
11 29 6 33 14 21 18 17 22 38
31 22 27 19 22 23 26 39 34 27
St
eps
1.
2.

3.Sel
ectst
art
ingpoi
nt.Taket
hemi
nimum whi
chi
s6t
henaddwi
dth6oni
ttoget
t
henextcl
assLCL.
6 12 18 24 30 36
4.Uppercl
assl
i
mit
.Si
nceuni
tofmeasur
ementi
sone. So11i
sthe
UCLoft
hef
ir
stcl
ass.
1 17 23 29 35 41
1
5.Compl
etet
hef
requencydi
str
ibut
ionas
Cl
assli
mit 6-11 12-
17 18-23 24-29 30-35 36-41
Cl
assBoundaries 5.
5-11.5 11. 5-
17.5 17.
5-23.
5 23.
5-29.
5 29.
5-35.
5 35.
5-41.
5
Fr
equency 2 2 7 4 3 2
Example:f
indthecumul
ati
vefrequencyofexampl
e2.3.

12
 2.4.Rel
ati
veFr
equencyDi
str
ibut
ion
2.
5Cl
assMar
k(Mi
d-poi
nt)
Ani
mpor
tantv
ari
ati
onoft
hebasi
cfr
equencydi
str
ibut
ionusesr
elat
ivef
requenci
es,
whi
char
eeasi
l
yfoundbydi
vi
dingeachcl
assf
requencybyt
het
otalofal
lfr
equenci
es.A
r
elat
ivef
requencydi
str
ibut
ioni
ncl
udest
hesamecl
assl
i
mit
sasaf
requencydi
str
ibut
ion,
butr
elat
ivef
requenci
esar
eusedi
nst
eadofact
ualf
requenci
es.Ther
elat
ivef
requenci
es
ar
esomet
imesexpr
essedasper
cent
s.

Rel
ati
vef
requencydi
str
ibut
ionenabl
esust
ounder
standt
hedi
str
ibut
ionoft
hedat
aand
t
ocompar
edi
ff
erentset
sofdat
a.
2.
6.Gr
aphi
calandDi
agr
ammat
icPr
esent
ati
onofDat
a
Wehav
edi
scussedt
het
echni
quesofcl
assi
fi
cat
ionandt
abul
ati
ont
hathel
pusi
n
or
gani
zi
ngt
hecol
l
ect
eddat
ainameani
ngf
ulf
ashi
on.Howev
er,
thi
swayofpr
esent
ati
on
ofst
ati
sti
caldat
adoesnotal
way
spr
ovet
obei
nter
est
ingt
oal
ayman.Toomany
f
igur
esar
eof
tenconf
usi
ngandf
ail
toconv
eyt
hemassageef
fect
ivel
y.
Oneoft
hemostef
fect
iveandi
nter
est
ingal
ter
nat
ivewayi
nwhi
chast
ati
sti
caldat
amay
be pr
esent
ed i
sthr
ough di
agr
ams and gr
aphs.Ther
e ar
e sev
eralway
sin whi
ch
st
ati
sti
caldat
amaybedi
spl
ayedpi
ctor
ial
l
ysuchasdi
ff
erentt
ypesofgr
aphsand
di
agr
ams.
2.
6.1.Di
agr
ammat
icdi
spl
ayofdat
a:Pi
e-char
tandBarchar
ts
I
. Pi
echar
t
Pi
echar
tcanusedt
ocompar
ether
elat
ionbet
weent
hewhol
eandi
tscomponent
s.Pi
e
char
tisaci
rcul
ardi
agr
am andt
hear
eaoft
hesect
orofaci
rcl
eisusedi
npi
echar
t.

13
Ci
rcl
esar
edr
awnwi
thr
adi
ipr
opor
ti
onal
tot
hesquar
erootoft
hequant
it
iesbecauset
he
ar
eaofaci
rcl
eis .Toconst
ructapi
echar
t(sect
ordi
agr
am)
,wedr
awaci
rcl
ewi
th

r
adi
us(
squar
erootoft
het
otal
).Thet
otalangl
eoft
heci
rcl
eis .Theangl
esofeach

componentar
ecal
cul
atedbyt
hef
ormul
a.

These angl
es ar
e made i
nthe ci
rcl
e bymean ofa pr
otr
act
ort
o show di
ff
erent
component
s.Thear
rangementoft
hesect
orsi
susual
l
yant
i-
clockwi
se.
Exampl
e2.
4:Thef
oll
owi
ng t
abl
egi
vest
hedet
ail
sofmont
hlybudgetofa f
ami
l
y.
Repr
esentt
hesef
igur
esbyasui
tabl
edi
agr
am.

Sol
uti
on:
Thenecessar
ycomput
ati
onsar
egi
venbel
ow:

14
 I
I. BarChar
ts
Thebargr
aph(
simpl
ebarchar
t,mul
ti
plebarchar
tandst
rat
if
iedorst
ackedbarchar
t)
usesv
ert
icalorhor
izont
albar
stor
epr
esentt
hef
requenci
esofadi
str
ibut
ion.Whi
l
ewe
dr
awbarchar
t,wehav
etoconsi
dert
hef
oll
owi
ngt
wopoi
nts.Thesear
e
 Maket
hebar
sthesamewi
dth
 Maket
heuni
tsont
heaxi
sthatar
eusedf
ort
hef
requencyequal
insi
ze
A.Asi
mpl
ebarchar
tisusedt
orepr
esent
sdat
ainv
olv
ingonl
yonev
ari
abl
ecl
assi
fi
ed
onspat
ial
,quant
it
ati
veort
empor
albasi
s.I
nsi
mpl
ebarchar
t,wemakebar
sofequal
wi
dthbutv
ari
abl
elengt
h,i
.e.t
hemagni
tudeofaquant
it
yisr
epr
esent
edbyt
he
hei
ghtorl
engt
hoft
hebar
s.Fol
l
owi
ngst
epsar
eunder
takeni
ndr
awi
ngasi
mpl
ebar
di
agr
am:
 Dr
aw t
wo per
pendi
cul
arl
i
nes one hor
izont
all
y and t
he ot
herv
ert
ical
l
y atan
appr
opr
iat
epl
aceoft
hepaper
.
 Taket
hebasi
sofcl
assi
fi
cat
ional
onghor
izont
all
i
ne(
X-axi
s)andt
heobser
ved
v
ari
abl
eal
ongv
ert
ical
li
ne(
Y-axi
s)orv
icev
ersa.
 Mar
kssi
gnsofequalbr
eat
hforeachcl
assandl
eav
eequalornotl
esst
hanhal
f
br
eat
hinbet
weent
wocl
asses.
 Fi
nal
l
ymar
kst
hev
aluesoft
hegi
venv
ari
abl
etopr
epar
erequi
redbar
s.
Exampl
e2.
5:Dr
awsi
mpl
ebardi
agr
am t
orepr
esentt
hepr
ofi
tsofabankf
or5y
ear
s.

15
B.Mul
ti
plebarchar
tsar
eusedt
woormor
eset
sofi
nter
-r
elat
eddat
aar
erepr
esent
ed
(
mul
ti
plebardi
agr
am f
aci
l
iti
escompar
isonbet
weenmor
ethanonephenomenon)
.
Thet
echni
queofsi
mpl
ebarchar
tisusedt
odr
awt
hisdi
agr
am butt
hedi
ff
erencei
s
t
hatwe use di
ff
erentshades,col
ors,ordot
sto di
sti
ngui
sh bet
ween di
ff
erent
phenomena.
Exampl
e2.
6:Dr
awamul
ti
plebarchar
ttor
epr
esentt
hei
mpor
tandexpor
tofCanada
(
val
uesi
n$)f
ort
hey
ear
s1991t
o1995.

C.St
rat
if
ied(
Stackedorcomponent
)BarChar
tisusedt
orepr
esentdat
ainwhi
cht
he

16
 t
otalmagni
tudei
sdi
vi
dedi
ntodi
ff
erentorcomponent
s.I
nthi
sdi
agr
am,f
ir
stwe
makesi
mpl
ebar
sforeachcl
asst
aki
ngt
otalmagni
tudei
nthatcl
assandt
hendi
vi
de
t
hesesi
mpl
ebar
sint
opar
tsi
nther
ati
oofv
ari
ouscomponent
s.Thi
sty
peofdi
agr
am
showst
hev
ari
ati
oni
ndi
ff
erentcomponent
swi
thi
neachcl
assaswel
lasbet
ween
di
ff
erentcl
asses.Sub-
div
idedbardi
agr
am i
sal
soknownascomponentbarchar
tor
st
akedchar
t.
Exampl
e2.
7:Thet
abl
ebel
owshowst
hequant
it
yinhundr
edkgsofWheat
,Bar
leyand
Oat
spr
oducedonacer
tai
nfor
m dur
ingt
hey
ear
s1991t
o1994.Dr
aw st
rat
if
iedbar
char
t.

Sol
uti
on:Tomaket
hecomponentbarchar
t,f
ir
stofal
lwehav
etot
akey
earwi
set
otal
pr
oduct
ion.Ther
equi
reddi
agr
am i
sgi
venbel
ow:

2.
5.2.Gr
aphi
calpr
esent
ati
onofdat
a:Hi
stogr
am andFr
equencyPol
ygon
St
ati
sti
cal
graphscanbeusedt
odescr
ibet
hedat
asetort
oanal
yzei
t.Gr
aphsar
eal
so
usef
uli
nget
ti
ngt
heaudi
ence’
sat
tent
ioni
napubl
i
cat
ionoraspeaki
ngpr
esent
ati
on.
Thet
womostcommonl
yusedgr
aphsi
nresear
char
e
(
1). Hi
stogr
am
Hi
stogr
am i
saspeci
alt
ypeofbargr
aphi
nwhi
cht
hehor
izont
alscal
erepr
esent
s
cl
assesofdat
aval
uesandt
hev
ert
icalscal
erepr
esent
sfr
equenci
es.Thehei
ghtoft
he

17
 bar
scor
respondt
othef
requencyv
alues,
andt
hedr
awnadj
acentt
oeachot
her(
wit
hout
gaps)
.
Wecanconst
ructahi
stogr
am af
terwehav
efi
rstcompl
etedaf
requencydi
str
ibut
ion
t
abl
eforadat
aset
.The axi
sisr
eser
vedf
ort
hecl
assboundar
ies.

Exampl
e2.
8:Taket
hedat
ainexampl
e2.
3.

7.
0

6.
0

5.
0
Fr
equency
4.0

3.
0

2.
0

1.
0

0.
0 5.
5 11.
5 17.5 23.5 29.
5 35.
5 41.
5
Classboundari
es
Rel
ati
vef
requencyhi
stogr
am hast
hesameshapeandhor
izont
al( )scal
easa

hi
stogr
am,butt
hev
ert
ical( )scal
eismar
kedwi
thr
elat
ivef
requenci
esi
nst
eadof

act
ual
frequenci
es.
(2)
. FrequencyPoly
gon
Af
requencypol
ygonusesl
i
nesegmentconnect
edt
opoi
ntsl
ocat
eddi
rect
lyabov
e
cl
assmi
dpoi
ntv
alues.Thehei
ght
soft
hepoi
ntscor
respondt
othecl
assf
requenci
es,
andt
hel
i
nesegment
sar
eext
endedt
othel
eftandr
ightsot
hatt
hegr
aphbegi
nsand
endsont
hehor
izont
alaxi
swi
tht
hesamedi
stancet
hatt
hepr
evi
ousandnextmi
dpoi
nt
woul
dbel
ocat
ed.
Exampl
e2.
9:Taket
hedat
ainexampl
e2.
3.

18
 7.
0

6.
0

5.
0

4.
0

3.
0

2.
0
2.
5 8.
5 14.
5 20.
5 26.
5 32.
5 38.
5 44.
5
Mi
dpoi
nts

UNI
TTHREE

SUMMARI
ZATI
ONOFDATA

I
ntr
oduct
ion

Afterthedat ahav ebeencol l

ectedandpr esent ed,t henextst epi stoanal yzet hedat a
andmaket hepr operi nt
erpr
etati
on.Classif
icat i
oni snotconv eni entf orcompar isonof
onedi st
ributionwi thanother.Thef r
equencydi st
ributioni t
selfmustbesummar i
zedor
condensedi nsuchamannert hatitsessencei sexpr essedasf ewf iguresaspossi ble.
Inagi vendi st
ributi
onofdat a,ifwemov efr om t hel owestv aluet ot hehi ghest ,the
numberofi temsateachsuccessi vestagei ncreases,r eachesmaxi mum andaswe
proceededf urt
hert henumberofi temsi sfoundi nthecent eroft hedi st
ribut i
on,t he
tendencyofst ati
sti
caldat atogetconcent ratedatacer tai
nv aluei scalledcent ral
tendencyandt hemet hodsofdet er
miningtheact ualv alueatwhi cht hedatat endt obe
concentratedar ecalledmeasur esofcentraltendency .

Measur
esofCent
ral
Tendency

Therearedif
fer
enttypesofmeasur esofcent
ralt
endenci
esofwhichwewi l
ldiscussthemean,
medianandmode.Thel ogicbehindofaddressi
ngmeasur esofcentr
altendenciesisrai
sed
becauseoftwoi mport
antreasons:Togetasingl
evaluethatcanbethebestrepresent
ati
veof
theenti
redataandtofaci
li
tatecompari
sons.

Not
e:Someti
meswemayuset heword‘av
erage’t
odenot
ethet
hreemeasur
esofcent
ral
t
endenci
es(
i.
e.Mean,
Medi
an,
andmode)
.

19
Cr
it
eri
aofaGoodMeasur
eofCent
ralTendency

Thef
oll
owi
ngar
edesi
rabl
echar
act
eri
sti
csofagoodmeasur
eofcent
ral
tendency
;

 Itshoul
dbebasedoneachandev er
yobser vat
ion.
 Itshoul
dnotbeaff
ect
edbyext r
emev alues.
 Itshoul
dbedefi
nedri
gidly
 Itshoul
dbeeasi
lycal
culated
 Itshoul
dbecapabl
eofbeingf urt
heralgebrai
call
ytr
eat
ed.
 Themean
Themostcommonmeasur eofcentraltendencyist hemean.( Thet
echni
calnameisthe
ari
thmeti
cmeant odi
stinguishitf
rom othermeanst hatwewillnotcov
erhere.
)Themeanis
whatinever
ydayconver
sationiscal
ledtheaverage.I
tiscal
cul
atedbysimpl
yaddingt
heval
ues
ofallt
heval
i
dcasestogetheranddivi
dingbythenumberofvali
dcases.

Meancanhavediff
erentf
ami
l
ieswhi
chareconduci
veforapart
icul
art
ypeandnat
ureof
dat
a.Thesear
eAr i
thmeti
cmean,Geometr
icmean,andHar monicmean.Wewi l
lnot
di
scussthel
atert
womeansinthemater
ial
.

Ar
it
hmet
icmean.I
tist
hesum ofv
aluesofobser
vat
iondi
vi
dedbyt
het
otalnumberof
obser
vati
on.I
fX1,X2,X3,....XN ar
eN obser
vedv
aluesoft
otalpopul
ati
on,t
he
popul
ati
onmeani
sgivenby

Theabov
efor
mul
aisusedt
ocomput
ethear
it
hmet
icmeanwhendat
aar
enotgr
ouped.

Exampl
e:comput
ethemeanoft
he10,
20,
30,
40,
40,
and50.

When dataaregr
ouped usi
ng f
requencydi
str
ibut
ion,t
hepopul
ati
onmeancanbe
computedas:

20
 t
h
,wher
eXir
epr
esent
sthecl
assmar
koft
hei cl
assandf
iist
he

f
requencyoft
hesamecl
ass.

Exampl
e:-Fr
om t
hepr
evi
ousexampl
e,f
indt
heav
erageef
fi
ciencyr
ati
ngofwor
ker
s?

C.
I Xi f
i Xif
i

24–33 28.
5 4 114

34–43 38.
5 12 462

44–53 48.
5 15 727.
5

54–63 58.
5 13 760.
5

64–73 68.
5 7 479.
5

2543.
5

Exer
cises

1.Comput
eAr
it
hmet
icmeanf
ort
hef
oll
owi
ngdat
a.

 2,
5,8,
6,9,
13

 8,
9,8,
7,0.
5,7,
2,5,
7.8,
3

2.Fort
hef ol
l
owingshowsfrequencydist
ri
buti
onofpr i
ceofapr
oductperki
l
ogr
am
anduni
tsofthepr
oductof
feredforsal
einthousands.

Pr
ice/
kg uni
tsof
fer
ed

20–29 2

30–39 12

40–49 15

50–59 20

60–69 18

21
 70–79 10

80–89 9

90–99 4

Comput
e:

1.Ar
it
hmet
icMean

B.Medi
an

Themediandenotedby ofasetofnobservati
onsX1,
X2,
...
,Xn,ar
rangedi
nanincr
easi
ngor
decr
easi
ngor derofmagni
tudei
sthemi
ddl
evalueifni
soddortheari
thmeti
cmeanofthetwo
middl
evalueswhenniseven.

Themedianist hemiddl
eval
uewhichdi
vi
desthedat
asetfort
woequalpar
tsappr
oxi
mat
ely
.It
canbecalcul
atedfr
om bothgr
oupedandungr
oupeddat
a.

TheMedi
anf
orungr
oupeddat
a:-I
fthedat
aar
eraw,t
hef
oll
owi
ngst
epsshoul
dbef
oll
owedi
n
cal
cul
ati
ngt
hemedi
an:

 Arr
anget
hedatai
nascendi
ngordescendi
ngor
der
 Obser
vewhet
herNisoddoreven.

 I
fNi
sodd,

 I
fNi
sev
en,

WhenNisodd,t
hemedianv al
uealwaysbel
ongstothev
aluesoft
hedat
a.Buti
fNi
sev
en,t
he
medi
anv
aluemaynotbelongtotheval
uesofthedat
a.

Exerci
se:Findthemedi
anageoft
hef
oll
owi
ngdat
aof11wor
ker
swhoar
ewor
kingi
nDashen
Bank,Aradabranch.

16 19 18 20 17 19 25 23 32 27 24

TheMedi
anval
uef
orgr
oupeddat
a:-I
nthi
scase,
themedi
anv
aluei
scal
cul
atedbyi
nter
pol
ati
on
whi
chcanbegi
venas:

Wher
e;Lmed=t
hel
owercl
assboundar
yoft
hemedi
ancl
ass

22
 Fi=t
hecumul
ati
vef
requencyofar
ecl
assespr
iort
othemedi
ancl
ass

f
i=f
requencyoft
hemedi
ancl
ass

W =t
hecl
asswi
dth

Themedi
ancl
assi
sthecl
asswhi
chcont
ainst
he obser
vat
ion.

Exer
cise:-Fi
ndt
hemedi
anef
fi
ciencyr
ati
ngoft
hef
oll
owi
ngdat
a.

C.
B f
i

23.
5–33.
5 4

33.
5–43.
5 12

43.
5–53.
5 15

53.
5–63.
5 13

63.
5–73.
5 7

C.TheMode

Themodei sthemostf r
equentv alueoftheobservat
ion.Themode ofthesetofn
observati
onsX1,X2,X3.....Xn i
st hatval
uewhichoccursmostoftenorwit
ht hegreat
est
frequencyandmorethanonce.Datamaynotbehav i
ngmodalv al
ueori
tmaybehav i
ngmor e
thanonemodal val
ues.Modecanbecomput edfr
om groupedandungr
oupeddat
a.

Comput
ingmodalval
uef
rom Ungr
oupeddat
a:-I
ncal
cul
ati
ngt
hemodalv
aluef
rom ungr
ouped
dat
a,t
hef
oll
owi
ngst
epsar
eimpor
tant
;

 Ar
ranget
hedatai
nascendi
ngordescendi
ngorder
.
 Pi
cktheval
uewhi
chismostfr
equentl
yappear
ing.

Exer
cise:Thef
oll
owi
ngar
ewagedi
str
ibut
ionofwor
ker
sinanor
gani
zat
ion.Fi
ndt
hemodal
val
ue?

50 300 200 220 70 200 100 500 180 150

Fi
ndi
ngt
heModalval
uef
orgr
oupeddat
a:i
tiscasecal
cul
atedusi
ngi
nter
pol
ati
onmet
hod.The
f
ormul
ausedt
ocomput
emodei
nthi
scasei
sgi
venby

Wher
e,

23
 Thedif
fer
encebetweenf
requenci
esoft
hemodalcl
assandt
hecl
assi
mmedi
atel
y
pr
ecedi
ngthemodalcl
ass

Thedi
ff
erencebet
weenf
requenci
esoft
hemodalcl
assandt
hecl
assi
mmedi
atel
y
succeedi
ngi
t

Thel
owercl
assboundar
yoft
hemodal
class

Thecl
asswi
dth

Thecl
asswi
tht
hehi
ghestf
requencyi
sthemodel
class.

Exampl
e:-Forthedat
aofef
fi
ciencyr
ati
ngf
or51wor
ker
s,f
indt
hemodalef
fi
ciencyr
ategi
ven
bythemanager
?

C.
B f
i

23.
5–33.
5 4

33.
5–43.
5 12

43.
5–53.
5 15

53.
5–63.
5 13

63.
5–73.
5 7

Measur
esofDi
sper
sion

Inthepr evioussecti
onswehav ediscussedhowtogetav al
uewhi chisthemostrepresent
ativ
e
oft hedat aset.Butbyst udyi
ngt hemeasureofcent
raltendencyonly,wecannothav eenough
andcl eari deaaboutt hedistri
butionofthedataaroundt hecentralt
endencyunlessallthe
valuesar et hesame.Ther efore,studyi
ng t
hemeasur esofdi spersi
on can gi
veaddi t
ional
i
nf ormation thatenablest he i
nv est
igat
ortojudge the r
eli
abil
ityofa measur e ofcentral
tendencyasr epr
esentati
veofdat a.

Di
spersi
oncanbedef i
nedast
heext
entofscat
ter
ednessorcl
osenessofi
temsar
ounda
measureofcent
ral
tendency
.

Ty
pesofMeasur
esofDi
sper
sion

Therearevar
iousmeasur
esofv
ari
ati
on.Thef
oll
owi
ng ar
ecommonl
yused measur
esof
var
iat
ion.

 Ther ange
 Quar t
il
edevi
at i
on
 Meandev iat
ion
 Variance
 Standarddeviati
on

24
  Coef
fi
cientofv
ari
ati
on.
I
. TheRange

Rangeist
hesimplestmeasur
eofv ari
ati
onwhi
chcanbeeasil
yunderst
ood.I
tisthedi
ff
erence
betweent
helargestandsmallestval
uesoftheobser
vat
ion.Itcanbecalcul
atedusi
ngt he
fol
l
owingf
ormula;

Range=L–S...ungr
oupeddat
a

Range=cl
assmar
koft
hel
astcl
ass–Cl
assmar
koft
hef
ir
stcl
ass.... Gr
oupeddat
a

I
I. Quar
ti
leDev
iat
ion

I
tistheav
erageofthedi
ff
erencebet he3rd and1stquar
weent ti
les.I
tiscomput
edbyusi
ngt
he
f
oll
owingf
ormula;

When quart
il
e dev
iat
ion i
s smal
l
,itexpr
esses t
he pr
esence ofhi
gh uni
for
mit
y among
obser
vat
ionval
ues.

I
II
. MeanDev
iat
ion

Meandevi
ati
onistheav
erageoft
heabsol
utedev
iat
iont
akenf
rom cent
ralt
endency
,usual
l
y
f
rom meanormedi
an.

,Wher
eAi
sameasur
eofcent
ralt
endency
.Her
e,usual
l
ywe

ar
et al
kingaboutmeandeviati
onabovemeanwhi chi
sdenot
edby .Wecanhav
e
di
ff
erentmethodsofcomput
ingmeandevi
ati
onaboutmean.

a) Ungr
oupedDat
a

Themeandev
iat
ionaboutmeancanbecal
cul
atedf
rom ungr
oupeddat
ausi
ngt
hef
oll
owi
ng
for
mul
a.

b) Gr
oupedDat
a

t
h
,
wher
eXii
sthecl
assmar
koft
hei cl
ass

Exampl
e:Cal
cul
atet
hemeandev
iat
ionf
rom meanoft
hef
oll
owi
ngdat
a.

25
 Sal
es Noofday
s

10–20 3

20–30 6

30–40 11

40–50 3

50–60 2

Sinceitisagr
oupeddat
a,wehav
etomakecomput
ethemeanbef
orecal
cul
ati
ngt
hemean
deviat
ion.

C.
I Xi f
i

10–20 15 3 18 54

20–30 25 6 8 48

30–40 35 11 2 22

40–50 45 3 12 36

50–60 55 2 22 44

204

I
nter
pret
ati
on:
EachVal
uedi
ff
ersf
rom t
hemeanonav
erageby8.
16

I
V. Var
iance

I
tisthemostcommonlyusedmeasur
esofvari
ati
onwit
hdesirablemat
hemati
calproper
ti
es.In
i
tscalcul
ati
ont
hedev
iat
ionsar
esquar
edget
tingri
dofnegat
ivesignswi
thnoal
gebraicvi
olence.

26
Varianceistheaver
ageofthesquar
eofdev i
ati
onsf
rom mean.Thedev
iat
ionsar
ecal
culated
from themeanal l
owingthesum ofsquar
estobethemi ni
mum ascompar edt
odevi
ations
takenfrom anyot
herval
ue.

I
fX1,
X2...,
Xnar
ethesetofobser
vedv
alues,
var
iancei
scal
cul
atedas:

Sampl
evar
iance

Popul
ati
onv
ari
ance

Theseformulasar
eusedt ocomput
ethev
ari
anceofungr
oupeddat
a.Fr
om gr
oupeddat
a,
var
iancecanbecal
cul
atedas:

t
h
,
wher
eXii
sthecl
assmar
kofi cl
ass

V. St
andar
dDev
iat
ion

Standarddev
iat
ioni
sthesquar
erootofv
ari
ance.I
tameasur
eofhowmuchv
ari
abi
l
ityexi
stsi
n
thedata.

Sampl
est
andar
ddev
iat
ion

Popul
ati
onst
andar
ddev
iat
ion

Standarddeviati
onisanimportantmeasureofhomogenei
tyoruniquenessofthedat
af r
om
centralt
endency.Themorestandarddevi
ati
on,t
hegreat
erhet
erogeneit
yexist
samongv al
ues
ofitemsandv icever
sa.

Exampl
e:Fort
hedat
agi
venbel
ow,
findt
hev
ari
anceandst
andar
ddev
iat
ion.

2 8 6 7 7

I
nor
dert
ofi
ndv
ari
ance,
fir
stwehav
etof
indt
hear
it
hmet
icmean.

.Af
tert
hemean i
scomput
ed,squar
ed

dev
iat
ionsf
rom meanandi
tssum ar
ecomput
edasf
oll
ows;

2 -
4 16

8 4 16

27
 7 1 1

7 1 1

34

VI
. Coef
fi
cientofVar
iat
ion(
C.V)

Coeff
ici
entofvar
iat
ioni
stherat
ioofstandar
ddevi
ati
ontomeanexpressedinper
cent
age.I
tis
oneoft hebestmeasuresofvari
abil
i
ty.Iti
sfreefr
om uni
tofmeasur ement
.Coeff
ici
entof
var
iat
ionisusedt
ocomparethevari
abil
it
yoftwoormoreset
sofdat
a.

Lessvalueofcoeff
ici
entofv
ariat
ionmeansl essvari
abi
l
ityorgreat
erhomogenei
tyandhi
gh
coef
fi
cientofv
ari
ati
onmeanshighv ar
iabi
l
ityorl
esshomogenei
ty.

Example:Twodi str
ibut
ionsAandBhav emean80and20i nchesandstandarddevi
ati
ons10
and15i nchesr espect
ivel
y.Whichdi
str
ibut
ionhasmoreunl i
keelements?Tocompar ethe
var
iabi
l
itybetweenAandB, wehavet
ocalcul
atet
hecoef
fi
cientofvar
iat
ionforeachdi
str
ibut
ion
andcompar eit
.

A :

B :

Si
nceC.
VofBi
sgr
eat
ert
hanC.
VofA,
Bhasmor
eunl
i
keel
ementt
hanA.

3.
3Measur
esofLocat
ion

Wehav edi
scussedmanyi ssuesthatcanchar
act
eri
zet
hedi
str
ibuti
onofanydataset
.St
il
lwe
wi
llhavetwoot herchar
acteri
sti
cssuchasskewnessandkurtosi
s.Theseconcept
swil
lbe
di
scussedi
nthissect
ion.

28
3.
3.1Skewness

Skewnessi
sthemeasur
eofasy
mmet
ryordepar
tur
efr
om sy
mmet
ry.

Theshapeofanydi
str
ibut
ionmost
lyf
oll
owsoneoft
hef
oll
owi
ng;
symmet
ryorskewed.

 Sy
mmet
ryorBel
lShaped

Iti
sthetypeofdi st
ribut
ioninwhichtheobser
vat
ionsareev
enlyscatt
eredar
oundt hemean.
Thatishalfoftheobservati
onsli
ebelow meanandhalfoft
heobservati
onsli
eabov eit
.The
sum ofnegati
vedeviati
onsisequalt
othesum ofposi
ti
vedev
iat
ions.I
nthisdi
str
ibut
ion,
mean,
modeandmedi anareequal t
oeachother.

 SkewedDi
str
ibut
ions

I
tisadistri
but
ionwhi
chisnotsati
sfyi
ngthepr
oper
tiesofsymmetr
icdist
ri
buti
on.Thati
s,i
ta
di
str
ibut
ionwhichi
snotsy
mmet r
ic.Ther
earet
wotypesofskeweddi
str
ibut
ions:

a) Negat
ivel
ySkewedDi
str
ibut
ion

I
tist
hesi
tuat
ioni
nwhi
chl
argev
aluesar
emostf
requent
.Thet
ail
ist
il
tedt
othel
eft
.

29

x
I
tisadi
str
ibut
ionsucht
hatmeansi
slesst
hanmedi
anandmedi
ani
slesst
hanmode.Thati
s,
.

b) Posi
ti
veSkewedDi
str
ibut
ion

I
tisthecondi
ti
ont
hatsmal
lnumber
sar
emostf
requentv
alues.Thet
aili
sti
l
tedt
owar
dst
he
r
ight
.

I
tisadi
str
ibut
iont
hat .

 Measur
esofSkewness
x
St
ati
sti
cal
l
yskewnessi
smeasur
edusi
ngt
hef
oll
owi
ngmet
hods:

1.Pear
son’sMeasureofSkewness
2.Bowley
’sMeasureofSkewness
3.Usi
ngcentr
almoment s.

1) Pear
son’
sMeasur
eofSkewness

Pear
son’
scoef
fi
cientofskewness(
SKP)i
sgi
venbyt
hef
ormul
a:

.I
fmodedoesnotexi
st,
wecanr
epl
acet
hisf
ormul
aby
:

Thev
alueofPear
son’
scoef
fi
cientofskewnessr
angesbet
ween-
1and+1.

 I
fSKP>0,
thereisposi
ti
veSkewness
 I
fSKP<0,
thereisnegat
iveskewness
 I
fSKP=0,thereisnoskewness

Bowl
ey’
sMeasur
eofSkewness

Bowl
eyusedquar
ti
lest
omeasur
etheext
entandnat
ureofskewnessofadi
str
ibut
ion.

30
Thef
ormul
adev
elopedbyBowl
eyi
sgi
venas

 IfSB=0,thereisnoSkewness.
 IfSB>0,
thereisposi
ti
veskewness
 IfSB<0,
thereisnegat
iveskewness
2) Usi
ngcentralmoments

I
nthi
scaseskewnesscanbecal
cul
atedusi
ngt
hef
ormul
agi
venas:

,wher
e and ar
ethet
hir
dand

secondcent
ral
moment
srespect
ivel
y.Thi
sfor
mul
acanbet
ransl
atedas:

 I
fV1=0,noskewness.
 I
fV1>0,
ther
eisposit
iveskewness
 I
fV1<0,
ther
eisnegati
veskewness

Exampl e:The mean and medi

an ofwage di
str
ibut
ion offact
oryworker
s are 10 and 13
respectiv
ely
.Ift
hestandar
ddevi
ati
oni
s12,computetheKarlPear
son’
scoeff
ici
entofskewness
ofwor kerswagedi
stri
buti
on?

,
whi
chi
snegat
ivel
yskewed

Exer
cises

1) Thefoll
owingdatapresentt
hemar ksobt
ainedby60st
udent
s.Pr
esentt
hedat
ain
f
requencydi
str
ibut
ionusi
ng10classes.

31
 41 17 83 63 55 92 60 58 70 6

67 82 33 44 57 49 34 73 54 63

36 52 32 75 63 33 9 79 28 30

42 93 43 80 3 32 57 67 84 64

63 11 35 28 10 23 8 41 60 32

72 53 92 88 62 55 60 33 40 57

2) Fort
hefoll
owingshowsfrequencydi
stri
buti
onofpr
iceofapr
oductperki
l
ogr
am and
uni
tsoft
heproductof
fer
edforsaleint
housands.

Pr
ice/
kg uni
tsof
fer
ed

20–29 2

30–39 12

40–49 15

50–59 20

60–69 18

70–79 10

80–89 9

90–99 4

Comput
e

 Arit
hmet i
cmean
 Medi an
 Variance
 Standarddevi
ati
on
3) Par
ti
cul
arsr
egar
dingt
hei
ncomeoft
wov
il
lagesar
egi
venbel
ow

Vi
ll
ageA Vi
ll
ageB

Numberofempl
oyees 600 500

Av
eragei
ncome 1750 1860

S.
Dofi
ncome 100 81

I
nwhi
chvi
l
lagei
sthevar
iat
ioni
nincomegr
eat
er?

32
4) Thefol
l
owi
ngdat
arepr
esentki
l
owat
thour
sofel
ect
ri
cit
yconsumedby100per
sonsi
n
Har
ar.

Ki
lowat
t user
s

0–10 6

10–20 25

20–30 36

30–40 20

40–50 13

Cal
cul
ate

o Ari
thmeti
cmean
o St
andarddeviat
ion
o Coeff
ici
entofskewness.
5) Themeanandmedianofacert
aindi
str
ibut
ionar
egi
venby10and12Uni
tsr
espect
ivel
y.
I
ftheC,Vofthesamedist
ri
but
ioni
s40%,compute
o Thevar
iance
o Pear
son’scoef
fi
cientofskewness
o Themode

33
 UNI
TFOUR

PROBABI
LITYANDPROBABI
LITYDI
STRI
BUTI
ONS

4.
1.Pr
obabi
li
tyTheor
y

Basi
cConcept
s
Probabili
tyisameasur eofthelikel
i
hoodorchancet hatanuncertai
neventwilloccur.Iti
sa
numerical measureofthechanceofanout come’soccur rence.I
tcanassumeav aluebetween0
and1,i nclusi
ve.Apr obabi
li
tynearzer oindi
catest hattheoutcomei sveryunli
kelytooccur,
whileapr obabil
it
ynear1i ndi
catesthattheev entisalmostcer t
aintohappen.I
fwegot othe
extr
eme,apr obabil
i
tyofsomet hi
ngwi l
lalwayst ohappen.Thus,probabil
i
tiesar
enon- negat
ive
properfracti
ons.Iti
sthebasisforinf
erent
ialst
atist
ics

Exper
iment

Anexper i
mentisanywelldefi
nedsi
tuati
onorprocedur
et hatresul
tsinoneormor epossi
ble
outcomes.Orsimplyi
tcanbedef i
nedasanypr ocessthatgenerat
eswelldefi
nedoutcomes.
Forinst
ance,
tossi
ngacoin,rol
l
ingadie,
footbal
lmatch,
etccanbet akenasexperi
ments.

Out
come

Anout comeisapar
ticularresul
tofanexperi
ment
.Forexample,get
ti
ngeitherheadortai
lisa
possibl
eoutcomeoft heexper i
menttossi
ng acoi
n.Wi nni
ng,loosi
ng ort i
e/dr
aw arethe
possibl
eoutcomesoft hef ootbal
lexperi
ment
,andget t
ing1,2,3,4,5,or6ar epossibl
e
outcomesofther
oll
i
ngadi eexperi
ment.

Ev
ent
s

Anev entisaspeci f
iccoll
ect
ionofbasicout
comes, t
hati
s,asetcontai
ningoneormor
eofthe
basicoutcomesf rom thesamplespace.Anexperi
mentident
if
iesoneormor eout
comesofan
experi
ment .Forexample,int
heroll
ingadieexper
iment,t
hesimplecoll
ecti
onoft
woormoreof
thesixpossibleoutcomescanbet akenasanevent.

34
Sampl
eSpace

Asamplespacei
sacomplet
er ost
erorli
sti
ngofallpossi
bleoutcomesofanexper
iment
.The
sampl
espaceofanexper
imentisusual
l
yill
ust
rat
edei t
herbyalistorsomet
ypeofdiagr
am –
Venndi
agr
amsandtreedi
agrams.

I
ll
ust
rat
ionofanexper
iment
,out
comes,
event
s,andsampl
espace.

Tossi
ng/
Fli
ppi
ngacoi
ntwi
ce……………….Exper
iment

HeadsorTai
l
s……………………………….
.Out
comes/
element
aryev
ent
s

HH,
HT,
TH,
TT…………………………….
.4Ev
ent
s

(
HH,
HT,
TH,
TT)
……………………………Sampl
eSpace

Exer
cise

I
dent
if
ytheexper
iment
,out
comes,
event
sandsampl
espacef
ort
hef
oll
owi
ngquest
ions.

1.Si
tt
ingf
oranexam ……………………………….Exper
iment

Scor
ingA,
B,C,
D,F…………………………….
..Possi
bleout
comes

[
A,B,
C,D,
F]…………………………………….
..Sampl
eSpace

Scor
ingBandabov
e………………………………Ev
ent

Candabov
e………………………………Ev
ent

Dorbel
ow……………………………….
.Ev
ent

2.Footbal
lgame………………………………………Exper
iment

Wi
n,Loose,
Tie/
Draw………………………………Out
comes

[
W,L,
T]………………………………………………Sampl
eSpace

Notwi
nni
ng(
L,D)
,Notl
oosi
ng(
W,D)……………Ev
ent
s

Ev
ent
s

1. I
ndependentev
ent
s

Twoormor eev ent

sarei ndependentiftheoccurr
enceornonoccurrenceofoneoft heevents
doesnotaf fecttheoccurrenceornonoccur r
enceoft heot
hers.Certai
nexperiment s,suchas
rol
li
ngdice,yiel
dindependentevents;eachdieisi
ndependentoftheother.Whethera6i srol
led
onthef i
rstdiehasnoi nf
luenceonwhet hera6i srol
ledontheseconddi e.Coi
nt ossesalway s
areindependentofeachot her
.Thepossi bi
l
ityofgetti
ngaheadont hef i
rstt
ossofacoi nin
i
ndependentofget t
ingaheadont hesecondtoss.

35
Thei mpactofindependentev ent
sont hepr obabili
tyisthat,i
ftwoev entsareindependent,the
probabil
it
yofat taini
ngt hesecondev enti st hesamer egardl
essoft heout comeoft hefir
st
event.Thepr obabili
tyoft ossi
ngaheadi salways½r egar
dlessofwhatwast ossedpr evi
ously.
Thus,ifsomeonet ossesacoi nsixti
mesandget ssixheads,t heprobabil
it
yoft ossi
ngahead
ont heseventhti
mei s½, becausecoi
nt ossesar eindependent.Int
ermsofsy mbolicnotat
ion,i
f
XandYar eindependent :P(X/Y)=P(X)andP( Y/X)=P( Y),whereP(X/Y)denotest heprobabil
it
y
ofXoccur ri
nggivent hatYhasoccur r
ed, andP( Y/X)denotestheprobabil
ityofYoccur r
inggiven
thatXhasoccur red.

2.DependentEv
ent
s

Twoormor eev ent

saredependenti ftheoccurrenceornonoccurr
enceofoneoft heevents
aff
ect
st heoccurr
enceornonoccurrenceoft heother
s.Cert
ainexperi
ments,suchasrol
l
inga
di
e,yiel
dsdependentev ent
s;theoccur r
enceofoneoft hesixeventsisdependentont he
occur
renceornonoccur
renceofotherevents.

Thei mpactofdependentev ent

sont heprobabili
tyisthat
,iftwoev entsaredependent,the
probabi
li
tyofattai
ningthesecondev entisdif
ferentfr
om thatoftheoutcomeoft hefi
rstevent
.
I
nt ermsofsy mboli
cnot at
ion,i
fXandYar edependent:P(X/Y)≠P(X)andP( Y/X)≠P(Y),where
P(X/Y)denotestheprobabili
tyofXoccurringgiv
ent hatYhasoccurred,andP(Y/X)denotesthe
probabi
li
tyofYoccur r
inggiventhatXhasoccur r
ed.

3.Mut
ual
lyexcl
usi
veev
ent
s/Di
sjoi
ntEv
ent
s–opposi
teofJoi
ntev
ent
s

Twoormor eev ent

sar emut ual
lyexcl
usi
veiftheoccur renceofoneeventpr
ecl
udest he
occurr
enceoftheotherevents.Thischar
act
eri
sti
cmeanst hatmutual
l
yexcl
usi
veev
entscannot
occursimul
taneousl
yandt herefor
ecanhavenointer
sect
ion.

Inthet ossofasi nglecoin,t

heeventsofheadsandt ail
saremut uall
yexclusive.Theperson
tossingthecoinget seitheraheadorat ailbutneverboth.Thepr obabil
it
yoft womut uall
y
exclusi
veev ent
soccur r
ingatthesamet i
mei szero.Inter
msofsetnot at
ion,ifevent
sXandY
aremut ual
lyexcl
usive,P(XnY)=0, ort
hepr obabi
li
tyofXinter
sect
ingYi szero.

4.Col
lect
ivel
yexhaust
iveev
ent
s

Al i
stofcol lect
ively exhaust
ive event
s contains al
lpossible elementar
y ev
entsforan
experi
ment.Thusal lsamplespacesar ecol
lecti
velyexhaust
iveli
sts.Thesamplespaceforan
experi
mentcanbedescr i
bedasal i
stofev ent
st hataremut ual
lyexclusi
veandcoll
ecti
vel
y
exhausti
ve.

 Compl
ement
aryev
ent
s

Thecompl ementofaneventAi sdenoted .Allel

ementaryeventsofanexperi
mentnoti nA
comprisei
tscomplement.Forexample,ifi
nrol
li
ngonedie,eventAisgett
inganev ennumber,
thecomplementofAi sgett
inganoddnumber .I
feventAisgetti
nga5ont her
ollofadie,the
complementofAi sgetti
nga1,2,3,4,or6.Thecompl ementofev entAcontainswhatever
port
ionofthesamplespacethatev
entAdoesnotcontain.

Usi
ngt
hecompl
ementofanev
entcanbehel
pfulsomet
imesi
nsol
vi
ngf
orpr
obabi
l
iti
es

36
becauseoft
her
ule:
P( )=1-P(
A).

Met
hodsofassi
gni
ngpr
obabi
li
ti
es

Thet
hreegener
almet
hodsofassi
gni
ngpr
obabi
l
iti
esar
e

 Theclassi
calmethod–theequal
l
yli
kel
yappr
oach
 Therelat
ivefr
equencymet
hod
 Thesubjecti
vemethod

Cl
assi
calMet
hod

Theclassicalmet hodofassi
gningpr
obabil
i
tiesi
sbasedont heassumpt i
onthateachout
come
i
sequal l
yl i
kelytooccur.Cl
assi
calpr
obabil
it
yutil
i
zesr ul
esandl aws.I
tinvol
vesanexper
iment
andanev ent.Thedefini
ti
onassumesthatallnpossibleoutcomeshavet hesamechancefor
occur
ring.Inthismethodprobabil
i
tyval
uesareassi
gnedasf oll
ows:

Asne cannev erbegreat

ert hanN( nomor ethanNout comesinthepopulati
oncoul dpossibly
possessattri
butee)
,thehi ghestvalueofanyprobabil
it
yis1.Iftheprobabil
i
tyofanout come
occurr
ingis1, t
heeventi
scer t
aintooccur.Thesmall
estpossi
bleprobabil
i
tyiszero.Ifnoneof
theoutcomesofNpossi bili
tiespossessesthedesi
redcharact
eri
stc,e,t
i heprobabil
it
yis0/ N=
0,and t he eventi
s certain notto occur.The range ofpossibi
li
ti
es forpr obabil
i
ties i
s:

Thusthepr obabi
l
iti
esar
enonnegat
ivepr
operf
ract
ionsornonnegat
ivedeci
malv
aluesl
ess
thanorequalto1.

Rel
ati
veFr
equencyofOccur
renceMet
hod

Therelat
ivefrequencyofoccur
rencemet hodofassi
gni
ngprobabi
lit
iesi
sbasedoncumul ated
hi
stori
caldata.Withthi
smethod,theprobabi
li
tyofanev
entoccurr
ingisequalt
ot henumberof
ti
mest heeventhasoccur r
edint hepastdivi
dedbythetot
alnumberofoppor tuni
ti
esforthe
eventt
ohav eoccurred:

Pr
obabi
l
ityby numberof

r
elat
ivef
requency =t
imesanev
enthasoccur
red

ofoccur
rence t
otal
numberofoppor
tuni
ti
es

f
ort
heev
entt
ooccur

37
Relat
ivef
requencyofoccur
rencei
snotbasedonr
ulesorl
awsbutonwhathasoccur
redi
nthe
past.

Subj
ect
ivemet
hod

Thesubj ecti
vemet hodofassi gni
ngpr obabi
lit
yisbasedont hefeel
i
ngsori nsight
sofaper son
determining the probabil
it
y.Subjective probabi
li
ty comes from the person’sintuit
ion or
reasoni
ng.Al thoughnotasci enti
ficapproacht oprobabil
it
y,t
hesubjectivemet hodof tenis
basedont heaccumul at
ionofknowl edge,underst
anding,
andexperi
encest oredandpr ocessed
i
nt hehumanmi nd.Att imesi ti
smer el
yaguess.Atot herti
mes,subjecti
v eprobabil
it
ycan
potenti
allyyi
eldaccurateprobabil
i
ties.

Subject
ive pr
obabi
l
itycan be a pot
enti
all
yusef ulwayoft appi
ng a per
son’
s exper
ience,
knowledge,andi
nsi
ghtandusingthem t
oforecasttheoccur
renceofsomeev ent
.E.
g.Weat her
for
ecast

Ty
pesofpr
obabi
li
ti
es

Ther
ear
efourt
ypesofpr
obabi
l
iti
es.Thesear
e:

B)Si
mpleprobabil
it
y
C)Joi
ntprobabi
li
ty
D)Margi
nalprobabi
li
ty
E)Condi
ti
onalprobabi
li
ty

Si
mpl
ePr
obabi
li
ty

Si
mplepr
obabi
li
ti
esarerel
ati
vel
ystr
aightf
orwar
dwhi
char
eobt
ainedusi
ngt
hef
ormul
aP(
A)=
n(A)
/n–rel
ati
vefr
equencymethod.

Mar
ginalpr
obabi
li
ty

Mar ginalprobabi
lityisdenotedbyP( E),wher
eEi ssomeev ent
.Amar gi
nalprobabili
tyisusual l
y
comput edbydi v
idingsomesubt otalbythewhol e.Anexampl eofmar ginalprobabil
ityist he
probabi l
it
ythatast udentisinfectedbyHI V/
AI DS.Thi sprobabi
li
tyi
scomput edbydi vi
dingt he
numberofst udent sinfectedbyHI V/AIDSbyt het otalnumberofstudents.Thepr obabil
ityofa
personwear i
nggl assesi salsoamar gi
nalprobabi l
ity
.Thisprobabi
li
tyi
scomput edbydi v i
ding
thenumberofpeopl ewear i
nggl assesbyt het otalnumberofpeopl e.Amar ginalprobabili
tyis
foundi nt hemar ginofanyj ointpr obabil
i
tytable.Iti sthesum ofthejointprobabili
ti
esf ora
singlecategoryofoneat tr
ibuteov erallpossi
blecat egori
esofanotheratt
ribut
e.

Exampl
e:

ABCCompanymanuf act
ureswi
ndowaircondi ti
onersinbot hadeluxemodel (D)andast andard
model( S).Anaudit
orengagedinacomplianceaudi tofthef i
rmi svali
dati
ngt hesalesaccount
forthemont hApril
.Shehascoll
ect
ed200i nv oi
cesf orthemont h,someofwhi chwer esentto
whol esal
ers(W)andt heremai
nder
st oretail
ers( R).Ofthe140r etai
linv
oices,28ar eforthe
standardmodel .Onl
y24oft hewholesal
ei nvoicesar eforthest andardmodel .I
ft heaudi
tor
selectsoneinvoi
ceatrandom,f
indt
hef ol
l
owi ngpr obabil
it
ies.

a)Thei
nvoi
cesel
ect
edi
sfort
hedel
uxemodel
.

38
 b)Theinvoi
cesel
ect
edi
sfort
hestandar
dmodel.
c) Thei
nvoi
cesel
ect
edi
sawholesal
einvoi
ce.
d)Theinvoi
cesel
ect
edi
saret
aili
nvoi
ce.

Sol
uti
on

Whol
esal
e,W Ret
ail
,R Tot
al

Del
uxe,
D 36 0.
18* 112 0.
56* 148 0.
74*
*

St
andar
d,S 24 0.
12* 28 0.
14* 52 0.
26*
*

Tot
al 60 0.
30*
* 140 0.
70*
* 200

P(
D)=148/
200=0.
74 P(
W)=60/
200=0.
30

P(
S)=52/
200=0.
26 P(
R)=140/
200=0.
70

P(
D)=P(
WnD)+P(
RnD) P(
W)=P(
WnD)+P(
WnS)

=0.
18+056 =0.
18+0.
12

=0.
74 =0.
30

*Joi
ntPr
obabi
l
iti
es

*
*Mar
ginal
Probabi
l
iti
es

Uni
onpr
obabi
li
ty

Asecondt ypeofprobabili
tyist heunionoft woev ents.Unionprobabili
tyisdenot edbyP( E1 U
E2)
, whereE1andE2ar etwoev ents.P( E1UE2)i stheprobabil
it
ythatE1wi l
loccurort hatE2wi l
l
occurort hatbothE1 andE2 willoccur.Anexampl eofunionprobabi l
i
tyist heprobabili
tythata
personisinfectedbyHI V/AI
DSorCancer .Toqualif
yfortheunion,theper sonhast obei nf
ected
wit
hatl eastoneoft hediseases.Anot herexampl eist
hepr obabi
li
tyofwear i
ngey eglassesori s
asol di
er.Allpeopl
ewear i
ngey egl assesar eincl
udedintheunional ongwi thallpeoplewhoar e
sol
di er
sandal lsol
dierswhowearey eglasses.

Joi
ntpr
obabi
li
ty

Athir
dtypeofprobabil
ityistheint
ersecti
onoftwoev ent
sorjoi
ntprobabil
it
y.Ajoi
ntprobabil
it
y
shows the probabi
li
tyt hatan observati
on willpossess t
wo ( ormor e)character
isti
cs
si
multaneousl
y.Thatis,itmeasurest heprobabi
li
tyoftwoormor eeventsoccurr
ingtogether.
Thejoi
ntprobabi
li
tyofev entsE1andE2occur ri
ngisdenotedP(E1nE2).Someti
mesP( E1nE2)

39
 i
sr eadast heprobabil
it
yofE1 andE2.Toqual ifyfortheintersect
ion,botheventsmustoccur.
Jointprobabil
i
tyrangesfrom 0t o1,inclusive[0,1].Thesum ofal ljointpr
obabil
it
iesmustbe
equalto1.0.Anexampl eofj ointprobabil
ityisthepr obabil
it
yofaper sontobei nfect
edwith
HIV/AIDSandCancer .Beingi nf
ectedwi thoneoft hedi seasesi snotsuf f
ici
ent.A second
exampleofj oi
ntprobabi
li
tyistheprobabili
tythatthepersonisasol dieraswellashe/shewears
eyeglasses.

Condi
ti
onalpr
obabi
li
ty

Thefourthtypeiscondit
ionalprobabil
it
y.Conditi
onalprobabi
li
tyi
sdenot
edbyP( E1/E2)
.This
expr
essionisreadas:theprobabili
tythatE1 wil
loccurgiv
enthatE2i
sknowntohav eoccurr
ed.
Theconditi
onalprobabi
li
tyofanev entE1,giveneventE2 i
stherat
ioofthej
ointprobabi
li
tyof
twoeventstothemar gi
nalprobabil
i
tyofE2.

e)

Exampl
e:

BlueNi l
eUniversi
tyr ecentl
yconductedasur veyofundergraduat
estudentsi nordertogat her
i
nf ormati
onaboutt heusageoft helibr
ary.Thepopul at
ionforthi
sst udyi ncl
udedal l4000
under gr
aduate students enrol
l
ed i nthe universi
ty.The li
brar
y off
icer
s ar ei nt
erested in
i
ncr easi
ngusage,par ti
cular
lyamongf emales(F)andseniors(S)attheuni v
ersit
y.Oft he4000
students,800studentsar eseni
ors,1800studentsarefemalesand450oft he1800f emalesar e
seniors.

Requi
red:

1.Whatisthepr
obabil
it
ythatast
udentselectedatrandom i
saseniorgi
vent
hatt
hesel
ect
ed
st
udenti
sfemale?
2.Whatistheprobabi
l
ityt
hatastudentselectedatrandom i
sfemal
egivent
hatt
hesel
ect
ed
st
udenti
sseni
or?

Sol
uti
on:

Seni
or,
S Non-
Seni
or,
N Tot
al

Femal
e,F 450 0.
1125 1350 0.
3375 1800 0.
45

Mal
e,M 350 0.
0875 1850 0.
4625 2200 .
055

Tot
al 800 0.
20 3200 0.
80 4000

40
 1.P(
S/F)=P(
SnF)
/P(
F)=0.
1125/
.45=0.
25
2.P(
F/S)=P(
SnF)
/P(
S)=0.
1125/
.20=0.
5625

Usi
ng condi
ti
onal pr
obabi
l
ity
,joi
nt pr
obabi
l
ity of X and Y i
s cal
cul
ated as:

LawsofPr
obabi
li
ty

Addi
ti
veLaw

Thegener
allawofaddi
ti
onisusedtofi
ndt
heprobabi
l
ityoft
heuni
onoft
woev
ent
s,P(
E1UE2)
.
Thegener
alLawofAddit
ioni
spresent
edasfol
l
ows:

Speci
alRul
eofAddi
ti
on

Iftwoev entsaremut ual

l
yexclusiv
e,thepr obabil
i
tyoftheuni onoft hetwoev ent
sisthe
probabil
i
tyoft he f
ir
stev entplus t
he probabil
i
tyoft he second ev
ent.Because mut
uall
y
exclusi
veevent
sdonoti nt
ersect
,not
hinghastobesubtract
edout .Theformul
aisshownbel
ow.

Exampl
e:

1.Ahusbandandawi fe,each20y ear

sol d,ar
edebatingwhethertosetupar eti
rementpr
ogram for
t
hemselves.Benef
itsarepaidtothemanorwomanatt heageof70.I fbothhavediedbefore
r
eachi
ngage70,nobenef i
tsarepaid.Assumet hattheprobabil
i
tyt hatamanaged20l i
vesup
t
oage70i sapproximatel
y0.7.Ift
hehusbandandwi fejoi
nthepr ogram,whatistheprobabi
li
ty
t
hatei
therthemanort hewomanwi llcoll
ectbenefi
ts?Assumet hatthechancesofthemanor
womandy i
ngareindependentofeachot her.

Sol
uti
on:

LetM=manl
i
vesupt
oage70,
W =womanl
i
vesupt
oage70.

P(
M)=0.
60 P(
W)=0.
70

P(
WUM)=P(
W)+P(
M)–P(
WnM)

=0.
70+0. 60–P( WnM).Sincet hetwoev ent
sarei
ndependent
,thejointpr
obabi
li
ty
thatboththemanandt hewomanl i
vesuptoage70i sequalt
ot heproductof
theindi
vi
dualmar
ginal
pr obabil
it
ies.P(WnM)=P(M)*P( W)

=0.
60*0.
70

41
 =0.
42

=0.
70+0.
60–0.
42

=0.
88

2. Accordingtoar ecentst
udyconductedbybusinessmen,76% ofal
lsharehol
der
shavesome
collegeeducat
ion.Supposethat37%ofal
ladult
shav esomecoll
egeeducati
onandthat22%of
alladult
sareshareholder
s.Forarandoml
yselect
edadult:

a)Whati stheprobabi
li
tythatthepersondi
dnotownsharesofstock?0.
78
b)Whati st heprobabi
li
tyt hatthepersonownssharesofstockorhadsomecol l
ege
educat
ion?0.
4228
c) Whatistheprobabil
it
yt hatthepersonhasnei
thersomecoll
egeeducati
onnorowns
shar
esofst
ock?P( )=1–P( AUB)=0.
5772
d)Whatist
heprobabi
l
ityt
hattheper
sondoesnotownshar
esofst
ockorhasnocol
l
ege
educat
ion?P( )=1–P( AnB)=0.8382
e) Whatistheprobabi
li
tythatt
hepersonownsonlyshar
esofstockorhadsomecol
l
ege
educat
ionbutnotboth?P(AUB)–P(AnB)=0.4227–0.1672=0.2556

3.A 1999 surveyof20, 000 sales pr

ofessi
onals conducted byEt
hiopi
an Tel
ecommunicati
on
Corpor
ation(ETC)f oundthat15%ofal lsal
espr ofessi
onal
susehomef axmachinesand35%
usemobi l
etelephones.Supposethat1%ofal lsalesprof
essi
onal
shavebothfaxmachinesand
usemobiletelephones.

6)Whati sthepr obabi

lit
yt hatar andomlysel ectedsalesprof
essi
onalhasahomef ax
machineorusesamobi letelephone?
7)Whatist heprobabil
i
t ythatar andoml yselectedsalesprof
essi
onalnei
therhasahome
faxmachi nenorusesamobi letelephone?
8)Supposet hatnosalespr ofessionalhasbot hahomef axmachineandusesamobi l
e
tel
ephone.Whati st hepr obabili
tythatar andomlyselect
edsalesprofessi
onalhasa
homef axmachi neorusesamobi letel
ephone?

Mul
ti
pli
cat
ivel
aw

Thepr obabi
l
ityoftheintersecti
onoftwoev ent
s( E1 E2)i
scall
edthejointpr
obabi
l
ity.The
generallawofmul t
ipl
i
cationisusedtofindtheprobabil
it
yoftheint
ersect
ionoftwoeventsor
j
ointprobabi
li
ty.Thegenerallawofmult
ipli
cat
ioni
sst at
edasfoll
ows:

Speci
alRul
eofMul
ti
pli
cat
ion

I
feventsE1 andE2 ar
eindependent
,aspeci
all
aw ofmul
ti
pli
cat
ioncanbeusedt
ofi
ndt
he
i
nter
sect
ionofE1andE2.

42
 Exampl
e:

1. Testt he matr
ixforthe 200 execut
iver
esponses t
o det
ermi
ne whet
heri
ndust
ryt
ype i
s
independentofgeogr
aphi
clocat
ion.

Geogr
aphi
cLocat
ion

I
ndust
ryt
ype Nor
thEast
,D Sout
hEast
,E Mi
dWest
,F West
,G Tot
al

Fi
nance,
A 24 10 8 14 56

Manuf
act
uri
ng,
B 30 6 22 12 70

Communi
cat
ion,
C 28 18 12 16 74

Tot
al 82 34 42 42 200

Sol
uti
on:

Sel
ectoneindust
ryt
ypeandonegeogr
aphi
clocat
ion(
SayA–Fi
nanceandG–West
).DoesP
(A/
G)=P( A)?

P(
A/G)=P(
AnG)
/P(
G)=0.
07/
0.21=0.
33

P(
A)=56/
200=0.
28

Si
nceP(
A/G)≠P(
A),
indust
ryt
ypeandgeogr
aphi
clocat
ionar
enoti
ndependent
.

2.Consi
der
ingt
heabov
epr
obl
em,
ifar
espondenti
srandoml
ysel
ect
edf
rom t
hesedat
a:

a)Whati st
hepr obabi
li
tythatt
hisexecuti
veisf
rom t
hemi dwest
?0.21
b)Whati st heprobabi
li
tythatar espondenti
sfrom thecommuni cat
ioni
ndust r
yorf r
om
nort
heast?0.
64
c) Whatistheprobabil
i
tythatarespondentisf
rom thesout
heastorf r
om f
inanceindustr
y?
0.36
d)Whati st
hepr obabi
li
tythatt
hisexecuti
veisf
rom t
hesoutheastorthewest?0.38

3. Theresul
tsofasur
veyaski
ng,“
Doy
ouhav
eacal
cul
atorand/
oracomput
eri
nyour
home?
”ar
easf ol
l
ows:

43
 Cal
cul
ator

Yes No

Comput
er
Yes 46 3

No 11 15

I
sthev
ari
abl
ecal
cul
atori
ndependentoft
hev
ari
abl
ecomput
er?Whyorwhynot
?NO

Lawsofcondi
ti
onalpr
obabi
li
ty

Condi
tionalpr
obabil
i
tiesarebasedonknowl edgeofoneofthev ar
iabl
es.I
fE1 andE2 ar
etwo
event
s,thecondit
ionalprobabi
li
tyofE1 occurr
inggi
venthatE2 i
sknownorhasoccur redi
s
expr
essedas .Thef
ormul
aforf
indi
ngacondi
ti
onal
probabi
l
ityi
sgi
venbel
ow.

Speci
alLawofCondi
ti
onalPr
obabi
l
ity

I
fE1 andE2 arei
ndependentevents,thecondit
ionalprobabil
it
yandmar
ginalpr
obabi
l
ityoft
he
t
woev ent
sar eequal
.Thatis,
P( E1/
E2)=P( E1)
,andP( E2/E1)=P(E2)
.

4.
2.Pr
obabi
li
tyDi
str
ibut
ions
Basi
cconcept
s

Av ari
ableisachar acteri
sti
ct hatcanhavedifferentvaluesorout comes.Avariabl
ewhose
numericalv
aluei sdeterminedbyanout comeofar andom experi
ment ,or
,avariabl
ewhose
out
comesoccurbychancei scalledarandom v ari
able.Dependi
ngont heval
uesar andom
var
iabl
ecant ake,t
herear et wot ypesofrandom v ari
ables:Di
screterandom vari
abl
esand
Conti
nuousrandom v ar
iabl
es.

Discr
eteRandom Var i
abl
es:thesearerandom var
iabl
eswhichcanonlyassumenon-negat
ive
wholenumber ssuchas0,1,2,3………,nf orexample,thenumberofstudentsi
naclass,the
numberoft el
ephonecall
sreceivedi
nagi venhour
,thenumberofpeopleli
vingi
ncer
tainarea
andthelikecantakeonlynon-negat
ivewholenumbers.Asaresul
t,t
herearegapsorvoidsin
them al
onganinterv
al.

Conti
nuousRandom Vari
abl
es:t
hesearerandom v
ar i
abl
eswhi chcantakeanyval
ue,t
hatis,i
t
cant akeanyvalueoveranint
erval
.Thus,conti
nuousr andom vari
ableshav
eno gapsor
unassumed val
ues.Thesearerandom vari
abl
est hatcan assumean uncountablyinf
ini
te

44
 numberofvalues.Forexampl
e,thehei
ghtofani
ndi
vi
dual
,thedi
stancetr
aveledbyatr
uckdriv
er
i
nagi venhour,thetemperat
ureofaroom onagiv
enday,andthelikeproducesacont
inuous
random v
ari
able.

NB:Conti
nuousr
andom v ar
iabl
est
ypi
cal
lyr
ecordthevalueofameasur
ementsuchasti
me,
wei
ght,vol
ume,orlength.Whiledi
scr
eterandom var
iabl
ecount
sthenumberoft i
mesa
par
ti
cul
aratt
ri
but
eisobserved.

Probabi
li
tyDistr
ibuti
on:isal istingofthepossi bl
evaluesthatar andom var
iablecanassume
al
ongwi tht
heirprobabil
it
ies.Itisanyrepresentati
onoft hevaluesofar andom v ar
iabl
eandt he
associ
atedprobabil
it
ies.Dependi ngont hetypesofr andom v ar
iableswit
hwhi chwedealwi th,
we do havet wo types ofPr obabil
it
yDi str
ibuti
ons.:Discrete Probabi
li
tyDi str
ibut
ions and
Conti
nuousProbabili
tyDistri
butions.

Discr
eteProbabil
ityDistri
buti
onisanyrepr
esent
ati
onoft
hev al
uesofdiscret
erandom var
iabl
e
andt heassociat
edpr obabil
it
ies.Themostcommonlyuseddi scr
eteprobabil
i
tydist
ri
buti
ons
i
ncludetheBinomial,
hy pergeometri
candthePoi
ssondi
str
ibuti
ons.

TheBi
nomi
alDi
str
ibut
ion
Per
haps the mostwidel
y known ofalldi
scret
e probabi
li
ty di
str
ibut
ion i
sthe bi
nomi
al
di
str
ibut
ion.Thebi
nomi
aldi
str
ibut
ionhast
hef
oll
owingunderl
yi
ngassumpt i
ons:

i.Theexper i
mentinv ol
vesnidenti
cal
tri
alsorsampl i
ngisdonewi threpl
acement.
i
i.Eachtr i
alhasonlytwopossibl
emut uall
yexclusi
veoutcomes.[
Bi =Two]
i
i
i.Eachtr i
ali
sindependentoftheprevi
oust r
ial
s
i
v.Thepr obabil
i
tyofsuccess(P)andfail
ure(q=1- P)remainconstantforeachtr
ial
.
v.Inntrial
s,onlyXsuccessesarepossiblewher eXisawholenumberbet ween0andn[0≤
X≤n]
v
i.Iti
sappl i
cablei
fthesamplesi zenislessthan5%oft hepopulationsi
zeNori fsampl
es
aretakenwi t
hreplacement.

Tocomput
etheprobabi
li
tyofoccur
rencesi
nbi
nomi
aldi
str
ibut
ionwedohav
etheBi
nomi
al
Formula.I
tisst
atedasfol
l
ows:

Exampl
e:

1.I
fwet
ossacoi
nthr
eet
imes,
whati
sthepr
obabi
l
ityofget
ti
ngexact
lyt
woheads?

45
 Sol
uti
on:

I
nasingletoss,t
heprobabi
lit
yofget
ti
ngaheadorat
aili
s0.
5.I
ntossi
ngt
hecoi
nthr
eet
imes,
t
hef
ollowingarethepossi
bleout
comes.

HHH,
HHT,
HTH,
HTT,
THH,
THT,
TTH,
TTH,
TTT

Thepr
obabi
l
ityofget
ti
ngexact
lyt
woheadsi
s,t
her
efor
e,comput
edas

=(
0.5*
0.5*
0.5)+(
0.5*
0.5*
0.5)+(
0.5*
0.5*
0.5)

=0.
125*3=0.
375

Usi
ngt
heBi
nomi
alf
ormul
a

P=0.
50 q=1–0.
50=0.
50 n=3 x=2

X 1-
x
x=2)=n
P( cx*P *q

=3c2*
0.52*
0.51
=
3(
0.25*
0.5)=3(
0.125)=0.
375
Ther
ear
ethr
eeway
sofchoosi
ngexact
lyt
woheadsf
rom at
otal
oft
hreet
ri
als.

2.Aresear
cherwant
stot
estthecl
aimt
hat10%ofallpeopl
earel
eft
-handedbyrandomlysel
ect
ing
fort
ystudent
satauni
ver
sit
y.Whati
stheprobabi
li
tyofget
ti
ngsixlef
thandedstudent
samong
fort
y?

Sol
uti
on:

P=0.
10 q=1–0.
10=0.
90 n=40 x=6

P(
x=6)=0.
1068

I
f10%ofthepopulat
ioni
slef
t-handed,about10.
68%oft
het
imet
her
esear
cherwoul
dgetsi
x
whoar
elef
thandedinasampleofforty
.

3.Basedonpastdat a,appr
oximatel
y30%oft heoilwel
l
sdr i
l
ledinareashavingacer t
ainfavor
abl
e
geologi
calformationhavest r
uckoil.Acompanyhasi denti
fi
ed5l ocat
ionst hatpossessthi
s
inf
ormation.Assumi ngthatthechanceofst r
iki
ngoilonanyl ocati
onisi ndependentofany
others,
calcul
atetheprobabil
i
tythatexact
ly2ofthe5wellsst
rikeoi
l.

Sol
uti
on:

P=0.
30 q=1–0.
30=0.
70 n=5 x=2

P(
x=2)=0.
3087

46
 I
fthepr
obabi
l
ityofget
ti
ngoili
nareashavi
ngcer
tai
nfavor
abl
egeol
ogi
calf
ormat
ioni
s0.
3,31%
oft
heti
mewecanget2dr i
l
lswhichhaveoi
linasampl
eof5dril
l
s.

4.Thequal i
tycontroldepar
tmentofamanuf acturert
est
edthemostr ecentbatchof1000cat al
yti
c
converter
s pr oduced and f ound 50 of t hem defect
ive. Subsequentl
y , an employee
unwitt
ingly/
unintenti
onall
ymixedt hedefect
iveconver
ter
swi ththenon- defecti
veones.Ifa
sampleoft hreeconv ert
ersi
sr andomlyselect
edfrom t
hemi xedbatch,whati stheprobabil
it
y
thatt
heempl oyeemaygetonedef ect
ivei
tem?

Sol
uti
on:

Beforewet r
ytosol vethisproblem wehav etocheckwhetheral
ltheassumptionsofaBi nomial
dist
ri
buti
onar esatisf
iedornot .Oneoft heassumpti
onsst at
esthatthesamplesi ze,nmustbe
l
esst hanfiv
eper centoft hepopulati
onsi ze,N.i
nourcase,thesamplesizeisl essthat5%of
thepopulati
onsize[3/ 1000=0. 003<0.05]sowecanuset hebi
nomialdist
ri
butiont osol
vethi
s
specif
icexerci
se.

N=1000 p=R/
N,wher
eR-t
henumberofsuccessi
nthepopul
ati
on,
N

n=3 =50/
1000=0.
05

x=1 q=1–0.
05=0.
95

P(
x=1)=0.
1354

I
f5% oft heproductcont
ainsdefect
iveconv
erter
s,13.54% ofthetimet
hequal
i
tycont
rol
depar
tmentwouldget1def
ecti
veit
em inasampleofthr
eeconvert
ers.

5.Atownhasthreeambul ancesforemergencytransport
ati
ontoahospi
tal
.Theprobabi
l
itythat
anyoneofthesewillbeav ai
l
ableatagi v
ent i
mei s0.75.ifaper
soncal
lsforanambulance,
whati
stheprobabi
l
itythatanambulancewill
beav ail
abl
e?

Sol
uti
on:

n=3 p=0.
75 q=0.
25

Probabi
li
tyofgett
ing(atl
east
)anambul
ancei
scal
cul
atedasonemi
nust
hepr
obabi
l
ityof
gett
ingnoambulance.

P(
ambul
ance)=1–P(
0ambul
ance)

0 3
3c0*
=1–( 0.75*
0.25)
=1–0.
0156

=0.
9844

Usi
ngI
ndi
vi
dual
Binomi
alPr
obabi
li
tyTabl
e

Tabl
eshav
ebeendev
elopedt
hatgi
vet
hepr
obabi
l
ityofxsuccessesi
nnt
ri
alsf
orabi
nomi
al

47
 experi
ment.Thesetabl
esaregeneral
lyeasytouseandquickerthantheBinomialFormul
a,
especi
all
ywhenthenumberoftri
alsi
nvol
vedorsampl
esi
ze,
ni slar
ge.I
nordertousethi
stabl
e,
i
tisnecessar
ytospecif
ythev
aluesofn,pandx.

TheBinomialt
abl
esonl
yshowv aluesupto0. 5.Thus,itwouldappearthesetabl
esar
ecannot
beusedwhent hepr
obabil
i
tyofsuccessexceedsp=0. 5.However,suchtabl
escanbeusedby
not
ingthatthepr
obabi
l
ityofn-
xfailur
esi
sal sot heprobabil
i
tyofxsuccesses.Thati
s,f
indi
ng
t
heprobabi
li
tyofxsuccessesi
sequalt
ofi
ndi
ngt
hepr
obabi
l
ityofn-
xfai
l
ures.ncxandncn-
xar
ealwaysequal
.

Exampl
e:

Supposethat70% ofal
lcol
adrinker
sselectnondietcolas.If10coladr
inker
sar
erandoml
y
sel
ected,
whatist
heprobabi
l
ityt
hat4ofthem wil
lbedietcoladri
nker
s?

Sol
uti
on:

Fi
ndingtheprobabi
l
ityof4di
etcol
adr
inker
sisequi
val
entt
ofi
ndi
ngt
hepr
obabi
l
ityof6nondi
et
col
adrinker
s.

n=10 p=0.
7 q=0.
3 x=6

P(
x=6)=0.
2001

Fi
ndi
ngtheProbabi
li
ti
est
hatt
heNumberofSuccessesXLi
eInaGi
venI
nter
val(
Cumul
ati
ve
Pr
obabi
li
ti
es)

Cumul
ati
vepr
obabi
l
iti
esarethesum ofi
ndivi
dualprobabi
l
ityvalues.TheBi nomialfor
mul
a
gi
vesust
hepr
obabi
li
tyofexactl
yxsuccessesinntri
als/samplesi
zen.
tofi
ndcumulat
iveprobabi
l
iti
essuchasP(x≥3)
,P(x≤2)
,P(x›
10)orP(
X1≤X≤X2)=P(
10≤X≤20)
,we
shoul
daddtherespecti
veexact
/i
ndiv
idual
probabi
li
tyval
ues.

Exampl
e:

1.Apr ojectmanagerhasdeter
minedthatasubcont
ract
orfail
stodeli
verst
andardorder
s20%of
thet i
me.Thepr oj
ectmanagerhassi
xorder
sthathissubcontr
act
orhasagreedtodeli
ver
.What
i
st hepr obabi
l
ityt
hat

 Thesubcontr
act
orwil
ldeli
verall
oft heorders?0.2621
 Thesubcontr
act
orwil
ldeli
veratleastfouroftheor ders?0.9011
 Thesubcontr
act
orwil
ldeli
verexactlyfi
veorders?0. 3932
 Thesubcontr
act
orwil
lfail
todeli
v eratmostt wooft heorders?0.9011
 Whatdoy ouconcl
udef r
om youranswer sinpar ts( b)and(d)?Findi
ngt
hepr
obabi
li
ty
ofxsuccessesi
sequalt
ofindingthepr obabili
tyofn- xf ai
lur
es.

2.About20% ofallprofoot
ballpl
ayersareinj
uredduri
ngagi v
enseason.At eam hasf
ourst
ar
pl
ayers.Whati
st hepr
obabil
i
tythatatl
eastoneofthestarpl
ayer
sgetsi
njur
ed?

Sol
uti
on:

48
 n=4 p=0.
2 q=0.
8 x≥1

P(
x≥1)=1–P(
X≤0)=P(
x=0)

=1–0.
4096=0.
5904

3.Alawy eresti
matest hat40%ofthecasesinwhichsherepresentedt
hedef
endantwerewon.If
thelawy eri
spresentlyrepr
esent
ing10def
endantsi
ndif
ferentcases,
whati
stheprobabi
l
ityt
hat
atleast5oft hecaseswi l
lbewon?Whatareyouassuminghere?

Sol
uti
on:

Theassumptionwear etaki
ngherei
sthecasesi
nwhi
cht
hel
awy
eri
srepr
esent
ingar
e
i
ndependent
.Wi t
hthi
sassumpti
on:

n=10 p=0.
4 q=0.
6 x≥5

P(
x≥5)=P(
x=5)+P(
x=6)+P(
x=7)+P(
x=8)+P(
x=9)+P(
x=10)

=0.
2007+0.
1115+0.
0425+0.
0106+0.
0016+0.
0001

=0.
3670

Exampl
e:

1.Accor
dingtoast udyconductedappr
oxi
mat
ely55%ofallhospi
tal
sinagivent
owncontai
ned
100ormor ebeds.Aresearcherdr
awsasampleof15hospit
alsbyrandoml
ysel
ect
ingnames
fr
om adir
ector
yofhospital
s.

a)Whati sthepr
obabi
l
ityofsel
ect
ing10ormorehospit
alsthathave100ormor ebeds?
b)Whati sthepr
obabi
l
ityofsel
ect
ingl
essthanf
ivehospit
alsthathave100ormor ebeds?
c) Whatisthepr
obabi
l
ityofsel
ect
ingfr
om si
xtotenhospit
als,i
nclusi
ve,
thathave100ormor
e
beds?

2. Amanuf actur
ingcompanypr oduces10,000pl asti
cpar t
sperweek.Thi scompanysuppl i
es
plast
icpartstoanot hercompany,whi
chpackagest heplasti
cpart
saspar tofpi
cnicsets.The
secondcompanyr andomlysamples10plast
icpartssentfr
om thesuppli
er.Ift
woorl essofthe
sampl edplast
icpart
sar edef
ecti
ve,t
hesecondcompanyaccept sthelot.Whatistheprobabil
it
y
thatthelotwil
lbeaccept edi
fthepartmanufacturi
ngcompanyact ual
lyproduci
ngpar t
sis10%
defectiv
e?20%def ectiv
e?30%defecti
ve?40%def ect
ive?

Mean,
Var
ianceandSt
andar
dDev
iat
ionofaBi
nomi
alDi
str
ibut
ion

Binomialprobabi
l
itydi
str
ibut
ionisadiscr
eteprobabil
it
ydist
ri
buti
on.Andhence,t
hemet hod
usedt ocomputemeanandst andar
ddevi
ati
onforadiscret
erandom v
ari
abl
eissimi
larwi
ththe
met hodusedtocompute and f orabi
nomialdi
str
ibuti
on.

Abinomialdi
stri
buti
onhasanexpectedval
ueorlongrunaver
age,whichi
sdenotedbyµ.The
val
ueofµi sdetermi
nedby n*p.t
helongrunaver
ageorexpect
edvaluemeansthatifnit
ems
aresampledoverandoveragai
nforalongper
iodofti
meandifPistheprobabi
l
ityofget
ti
nga

49
successononet
ri
al,
theav
eragenumberofsuccesspersampl
eisexpect
edt
oben*
p.

Li
keforotherdiscret
ev ar
iabl
es,t
hevari
anceofthebi nomi
aldist
ri
but i
oniscalcul
atedasδ2 =
∑(
X-µ)2*P(
X)whi chisalsoequaltonpq.Thest
andarddev i
ati
onoft hebinomialdi
stri
but
ionis
2
al
socalcul
atedbytakingthesquar
erootofthev
ariance.δ=√∑(X-µ)* P(X)=√npq.

ThePoi
ssonDi
str
ibut
ion

ThePoi ssondistr
ibuti
oni snamedaf t
ertheFr enchMat hematici
anSimeonDeni
sPoi sson(1781
-1840),whopubl i
shedanar t
icl
ein1837di scussingthedistr
ibuti
on.ThePoi
ssondistri
buti
onis
anotherdiscretepr obabil
it
ydistri
but
ionwhi chi susedt odescribeanumberofpr ocesses,
i
ncludingthedi st
ri
but i
onoft el
ephonecall
sgoi ngthroughaswi t
chboardsyst
em,t hedemand
ofpatientsforservi
ceataheal t
hinsti
tuti
on,t hearri
valoftr
ucksandcarsatat oolbooth,and
thenumberofacci dent sataninter
secti
on.

Whil
eabi nomialr
andom var
iablecountsthenumberofsuccessesthatoccurinafi
xednumber
oft
ri
als,aPoissonrandom v
ariablecountsthenumberofrar
eev ent
s(successes)t
hatoccuri
n
aspecif
iedconti
nuousti
meintervalorspeci
fiedr
egi
on.

ThePoi
ssondi
str
ibut
ionhast
hef
oll
owi
ngchar
act
eri
sti
cs.

1.Theprobabil
i
tyofanoccurr
enceisthesamet hroughouttheti
meintervalorspaceperuni t
.
2.Thenumberofoccur r
encesinoneintervali
sindependentofthenumberofoccur rencesin
anot
herint
erval
.
3.Theprobabil
i
tyoftwoormor eoccur
rencesinasubi nt
erv
alissmallenought obeignored.
4.Itmustbepossibl
etodiv
idetheti
mei nt
ervalofint
eresti
ntomanysubi nter
val
s.
5.Theexpectednumberofoccurr
encesinani nt
ervali
sproporti
onal
tot hesizeoftheinterv
al.

Exampl
esofPoi
ssonr
andom v
ari
abl
e

 Thenumberofai
rpl anesar r
ivi
ngatanai r
por
tinanhour
.
 Thenumberofacci dent satafactoryi
naday.
 Thenumberofcar scr ossingabridgeduri
ngafiv
esecondi
nter
val
 Thenumberofmi spr i
ntsonapageofnewspr i
nt
 Thenumberofwhi tebl oodcell
sinabloodsuspensi
on.
 Thenumberoftypogr aphicalerr
orsonapage.
 Thenumberofbact eriainanounceoff lui
d.

Thef
ormul
aforPoi
ssondi
str
ibut
ion

Exampl
e:

50
1. Assumethatabankknowsf rom pastexperi
encethatbetween10and11a.m.ofeachday ,
the
meanarri
valrateis60customer sperhour .Supposethatthebankwantstodet erminethe
pr
obabi
li
tythatexactl
ytwocust
omer swillar
riv
einagivenmi nut
eti
meminuteinter
valbetween
10and11a. m.Ar r
ival
sareassumedt obeconst antoveragi v
enti
meinter
val.Calcul
atethe
pr
obabi
li
ty.

Sol
uti
on:

λ=60cust
omer
s/hr t
=1mi
nut
e x=2cust
omer
s

µ=λ*ι=60cust
omer
s/60mi
nut
es*60mi
nut
es=1

P(
x=2)= = =0.
1839

Theprobabi
l
ityofget
ti
ng2customersduri
ngthenextonemi
nut
einabankis0.
1839.Ort
her
e
i
s18.39%chancethatexact
ly2cust
omerswil
lar
ri
veinonemi
nut
eatabank.

2.Supposethatbankcustomersar
ri
verandomlyonweekdayafter
noonsatanaver
agerat
eof3.2
customersev er
yfourminut
es.Whatistheprobabi
l
ityofget
ti
ng10customer
sduri
nganeight
minuteint
erval?

λ=3.2cust
omers/
4mi nut
e t
=8minut
es x=10cust
omer
sµ=λ*ι=3.
2
cust
omers/
4minutes*8minut
es=6.
4cust
omers

P(
x=10)= = =0.
0528

Theprobabil
i
tyofgett
ing10customersdur
ingt
henextei
ghtminut
esinabankis0.0528.Or
ther
eis5.28%chancet
hatexact
ly10cust
omerswil
lar
ri
vei
neightmi
nut
esatabank.

3.I
fareal
est
ateoff
icesel
ls1.
6housesonav er
ageweekdayandsal
esofhousesonweekday
sar
e
Poi
ssondi
str
ibut
ed,whati
stheprobabi
l
ityofsel
l
ing:

 Fourhousesinaday?0.0551
 Nohousei naday?0.
2019
 Morethanfiv
ehousesinaday ?0.0060
 Tenormor ehousesi
naday ?1–1=0. 000
 Fourhousesintwodays?0.1781

4.Asecret
arytypes75wordsperminuteandaver
agessi
xerrorsperhouroftypi
ng.Assuming
err
oroccur
rencesareaPoissonpr
ocess,whati
sthepr
obabil
i
tythata225-wordlet
terwi
llbe
ty
pedwithouterr
or?0.
7408

5.Apencompanyaver
ages1.
2defecti
vepenspercar
tonpr
oduced(
200pens)
.Thenumberof
def
ect
spercar
tooni
sPoi
ssondi
str
ibuted.

51
 a)Whati stheprobabi
l
ityofselecti
ngacar toonandfindi
ngnodef ect
ivepen?0.
0312
b)Whati stheprobabi
l
ityoffi
ndingei ghtormoredefecti
vepensinacar t
oon?0.0000
c) Supposethatapur chaseroft hesepenswi llqui
tbuy ingfr
om thecompanyi facart
oon
cont
ainsmoret hant hr
eedef ecti
v es.Whatisthepr obabi
l
itythatthepurchaserwil
lqui
t
buyi
ngfrom t
hiscompany ?0.0338

6. A cer
tainmanufactur
ersell
samachi net hathasnumer ousmov i
ngpar t
s.A quali
tycontr
ol
inspectorcountst he numberofmov ing par
tst hatar e mi
sal
igned as the numberof
nonconf or
miti
esforaparti
cularmachine.Iti
sbeli
evedthatthenumberofnonconformit
iesper
machi nefol
lowsaPoissondistr
ibut
ion,withanaver
ageoft hr
eenonconfor
miti
espermachine.

a)Deter
mine the probabil
i
tythatt he qual
it
y cont
rolinspect
orfinds no mor
et han one
nonconf
ormityonapar ti
cul
armachineselect
edatrandom.0.0996
b)Whatisthepr obabi
li
tythatthr
eeormor enonconformit
iesmaybeobt ai
nedbythequalit
y
contr
oli
nspectoronthreemachines?0.9938

7.The numberofpai ntbli

sters pr
oduced byan aut
omated pai
nti
ng process atAssoci
ated
Indust
ri
esisPoissondistr
ibutedwit
har ateof0.06bl
i
st er
spersquarefeet.Theprocessis
abouttobeusedtopaintanitem t
hatmeasures9by15feet.

a)Whatisthepr
obabi
l
ityt
hatt
hef
ini
shedsur
facewi
l
lhav
enobli
steri
nit
?0. 0003
b)Whatisthepr
obabi
l
ityt
hatt
hef
ini
shedsur
facewi
l
lhav
ebet
ween5and8, inclusi
ve?0.4846
c) Whati
sthepr
obabi
l
ityt
hatt
hef
ini
shedsur
facewi
l
lhav
emorethan2bli
sters?0.9873

8.Thedefectsi
nanaut omat
edweavingprocessatSharpI
ndust
ri
esar
ePoissondist
ri
but
edata
meanr at
eof0.00025persquar
efoot.Theprocessi
stobeusedtoweav
eapi eceofmater
ial
s
thati
s5by16y ards.

a)Whati
sthepr
obabi
l
ityt
hatt
hispi
ecewi
ll
havenodef
ect
s?
b)Whati
sthepr
obabi
l
ityt
hati
twil
lhav
eonedefect
?

MeanandVar
ianceofPoi
ssonDi
str
ibut
ion

ThePoi
ssondistr
ibuti
onhasonlyoneparameter,
theexpect
edvalueλι=µ.Addi
ti
onal
l
y,fort
he
poi
sondi
stri
buti
on,theexpect
edvalueandvar
iancear
eequal.Theexpect
edval
eandvari
ancea
Poi
ssonprobabi
li
tydist
ri
buti
onareE(X)=µ=λ=δ2.

Poi
ssonAppr
oxi
mat
iont
oBi
nomi
alPr
obabi
li
tyDi
str
ibut
ion

ThePoissonprobabi
li
tydi
str
ibut
ioncanbeusedasanappr oxi
mati
ont othebinomi alpr
obabil
it
y
di
str
ibuti
onwhenP, thepr
obabil
ityofsuccessissmallandn, t
henumberoft r
ial
s/samplesize,
i
slar
ge.Simplysetµ=npanduset hePoissontabl
es.Asar ul
eofthumb, theapproxi
mat i
onwill
begoodwhenev erP≤0.
05andn≥20.Howev er
,thi
sapproxi
mationi
sr easonabl
yaccur at
eifn>20
andnp≤5.

Binomial
tabl
esar
eoftennotav
ail
abl
eforlar
geval
uesofn,sointhesecasestheapproxi
mati
on
canbeusef ul
.Soincaseswher eP≤0.05andn≥20,substit
utet hemeanoft hebinomial
dist
ri
buti
on(µ=np)i
nplaceofthemeanofthePoi
ssondist
ri
buti
on( µ=λι),
sothattheformul
a

52
 becomesP(
X)=

I
ngener
al,
thel
argerni
sandsmal
l
erpi
s,t
hebet
terwi
l
lbet
heappr
oxi
mat
ion.

Whyappr
oxi
mat
ion?

- ThePoi
ssonfor
mulaiseasi
ertousethanthebinomial f
ormula.
- I
tcanbetabul
atedmoreeff
ici
ent
lythanbinomialprobabil
it
iesbecausePoissondist
ri
but
ion
hasonl
yoneparamet
erµ(λι),
whereasbinomialdistr
ibuti
onhastwopar ametersnandp.

Exampl
e:

n=500 p=0.
02 µ=np=500*
0.02=10

n=1000 p=0.
01 µ=np=1000*
0.01=10

I
fwewanttocal
culat
eP(X)forbot
hcaseswecantabul
ateonasi
ngl
ecol
umn-Poi
sson.Hadi
t
beenbi
nomi
alf
ortheabovecasesweshoul
dhavef
ormulatedt
wocol
umns.

1.Acompanysel lsinsurancepoli
ciest
oarandom sampl
eof1000menwhoare35y ear
sofage.
Thepr obabi
li
tythata35- yearol
dmandieswi t
hinayearisappr
oxi
mat
ely0.
002.Whatist
he
probabil
it
ythattheinsurancecompanywi
llhav
etopayclai
mson2ormorepol
i
ciesnexty
ear
?

Sol
uti
on:

St
eps:
1.Makesur
eP≤0.
05andn≥20

P=0.
002 n=1000………….Bot
hrequi
rement
ssat
isf
ied

2.Cal
cul
ateµ=np=1000*
0.002=2

3.Cal
cul
ateP(
X)

P(
X≥2)=1–[
P(X=0)+P(
X=1)
]

=1-[ +[ ]

=1–(
0.1353+0.
2707)

=0.
5940

Theex
actbi
nomi
alpr
obabi
l
ityi
s0.
5942.

2. Supposet
hatthepr
obabi
li
tyofabankmakingami stakeprocessi
ngadeposi
tis0.
0003.I
f
10,
000deposi
tsar
eaudit
ed,whati
stheprobabi
l
itythatexactl
ysixmist
akesweremadein

53
pr
ocessi
ngdeposi
ts?

Sol
uti
on:

St
eps:
1.Makesur
eP≤0.
05andn≥20

P=0.
0003 n=10,
000………….Bot
hrequi
rement
ssat
isf
ied

2.Cal
cul
ateµ=np=10,
000*
0.0003=3

3.Cal
cul
ateP(
X)

P(
X=6)=

=0.
0504

Cont
inuousPr
obabi
li
tyDi
str
ibut
ion

Upt ot hispoint,wehav efocusedourat tenti

onondi scretedistr
ibuti
onsofr andom v ariables
thathav eeitheraf initenumberofpossi bleval
ue( E.g.0,1,2,3…n)oracount ablyi nfi
nite
numberofv alues( E.g.0,1,2,3…) ,andwecanal solistallofthepossiblev aluesofadi screte
random v ari
ableandi tismeani ngfult
oconsidert
hepr obabili
tythatapar t
icularindividualv alue
willbeassumed.I ncont r
ast,acont i
nuousrandom variablehasanuncount ablyinfi
ni t
enumber
ofpossi blevaluesandcanassumeanyv alueintheintervalbetweent wopoi ntsandb( a<x<b).
Asar esultt
heonl ymeani ngfulwayt ocomput eaprobabi l
it
yisthepr obabili
tythatt hev ar i
able
willfal
lwithi
naspeci fiedregion.Thatis,t
heprobabi
li
tyt hatacontinuousrandom v ariableXwi l
l
assumeanypar t
icularv al
ueiszer o.

Itisanyr epr
esentati
on ofthev al
uesofcontinuousrandom vari
ableand t
heassociated
probabi
li
ti
es.The continuous pr
obabi
l
itydi
str
ibuti
on i
ncludes t
he normaldi
str
ibut
ion and
exponenti
aldi
str
ibut
ion.

TheNor
malDi
str
ibut
ion

Thenormaldistr
ibuti
onisacont inuousdistri
buti
onthathasabellshapeandisdeterminedby
i
tsmeanandst andarddevi
ation.I toccupiesapl aceofcent r
alimpor
tanceincont i
nuous
probabi
l
itydi
str
ibuti
oninpart
icularandst at
isti
csingeneral
.Iti
sthemostimport
anttheoret
ical
di
stri
buti
onbecauseoft hef
oll
owi ngthreereasons:

1.Thenormaldistri
but
ionapproximatestheobser
v edfr
equencydi str
ibut
ionsofmanynatur
al
andphysi
calmeasur ements,
suchas, I
QS,wei
ghts,hei
ghts,sales,pr
oductli
fet
imes,
etc.
2.Thenormaldist
ribut
ioncanoftenbeusedt oest
imatebinomi alpr
obabil
iti
eswhenn(sample
si
ze)i
sgreaterthan20.
3.Thenormaldistri
but
ionisagoodappr oxi
mationofdistr
ibuti
onsofbot hsamplemeansand
samplepr
opor t
ionsoflar
gesampl es(n>30).

Char
act
eri
sti
csofnor
mal
dist
ri
but
ion

54
  Iti
sacont i
nuousdi stri
bution.
 Ithasabel lshapeandi ssy mmet ri
cal abouti t
smean.
 Iti
sasy mpt otictotheX-axi s.
 Itext
endsi nfinit
elyinei t
herdi recti
onf rom themean.
 Itisdefinedbyt wopar amet ers:µandδ.Eachcombi nationoft heset woparamet ers
specif
iesauni quenor maldi stri
bution.Thev al
ueofµi ndicateswher ethecenteroft he
belll
i
es,whi leδr epr esentshowspr eadout( orwide)thedi str
ibuti
onis.
 Itismeasur edonacont inuousscal eandt hepr obabili
tyofobt ai
ningapr eci
sev alueis
zero.
 Thetotal ar
eaundert hecur veisequal to1.0or100%; 50%oft heareaisabovethemean
and50%i sbel owthemean.
 Thepr obabi l
it
yt hatar andom v ari
ablewi llhaveav aluebet weenanyt wopoi ntsis
equalthear eaundert hecur vebet weent hosetwopoi nts.

Eachcombi nati
onofµandδspecifi
esauni quenor maldist
ributi
on.Thisbr i
ngsabouthavingan
i
nfini
tef amil
yofnor maldist
ri
butions.This problem ofdeal ing with an i
nfi
nit
ef amilyof
di
stri
buti
onscanbesol vedbyt ransfor
mi ngallnor maldistri
buti
onst ot hestandardnormal
di
stri
buti
on,whichhasameanequalt o0andast andar
ddev iati
onequalt o1.StandardNormal
Distr
ibut
ionisanormaldi
str
ibut
ioninwhi chthemeani s0andt hest andarddeviat
ionis1.Iti
s
denotedbyz.

Anynormaldistr
ibuti
oncanbeconv er
tedtot hest
andar
dnor
maldi st
ri
buti
onbystandardi
zing
eachofitsobser
v at
ionsintermsofZ-val
ues.TheZ-val
uemeasuresthedist
anceinstandard
devi
ati
onsbetweent hemeanoft henormalcurveandtheX-valueofinter
est
.Anyr andom
vari
abl
ecanbet r
ansf ormedtoast
andardrandom v
ari
abl
ebysubt
racti
ngthemeananddi vi
ding
bythestandar
ddev i
at i
on.

I
far andom v
ari
abl
eXhasmeanµandst
andar
ddev
iat
ionδ,t
hest
andar
dizedv
ari
abl
eZi
s
def
inedas:

AZ-scoreisthenumberofst andarddeviati
onsthatavalue,X,isawayf rom themean.I fthe
val
ueofXi slessthanthemean,t heZ-scoreisnegat
ive;ift heval
ueofXi sgr eatert
hant he
mean,t
heZ-scoreisposit
ive.Z-scor
ei salsoknownasz-v alue.Astandar
dizedscor ei
nwhi ch
themeaniszeroandthestandarddeviati
onis1.TheZscor eisusedtorepresentthestandard
normal
dist
ri
bution

Thepr obabi
li
tycal
cul
ati
onsinnor maldist
ri
buti
onar emadebycomput ingareasunderthe
graph.Thus,tofi
ndtheprobabi
li
tythatarandom v
ar i
ableli
eswit
hinanyspeci
fi
cint
ervalwe
mustcomput ethear
eaunderthenormalcur
veoverthatint
erval
.

Nor
malRandom Var
iabl
es

Themosti
mpor
tantt
ypeofr
andom v
ari
abl
eist
henor
malr
andom v
ari
abl
e.Thepr
obabi
l
ity

55
densi
tyf
uncti
onofanormalrandom v
ari
abl
e i sdet
erminedbyt
wopar
amet
er heexpect
s:t ed
val
ueandthestandar
ddevi
ati
onof .Wedesignat
etheseval
uesas and ,r
espect
ivel
y.

And

Thenor
malpr
obabi
l
itydensi
tyf
unct
ioni
sabel
l
-shapeddensi
tycur
vet
hati
ssy
mmet
ri
cabout
t
hev
alue ;i
tsv
ari
abi
l
ityi
smeasur
edby .Thel
arger i
s,t
hemor
evar
iabi
l
ityt
her
eisi
nthe
cur
ve.

Si
ncet
hepr
obabi
l
itydensi
tyf
unct
ionofanor
malr
andom v
ari
abl
e i
ssy
mmet
ri
cabouti
ts

expect
edv
alue ;
itf
oll
owst
hat i
sequal
l
yli
kel
ytobeonei
thersi
deof .Thati
s,

Notallbel
l-
shaped sy
mmet r
icdensi
tycur
vesar
enor
mal
.Thenor
maldensi
tycur
vesar
e
speci
fi
edbyaparti
cul
arfor
mula:

Gener
alpr
ocedur
eis

 Dr
awt
hepi
ctur
e

 Shadet
hear
eadesi
red

 Fi
ndt
hecor
rectf
igur
e

 Fol
l
owt
hedi
rect
ion

Exampl
e6:Fi
ndt
hear
eaundert
henor
mal
dist
ri
but
ioncur
vebet
ween and

Sol
uti
on:Dr
awt
hear
eaasf
oll
ows:

Si
nce t
abl
egi
vest
hear
eabet
ween0andany v
aluet
other
ightof0,
oneneedl
ookupt
he

56
val
uei
nthetabl
e.Fi
nd2.
3intheleftcol
umnand0. 04i
nthet
opr
ow.Thev
aluewher
ethe
col
umnandr
owmeetint
hetabl
eistheanswer
,0.
4904.

0.
0 0.
0 0.
0 0.
0 0.
04 …
0 1 2 3

0.
0

0.
1

0.
2

2.
2

2. 0.
490
3 4

Exampl
e:Fi
nd

A.

B.

Sol
uti
on:Dr
awt
hear
eaasf
oll
ows:

57
 0 0 1.
5

A.Dr
awt
her
equi
redar
eaasf
oll
ows:

0 0.
8

0 0.
8 0 0 0.
8

58
Exampl
e7:Fi
nd

A.

B.

Sol
uti
on:

A.Dr
awt
hegr
aphasf
oll
ows:

0 1 2

01 2 0 2 01 2

59
 B.Dr
awt
hegr
aphasf
oll
ows:

-
1.50 2.
5

-
1.5
0 2.
5 -
1.0
5 0 2.
5

Exampl
e:

1)TheGr aduateManagementAdmi ssionTest(GMAT)iswi del

yusedbygraduateschoolof
businessasanent rancerequirement .I
noneparti
cul
aryear,t
hemeanscor efort
heGMAT
was485,wi t
hast andarddev i
ati
onof105.assumi ngt hatGMAT scoresarenor mall
y
di
stributed,whatist heprobabili
tythatarandomlysel
ectedscorefr
om t
hisadminist
rat
ion
oftheGMAT:
1)Fal
lsbet ween600andt hemean, i
nclusi
ve?
1)I
sgr eatert han650?
1)I
slesst han300?
1)Fal
lsbet ween350and550, incl
usive?
1)I
slesst han700?
1)I
sexact l
y500?
1)I
f500appl icantst
akethetest,howmanywoul dyouexpecttoscor
e590orbelow?

Sol
uti
on:

St
epst
ofi
ndt
hepr
obabi
l
ityv
alueofar
andom v
ari
abl
ewhi
chl
i
esov
erani
nter
val
:

 Calcul
atetheappropri
atezv al
ues
 Fi
ndt heareas(pr
obabil
iti
es)inthet
abl
e
 I
nterprety
ourresult
s

µ=485 δ=105 485≤X≤600

a)P(
485≤X≤600)=?

60
 1.Fi
rstconv
ertXv
aluesi
ntoZ-
scor
eusi
ngt
hef
ormul
a

Z485=0

Z600= =+1.
10

P(
485≤X≤600)=P(
0≤Z≤+1.
10)

=P(
0to+1.
10)

=0.
36433

b)P(
X>650)=?

1.Fi
rstconv
ertXv
aluesi
ntoZ-
scor
eusi
ngt
hef
ormul
a

Z650= =+1.
57

P(
X>650)=P(
Z>+1.
57)

=0.
5-P(
0to+1.
57)

=0.
5-0.
44179

=0.
05281

c) P(
X<300)=?

1.Fi
rstconv
ertXv
aluesi
ntoZ-
scor
eusi
ngt
hef
ormul
a

Z300= =-
1.76

P(
X<300)=P(
Z<-
1.76)

=0.
5-P(
0to-
1.76)

=0.
5-0.
46080

=0.
03920

d)P(
350≤X≤550)=?

61
 1.Fi
rstconv
ertXv
aluesi
ntoZ-
scor
eusi
ngt
hef
ormul
a

Z350= =-
1.29

Z550= =+0.
62

P(
350≤X≤550)=P(
-1.
29≤Z≤-
1.29)

=P(
0to-
1.29)+P(
0to0.
62)

=0.
40147+0.
23237

=0.
63384

e) P(
X<700)=?

1.Fi
rstconv
ertXv
aluesi
ntoZ-
scor
eusi
ngt
hef
ormul
a

Z700= =+2.
05

P(
X>300)=P(
Z<+2.
05)

=P(
X<485)+P(
485≤X<700)

=0.
5+P(
0to+2.
05)

=0.
5+0.
47982

=0.
97982

f) P(X=500)=0.Thepr obabi
li
tyofanexact/si
ngleval
ueofacont
inuousr
andom v
ari
ableis
zero.Consequentl
y,t
heprobabi
li
tyofanintervali
sthesamewhethert
heendpoint
sar e
i
ncludedornot.
g)Tof i
ndtheexpectednumberofappl
icant
swhoscor e590orbel
ow,wefi
rstfi
ndP(X≤590)
andwemul ti
plyitbyt
henumberofappli
cants.

P(
X≤590)=?

1.Fi
rstconv
ertXv
aluesi
ntoZ-
scor
eusi
ngt
hef
ormul
a

62
 Z590= =+1.
00

P(
X≤590)=P(
Z≤+1.
00)

=P(
X<485)+P(
485≤X<590)

=0.
5+P(
0to+1.
00)

=0.
5+0.
34134

=0.
84134

I
f500appl i
cant
st akethetest
,thenumberofst
udent
sexpect
edt
oscor
e590orbel
ow i
s
500(
0.84134)=420.65or421st
udent
s.

2.Theresul
tofanexam scoref oragi v
enclassi
snormal
lydi
str
ibut
ed.Ifthemeanscoreis85
poi
ntsandt hestandar
ddevi
ationisequalt
o20point
s,f
indt
hecutoffpassi
nggradesuchthat
83.
4%oft hosetakingt
het
estwi l
lpass.

Sol
uti
on:

µ=85 pr
ob.Ofpassi
ng=83.
4%

δ=20 cut
offpoi
nt=?

Since83.
4%isgreatert
han50%,
thecut
offpoi
ntshoul
dbelesst
hant
hemean,andhencet
heZ-
valuei
snegat
ive.Andthi
scal
lsf
orthei
nver
seuseofthest
andar
dnor
malt
able.

(
Z/P=0.
334)=-
0.97

-
0.97=

-
19.
4=X-
85

X=65.
6Poi
nts–Mi
nimum poi
ntt
opasst
het
est
.

3.Dat
aaccumul at
edbytheNat i
onalCl
imati
cDataCent ershowsthattheaver
agewindspeedin
mil
esperhourf orAddisis9.7mph.Supposet hatwindspeedmeasur ementsarenormall
y
di
str
ibuted f
ora given geographi
callocat
ion.If22. 45% oft he t
ime the wi
nd speed
measurementsaremorethan11.6mph,
whatisthestandarddevi
ati
onofwindspeedi
nAddis?

Sol
uti
on:

µ=9.
7mph δ=? X>11.
6

P(
X>11.
6)=22.
45%

63
 (
Z/P=0.
2755)=+0.
76

+0.
97=

0.
97δ=1.
9

δ=2.
5

4.Thecy l
indermaki
ngmachi
nehasδ=0. 5mm andµ=25mm.wi t
hinwhati
nter
valofv
alues
cent
eredatthemeanwi
l
l,t
hedi
amet
ersof80%oft
hecy
li
nderl
i
e?

Sol
uti
on:

µ=25mm δ=0.
5mm

From t
hest
atementi
tiscl
eart
hatt
heint
erv
ali
scent
eredatt
hemean;
i.
e.,
50%oft
he80%(
40%)
l
iesbelowt
hemeanand50%l i
esabov
ethemean.

(
Z/P=0.
4)=±1.
28

X1=µ-Zδ X2=µ+Zδ

-
1.28= +1.
28=

-
0.64=X1-
25 +0.
64=X2-
25

X1=24.
36mm X2=25.
64mm

80%oft
hedi
amet
eroft
hecy
li
nderl
i
esbet
ween24.
36mm and25.
64mm.

5.Theli
vesofl
i
ghtbul
bsfol
l
owanor maldi
str
ibut
ion.I
f90%oft
hebul
bshaveli
vesexceedi
ng2000
hr
sand3% haveli
vesexceedi
ng6000hrs.Whatar et
hemeanandstandar
ddev i
ati
onofthe
l
iv
esofli
ghtbul
bs?

Sol
uti
on:

P(
X>2000)=0.
90 P(
X>6000)=0.
03

µ=? δ=?

(
Z/P=0.
4)=-1.
28 (
Z/P=0.
47)=+1.
88

-
1.28= +1.
88=

-
1.28δ=2000-µ +1.
88δ=6000-µ

64
 µ=2000+1.
28δ µ=6000-1.
88δ

Usi
ngsi
mul
taneousequat
ion,

µ=6000-1.
88δ

µ=2000+1.
28δ

3.
16δ=4000

δ=1265.
82

µ=2000+1.
28δ

=2000+1.
28(
1265.
82)

=3620.
25poi
nts

6.Onacivi
lser
viceexam,t
hegr adesarenormall
ydi
str
ibut
edwithµ=70poi
ntsandδ=10points.
Thepolicedepart
menthi r
est heappli
cantswhosegr adesar
eamongt hetop10% ofthe
popul
ati
on.Whatistheminimum graderequi
redt
obehired?

Sol
uti
on:

µ=70poi
nts δ=10poi
nts

(
Z/P=0.
4)=+1.
28

+1.
28=

+1.
28=

12.
8=X-µ

X-70=12.
8

X=82.
8–t
hemi
nimum gr
ader
equi
redt
obehi
red.

7.Abakeryshopsellsl
oavesoff reshl
ymadebr ead.Anyunsol
dl oav
esatt heendofthedayare
ei
therdi
scardedorsoldelsewhereataloss.Thedemandf orthisbreadhasfol
lowedanormal
di
stri
but
ionwithµ=35l oav esandδ=8loaves.Howmanyl oavesshouldthebakerymakeeach
daysothattheycanmeett hedemand90%oft hetime?

Sol
uti
on:

µ=70l
oav
es δ=8l
oav
es

65
(
Z/P=0.
4)=+1.
28

+1.
28=

+1.
28=

10.
24=X-35

X=45.24≈46-bystocki
ng46l
oav
esofbr
eadseachday
,thebaker
ywi
l
lmeett
hedemandf
or
t
hispr
oduct90%oft
het i
me.

Ar
eabet
ween0andz

Z 0.
00 0.
01 0.
02 0.
03 0.
04 0.
05 0.
06 0.
07 0.
08 0.
09

0.
0 0.
00000.
00400.
00800.
01200.
01600.
01990.
02390.
02790.
03190.
0359

0.
1 0.
03980.
04380.
04780.
05170.
05570.
05960.
06360.
06750.
07140.
0753

0.
2 0.
07930.
08320.
08710.
09100.
09480.
09870.
10260.
10640.
11030.
1141

0.
3 0.
11790.
12170.
12550.
12930.
13310.
13680.
14060.
14430.
14800.
1517

0.
4 0.
15540.
15910.
16280.
16640.
17000.
17360.
17720.
18080.
18440.
1879

0.
5 0.
19150.
19500.
19850.
20190.
20540.
20880.
21230.
21570.
21900.
2224

0.
6 0.
22570.
22910.
23240.
23570.
23890.
24220.
24540.
24860.
25170.
2549

66
0.
7 0.
25800.
26110.
26420.
26730.
27040.
27340.
27640.
27940.
28230.
2852

0.
8 0.
28810.
29100.
29390.
29670.
29950.
30230.
30510.
30780.
31060.
3133

0.
9 0.
31590.
31860.
32120.
32380.
32640.
32890.
33150.
33400.
33650.
3389

1.
0 0.
34130.
34380.
34610.
34850.
35080.
35310.
35540.
35770.
35990.
3621

1.
1 0.
36430.
36650.
36860.
37080.
37290.
37490.
37700.
37900.
38100.
3830

1.
2 0.
38490.
38690.
38880.
39070.
39250.
39440.
39620.
39800.
39970.
4015

1.
3 0.
40320.
40490.
40660.
40820.
40990.
41150.
41310.
41470.
41620.
4177

1.
4 0.
41920.
42070.
42220.
42360.
42510.
42650.
42790.
42920.
43060.
4319

1.
5 0.
43320.
43450.
43570.
43700.
43820.
43940.
44060.
44180.
44290.
4441

1.
6 0.
44520.
44630.
44740.
44840.
44950.
45050.
45150.
45250.
45350.
4545

1.
7 0.
45540.
45640.
45730.
45820.
45910.
45990.
46080.
46160.
46250.
4633

1.
8 0.
46410.
46490.
46560.
46640.
46710.
46780.
46860.
46930.
46990.
4706

1.
9 0.
47130.
47190.
47260.
47320.
47380.
47440.
47500.
47560.
47610.
4767

2.
0 0.
47720.
47780.
47830.
47880.
47930.
47980.
48030.
48080.
48120.
4817

2.
1 0.
48210.
48260.
48300.
48340.
48380.
48420.
48460.
48500.
48540.
4857

67
2.
2 0.
48610.
48640.
48680.
48710.
48750.
48780.
48810.
48840.
48870.
4890

2.
3 0.
48930.
48960.
48980.
49010.
49040.
49060.
49090.
49110.
49130.
4916

2.
4 0.
49180.
49200.
49220.
49250.
49270.
49290.
49310.
49320.
49340.
4936

2.
5 0.
49380.
49400.
49410.
49430.
49450.
49460.
49480.
49490.
49510.
4952

2.
6 0.
49530.
49550.
49560.
49570.
49590.
49600.
49610.
49620.
49630.
4964

2.
7 0.
49650.
49660.
49670.
49680.
49690.
49700.
49710.
49720.
49730.
4974

2.
8 0.
49740.
49750.
49760.
49770.
49770.
49780.
49790.
49790.
49800.
4981

2.
9 0.
49810.
49820.
49820.
49830.
49840.
49840.
49850.
49850.
49860.
4986

3.
0 0.
49870.
49870.
49870.
49880.
49880.
49890.
49890.
49890.
49900.
4990

68