1 1 Testing of Hypotheses
LEARNING GOALS
‘Air complaing he apes you will beable
+ Undesand the need frtesngalypohess + Do several ypes of small sample ces:
+ Undestand the bse concepts and apps, Fees. Chiagur, goodness ffi
cation of teing of hypothe + Do tes ofhomogency and proportions.
+ Do see pes of normale for ge
simple
EEED Introduction
Tr ean of ian we hve cn dh whet pepulson
pornos fic patameers canbe ese eer wid gle eal ae
sine unknown
‘ox two el values within which she parsmerr of res fle wih pcicd pbb Geral
‘timo rm be noted tha thet eximates arin fc, sed on singe simpe sp)
fae ns Now
Tet ws ammmesstuton in which we hee ths raph on tnd whee Popa
on isnt Kn a suspeced abe fom» popeation with parnmeer Inthe
ery whee the spl den iin at, Fm the population with parce @, we eb
‘ismean af the population, then the ample sic mas
tha be the mens ofthe sample, if isthe population propor then the ttc mst al
‘he proportion of the sumple and 0 on) and pt
Adan nthe populon with pr
"mpl, then we conclude the the sample
‘cherie we cold ho hs eos ers seme ee popelcon
“The proces of assuming a owe simple and vexing for
lw at toting of pes Hewere she sample col ei om pop
m tail st to know wheter che
che et eure in aor a the
come fom the population who parameters 8,
nme Bop no, he decsion oul nro sme so: of eter. This is msily dew the fc
{hac rom he original populacon fom whete the spe would hve come from one can da a
‘many samples athe mpl ati, being varble by ial bined fom exch sample could
‘rl indies conclrions. Homeve, ih experiments may pce oR intel or dhe
probably of comming ack an eroe This the lvl a wich the sp
iia ifr or do nor dif rom he aesmed poplon parm
etic ny cher
Scanned with OKEN Scanner 292 ITY AND STATISTICS FOR ENGINEERS
This pene lee is cw atthe le fein and ies to tha ofthe confidence
cocci in which wil (I~ 0 gives the probity har the population parameters within
tosample sie vals one sat andthe ethers igh The lene pretty (a2 +
(13) = @demonsraes the lee which the experienc cannot be conde, Her, whe the
‘ea coreponing te (= called asthe ape rn dhe ea comesponng to scl
25 the reo pon fe to Figure 111 gen for a tir dsebaton)
icy ppl
11.1.1. Relation Between Confidence Interval and Testing of Hypothesis
szsumed aan eximate of 8 lls within the confidence limit sy and Gy than we conclude
‘hat ievould hae come from he populcon whom parameters 8 Others ifthe sample aie
tefl ouside the confidence iit ten we conclude da the eample no fom the populaion
rho parame Therefore we have
iG, <0<0,)=
=a, rb
d)=$
Then P(0,<0<0,)=1=a is termed asthe acceptance gin and
P04,)- 84 Sa0
is emda the refeion region (fet Figure 1.1
11.12 Level of Significance and p-Value
Lew comers sation where we woul kee kom whether x sample of dh metres
‘feces of fah acl bas come from sh population whan verge wh between
14 inches and 20 inches. Tha if me cpneider Othe average wth ofthe fh poplin,
then we havel4 $6520. Here, in Gc, 6) =14 ape Gy =20- If ie she mean caine’ on
the sample wth cbacration, then if <4 8>20 we conclude that ths sample dae nat
felon to the fish population whose average width i inthe range 14 inches and 20 inches,
Scanned with OKEN Scanner APTER H-TESTING OF HYPOTHES! 293
Figure 11.2 Acceptance and ejection
Or otherwise, if 145852
we concie that cis sample hat come from the fish population
ne average wih isin the range Lt inches ad 20 inches. By this, we mean thatthe ifference
Beoween Band 8 ie not significant. IF the undying population ie normaly diibated with
rein and waiance 0 that sf ls 8 ia otal density, then thi station can be depicted
ple mean obtained fiom sample of em, then we ave
band VO
‘Now scorn the cena oem we ave
1
Zz
and Zsa wandard nema vaviate with mean O and standard deviation 1, Now cont
Pussdcana1—a
Then
pf idan < d-a
From Figure 1.2, the acepance region it
Pudsds20)=
(ee a Of Pig: $ZS44,,)=1-0
© scanned with OKEN Scanner wen EE aod 8
and can be obtained for dierent poied owl. For
iy if @> 20, shen the comesponding vale of Zcan be elated
nd thi ills inthe rejection repo i the sigh tl of che dinsibucon. Heres, we have
ome it maybe woe that when me compute ee of et sei we ihr so
both This implies 9 MZ ) and
lon asthe pobublry value or apy pale. dented by which give the scl ee
re tha is the sumed es ict Figure 11.3), However i, 2
hen sceeogy we il ave Ser
Figure 13 Aan ees
Scanned with OKEN Scanner Testing of Hypothesis
tecween he sample sate
drlaive to the standard err of @ is reasonably small then
ould Have come fom the population whose patamete i 8
lcs he peeing resting of hype deals wih making
mc By 8 By of the poplaon Fyre were «sample hs cme
then using he ample informacion in the Toe ofsample tiie 6, sail en te peso
foe cal
Hs9=8, vers Hs 06
Type Lor) = PUR |, ive) = PR 4) = sy)
Table 11-1 Decisions taken by the experimenter and the outcomes
Deca ba
(Gide apt Gude wot oe
edity cops omg
© scanned with OKEN Scanner Type leon) =~ PAcept Hy,
PiReet Hi ife)=1~
-b
is ely noe gy Thi ero leads 0 Typel
is ely uly hen Type en Bases A the same ines he
fly wen hein eally lly thi gves he peer ~
dunt resin in Type
ingin Type I eon J. Hwee fom te tomer ple of webb x pod nie i
11.2.1 Test Statistics
isfoms normal population with met neo" 18 isthe ple
ample of acm hen unde the umpson, te, 8= hor = He we bane £18)
sn V(Q)=V(@)=@ ln Since =F may be le han o equal to reser than the tuned
=a _le-1
Heal
then we say hat © ie aay fom with profiel Typ ero probably 2 Theron
EtSample sis
ror of sample ac
© scanned with OKEN Scanner WZen)oNZ>
gare 114 Sh ate shoms PIZ < ty
i rejected, hen thr ae ens wher one interested eo know wheter 88, oly or > ol.
thera vale such tha PILZ <=) =
and then were Hi 0= 0 if, <~ey. whre
nd
hiveawe MZ 8,
6-0 6-8,
and inthincane MZ > 2,)= pc eer a Figue 118)
Scanned with OKEN Scanner Figure 1.6 Shed se shone PLZ > +24) igh),
11.22 Steps for Hypothesis Testing (Normal Test)
The filling ae tee for hypothe ting
Step 1: Prope the ml hypothe Hy sy HO 8, who isthe unknown prasad i
Sie re hypothesis H, aflame:
(0) 11:00, for swe ll of significance rwith efi pobaii a2 ane
(©) 14:06, fr lef ex ofipiticance wth eal roabily is
(0 8204 te bev of
Sep 3 Asu 05 55) and cr=0.01 (1%) and aban he
© scanned with OKEN Scanner Sample seine (Sample sane) 0-0 _ 4-8
Sambrdenorof amples" oMdn oF
and compute Z value wih wion of 8 @and of vale
Figure 11.7 Clasifcin of bpote
© scanned with OKEN Scanner Large Sample Tests
11.8.1 Large Sample Tests for Single Mean (Variance ie Known)
eX hea rnd
ra age sample yp) af nem (n> 30) which pected co ave dee en she
population with mean J i ay) and known standard deviation. The wal pothesis ane
eae
res Hep ty (loreworai es)
Hawa nh, vere Howey, (orkid
Hemi, vere Hsu> th, lor igh)
pa b-O 5-H 2
Reject under igh et i >
otherwise accept
11.42 Large Sample Tests for Diference of Two Means (Variances are Known)
tmcan nf ed snd devniom ty Lc ta eoppe that wee acter ge sac
14) of sem devon fora population whose mean i and andar devon i 0
fae! sume ha thee ewe populations se ndependent ad ther standard deviations
an, ves Heth th ai
4, vers Heh, (rll)
Hea = Hy versus Ha> hy lor sghesil es)
© scanned with OKEN Scanner