Chap 009
Chap 009
ANALYSIS OF VARIANCE
9-1. H0: H1: X X X X X X X X X X X X X X X X X X X X
CHAPTER 9
9-2.
A$%VA ass #ptions: nor#al pop lations &ith eq al 'ariance. (ndependent rando# sa#plin) fro# the r pop lations. *eries of paired t-test are dependent on each other. +here is no control o'er the pro"a"ility of a +ype ( error for the ,oint series of tests. r = 5 n1 = n2 = . . . = n5 = 21 n =105 df-s of F are 4 and 100. Co#p ted F . 3./. +he p-'al e is close to 0.01. 0e,ect H0. +here is e'idence that not all 1 plants ha'e eq al a'era)e o tp t. F Distribution
1!T"i#$ F! Criti%"# 10% 2.0019 5% 2.4626 1% 3.5127 0.50% 3.9634
9-3.
9-4.
9-1.
r . 4 n1 . 12 n2 . 32 n3 . 43 n4 . 43 Co#p ted F . 12.13. 0e,ect H0. +he a'era)e price per lot is not eq al at all 4 cities. 4eel 'ery stron)ly a"o t re,ectin) the n ll hypothesis as the critical point of F 53613/7 for . .01 is appro8i#ately 3.2. F Distribution
1!T"i#$ F! Criti%"# 10% 2.1152 5% 2.6559 1% 3.8948 0.50% 4.4264
9-/.
%ri)inally6 treat#ents referred to the different types of a)ric lt ral e8peri#ents "ein) perfor#ed on a crop! today it is sed interchan)ea"ly to refer to the different pop lations in the st dy.9rrors are the differences "et&een the data points and their sa#ple #eans. :eca se the s # of all the de'iations fro# a #ean is eq al to 0.
9.3.
9-1
9-2. 9-9.
= +otal de'iation . xij ; = x . 5 x i ; x 7 < x ij x i . treat#ent de'iation < error de'iation. +he s # of sq ares principle says that the s # of the sq ared total de'iations of all the data points is eq al to the s # of the sq ared treat#ent de'iations pl s the s # of all sq ared error de'iations in the data.
9-10.
An error is any de'iation fro# a sa#ple #en that is not e8plained "y differences a#on) pop lations. An error #ay "e d e to a host of factors not st died in the e8peri#ent. :oth =*+0 and =*9 are sample statistics )i'en to nat ral 'ariation a"o t their o&n #eans. 5(f x > 0 &e cannot i##ediately re,ect H0 in a sin)le-sa#ple case either.7 +he #ain principle of A$%VA is that if the r pop lation #eans are not all eq al then it is li?ely that the 'ariation of the data points a"o t their sa#ple #eans &ill "e s#all co#pared to the 'ariation of the sa#ple #eans a"o t the )rand #ean. @istances a#on) pop lations #eans #anifest the#sel'es in treat#ent de'iations that are lar)e relati'e to error de'iations. Ahen these de'iations are sq ared6 added6 and then di'ided "y df-s6 they )i'e t&o 'ariances. Ahen the treat#ent 'ariance is 5si)nificantly7 )reater than the error 'ariance6 pop lation #ean differences are li?ely to e8ist. a7 de)rees of freedo# for 4actor: 4 ; 1 . 3 "7 de)rees of freedo# for 9rror: 20 ; 4 . 3/ c7 de)rees of freedo# for +otal: 20 ; 1 . 39 **+ . **+0 < **96 " t this does not eq al =*+0 < =*9. A co ntere8a#ple: Bet n . 21 r./ **+ . 100 **+0 . 21 **9 . 11 +hen **+ . **+0 < **9 . 21 < 11 . 100. : t . =*+0 + =*9 = **+0 **9 21 11 **+ + = + = 12 r 1 n r 1 11 n 1
9.11. 9-12.
9-13.
9-14.
9.11
9-1/.
Ahen the n ll hypothesis of A$%VA is false6 the ratio =*+0C=*9 is not the ratio of t&o independent6 n"iased esti#ators of the co##on pop lation 'ariance 2 6 hence this ratio does not follo& an F distri" tion. 4or each o"ser'ation xij 6 &e ?no& that
9-13.
xij ; = . 5 x i ; = 7 < x ij x i 5tot.7 . 5treat.7 < 5error7: x x *q arin) "oth sides of the eq ation: 2 2 =2 = 5 xij ; = x 7 . 5 x i ; x 7 < 25 x i ; x 75 xij ; x i 7 < 5 xij ; x i 7
9-2
$o& s # this o'er all o"ser'ations 5all treat#ents i . 16 . . . 6 r! and &ithin treat#ent i6 all o"ser'ations j . 16 . . . 6 ni:
i =1 j =1
ni
2 5 xij ; = x7 .
5 x
i= 1 j= 1
ni
2 ; = x7 <
25 x
i= 1 j= 1
ni
; = x 75 xij ;
xi 7<
i =1 j =1
ni
5 xij
x i 72
n5x
i
i =1
s ##and doesn-t 'ary o'er each of the ni7 'al es of j. *i#ilarly the second s # is 2
i =1
D5 x i ; = x 7 5 xij ;
j= 1
ni
xi
5x
j= 1
ni
ij
x i 7 . 0 since this is ,
st
the s # of all de'iations fro# the #ean &ithin treat#ent i. +h s the &hole second s # in the lon) 0.H.*. a"o'e is 06 and the eq ation is no&
i =1 j =1
ni
2 5 xij ; = x7 .
i =1
2 ni5 x i ; = x7 <
i =1 j =1
ni
5 xij ;
x i 72
&hich is precisely 9q ation 59-127. 9-12. 54ro# =inita"7: *o rce df ** =* F +reat#ent 2 321123 1901/3 20.31 9rror 23 2424/0 9202 +otal 29 /29123 +he critical point for F 526237 at . 0.01 is 1.49. +herefore6 re,ect H0. +he a'era)e ran)e of the 3 prototype planes is pro"a"ly not eq al.
5% '( )S F F%riti%"# p!*"#u& 2 190563.33 20.7084038 3.3541312 0.0000 R&+&%t 27 9202.2222 29
9.19.
9-3
S = 2.333
R-Sq = 55.19
Poole" ,e(el Mea& .--------/0 8 60.160 1-----$ Me2 8 58.390 /3E 8 55.190 45a& 8 54.124 .--------52.5 Poole" StDe( = 2.333 55.0 57.5 60.0 StDe( 2.535 2.405 2.224 2.149 StDe( .---------.---------.--------!-----!------1-----$ !------1------$ !-----1------$ .---------.---------.---------
Critical point F 536227 for . 0.01 is 2.94/3. +herefore &e re,ect H0. +here is e'idence of differences in the a'era)e price per "arrel of oil fro# the fo r so rces. +he 0otterda# oil #ar?et #ay not "e efficient. +he concl sion is 'alid only for 0otterda#6 and only for Ara"ian Bi)ht. Ae need to ass #e independent random samples fro# these pop lations6 nor#al pop lations &ith eq al pop lation 'ariance. %"ser'ations are ti#e-dependent 5days d rin) 4e"r ary76 th s the ass #ptions co ld "e 'iolated. +his is a li#itation of the st dy. Another li#itation is that 4e"r ary #ay "e different fro# other #onths.
9-20.
An F5.016261017 . 3./1 res lt6 relati'e to a critical 'al e of 3.02/336 indicates a si)nificant difference in their perceptions on the roles played "y African A#erican #odels in co##ercials.
df 2 32 40
=* 41.1213 3./9212
F 12.31
9-4
p-'al e . .0001. Critical point for F 526327 at . .01 is 3.241. +herefore6 re,ect H0. +here is a difference in the len)th of ti#e it ta?es to #a?e a decision.
ANOVA T"b#& 5% Sour%& SS '( )S F F%riti%"# p!*"#u& Between 91.0426 2 45.521302 12.3093042 3.2448213 0.0001 R&+&%t Within 140.529 38 3.6981215 Total 231.571 40
9-22.
An F5.01626117 . 12.323 res lt6 relati'e to a critical 'al e of 3.1/16 indicates a si)nificant difference in the #onetary-econo#ic reaction to the three inflation fi)htin) policies. +he test res lts e8ceed the critical 'al e of F5.0163623/7 . 3.2//. +he res lts indicate that the perfor#ances of the fo r different portfolios are si)nificantly different. 91F C.(. for the #ean responses: =artiniq e: x2 t C 2 =*9 C n2 . 31 1.9/ 104.4 C 40 . D/2.046 21.9/E 9le thera: 33 1.9/ =*9 C n3 . D//.046 39.9/E Garadise (sland: 91 1.9/ =*9 C n4 . D24.046 93.9/E *t. B cia: 21
9-23.
9-24.
1.9/
9-21.
Ahere do differences e8ist in the circle-sq are-trian)le pop lations fro# +a"le 9-16 sin) + ?eyH 4ro# the te8t: =*9 . 2.121 x1 . / trian)les: n1 . 4 sq ares:n2 . 4 circles: 4or
x 2 . 11.1
n3 . 3 . .016 q
5r6nr7
x3 . 2 . q 0.0153627 . 1./3 *#allest ni is 3: > 4.332 si). > 4.332 si). J 4.332 n.s.
9-2/.
4ind &hich prototype planes are different in Gro"le# 9-18: x A . 46403 x B . 46230 =*9 . 96202 ni . 10 for all i 4or
. .016 q
x C . 46131
536237
96202 C 10
. 10/.431
9-1
I x A x B I . 133
> 10/.431
si). n.s.
I x B x C I . 91 J 10/.431
I x A x C I . 232 > 10/.431 si). Grototype A is sho&n to ha'e hi)her a'era)e ran)e than "oth : and C. Grototypes : and C ha'e no si)nificant difference in a'era)e ran)e 5all concl sions are at . 0.017.
Tu4&- t&st (or 5"ir,is& %o35"rison o( 6rou5 3&"ns A r 3 7 Si6 7 n-r 27 C Si6 C q0 T 3.51 106.476
9.23.
*ince H0 &as re,ected in Gro"le# 9-196 there are si)nificant differences. T . q0.01546227 1.4433 C 2 . 3.332 IMN ; =9XI . I/0.1/ ; 12.39I . 1.33 IMN ; MA9I . I/0.1/ ; 11.19I . 4.97 IMN ; %=A$I . I/0.1/ ; 14.1232I . 6.0362 I=9X ; MA9I . I12.39 ; 11.19I . 3.2 I=9X ; %=A$I . I12.39 ; 14.1232I. 4.2662 IMA9 ; %=A$I . I11.19 ; 14.1232I . 1.0//2 All are J 0.226 th s not si)nificantas e8pected.
Tu4&- t&st (or 5"ir,is& %o35"rison o( 6rou5 3&"ns /0 r 4 )&2 )&2 n-r 28 /AE Si6 /AE q0 4.04 O3"n Si6 Si6 T 3.33248
9.22. 9-29.
5O estion has no rele'ance to 9-207 @e)rees of freedo# for 4actor: 3-1 . 2 @e)rees of freedo# for 9rror: 113 ; 3 . 114 @e)rees of freedo# for +otal: 113 ; 1 . 11/ +he o'erall 4 test indicates that there is a difference in the )ro ps- reaction to pricin) tactics. +he s "seq ent infor#ation also indicates that there is a si)nificant difference "et&een each of the )ro ps- reactions. a7 +otal sa#ple siPe . 231 "7 +he critical 'al e for 45.016 26 2327 is 3.029! therefore the o'erall A$%VA test is 'ery si)nificant. c7 =onopoly prices are si)nificantly different than li#ited co#petition and stron) co#petition.
9-30.
9-/
9-31.
Ae cannot e8tend the res lts to planes " ilt after the analysis. Ae sed fi8ed effects here6 not rando# effects. +he 3 prototypes &ere not rando#ly chosen fro# a pop lation of le'els as &o ld "e req ired for the rando# effects #odel. A rando#iPed co#plete "loc? desi)n is a desi)n &ith restricted rando#iPation. 9ach "loc? of e8peri#ental nits is assi)ned to treat#ents &ith rando#iPation of treat#ents &ithin the "loc?. 4ly all 3 planes on the sa#e ro te e'ery ti#e. +he ro te 5flo&n "y the 3 planes7 is the "loc?. Boo? at the resid als. (f the spread of the resid als is not eq al6 &e pro"a"ly ha'e neq al 2 6 the ass #ption of eq al 'ariances is 'iolated. A histo)ra# of the resid als &ill re'eal nor#ality 'iolations. %ther&ise yo are not rando#ly sa#plin) fro# a pop lation of treat#ents6 and inference is not 'alid for the entire Kpop lation.L $o. 0otterda# 5and Ara"ian Bi)ht7 &as not rando#ly chosen. (f the locations and the artists are chosen rando#ly6 &e ha'e a rando# effects #odel. 1. +estin) for possi"le interactions a#on) factor le'els. 2. 9fficiency. Bi#itations and pro"le#s: 517 Ae don-t ?no& the o'erall si)nificance le'el of the 3 tests! 527 (f &e ha'e 1 o"ser'ation per cell then there are 0 de)rees of freedo# for error. Also6 for a fi8ed sa#ple siPe there is a red ction of the df for error. 1. As #ore factors are incl ded6 df for error decreases. 2. As #ore factors are incl ded6 &e lose the control on 6 and the pro"a"ility of at least one +ype ( error increases. *ince there are interactions6 there are differences in e#otions a'era)ed o'er all le'els of ad'ertise#ents.
9-32.
9-33. 9-34.
9-31.
9-39.
9-40.
9.41.
9-42.
At = 0.05: Bocation: F . 10./6 si)nificant Qo" type: F . 10.2126 si)nificant (nteraction: F . 2.146 n.s.
9-3
ANOVA T"b#& Sour%& SS '( )S F Location 2520.988 2 1260.49 50.645 Job Type 2499.432 2 1249.72 50.212 nte!action 212.716 4 53.179 2.1367 "!!o! 1792 72 24.8889 Total 7025.136 80
A:C 10 10 10
C:* 10 10 10
$:C 10 10 10
df 2 2 4 441 449
=* 32.1 20 /0 14.0/
4ro# ta"le: F 0.015464007 . 3.3/ F 0.015264007 . 4.// +herefore6 all are si)nificant at . 0.01. +here are interactions. +here are $et&or? #ain effects a'era)ed o'er $e&sti#e le'els. +here are $e&sti#e #ain effects o'er $et&or? le'els. 9-44. a. ". c. d. e. a. ". c. d. e. f. ). Be'els of tas? diffic lty: a ; 1 . 1! therefore a . 2 Be'els of effort: b ; 1 . 1! therefore b . 2 +here are no tas? diffic lty #ain effects "eca se p-'al e . 0.1313 +here are effort #ain effects "eca se p-'al e J 0.0001 +here are no si)nificant interactions6 as p-'al e . 0.1/49.
9-41.
98plained is K+reat#entL: +reat . 4actor A < 4actor : < 5A:7 Be'els of e8ercise price: a ; 1 . 2! therefore a . 3 Be'els of ti#e of e8piration: b;1 . 1! therefore b . 2 ab5n ; 17 . 1446 a . 36 b . 2! therefore n ; 1 . 246 n . 216 N . 21 / . 110 n . 21 +here are no e8ercise-price #ain effects 5F . 0.42 J 17. +here are ti#e-of-e8piration #ain effects at . 0.01 " t not at . 0.01 "eca se F 5161447 . 4.241. 4ro# the F ta"le6 for df-s . 16 110: critical point for . 0.01 is 3.91 and for . 0.01 it is /.21. h. +here are no interactions: F . .193 J 1 i. +here is so#e e'idence for ti#e-of-e8piration #ain effects. +here is no e'idence for e8erciseprice #ain effects or interaction effects. ,. 4or ti#e-of-e8piration #ain effects6 .01 J p-'al e J .01. 4or the other t&o tests6 the p-'al es are 'ery hi)h. ?. Ae co ld se a t-test for ti#e-of-e8piration effects: t 251447 . F 5161447
9-2
9-4/.
*ince there are interactions " t neither of the #ain factors ha'e si)nificant 4-tests6 a li?ely concl sion is that the t&o factors &or? in opposite directions6 i.e.6 in'erse to each other. Ad'anta)es: red ced e8peri#ental errors 5the effects of e8traneo s factors7 and )reater econo#y of sa#ple siPes. Mse "loc?in) "y fir#6 to red ce the error contri" tions arisin) fro# differences "et&een fir#s. Co ld se a rando#iPed "loc?in) desi)n: 4 o"ser'ations6 MN6 =e8ico6 MA96 %#an at 4 locations and 4 different dates. A )ood "loc?in) 'aria"le &o ld "e siPe of fir# in ter#s of total assets or total sales6 etc. Res. Ha'e people of the sa#e occ pationCa)eCde#o)raphics se s&eaters of the 3 ?inds nder st dy. 9ach )ro p of 3 people are a "loc?. As stated in 9-236 a )ood "loc?in) 'aria"le &o ld "e so#e #eas re of di'ersity in the portfolio. Ae co ld )ro p the e8ec ti'es into "loc?s accordin) to so#e choice of co##on characteristics s ch as a)e6 se86 years e#ployed at c rrent fir#6 etc. +he different "loc?s for the chosen attri" te &o ld then for# a third 'aria"le "eyond Bocation and +ype to se in a 3-&ay A$%VA. Ae # st ass #e no "loc?-factor interactions. **+0 . 36233 **9 . 12632/ n . 100 "loc?s df error . 5n ; 175r ; 17 . 99527 . 192 df treat#ent . r ; 1 . 2 F . =*+0C=*9 .
36233 C 2 . 21.24 12632/ C 192
9-43.
9-42. 9-49.
9-10. 9-11.
9-12. 9-13.
9-14. 9-11.
0e,ect H0. p-'al e is 'ery s#all. +here are differences a#on) the 3 s&eeteners. *ho ld "e 'ery confident of res lts. :loc?in) red ces e8peri#ental error here6 as people of the sa#e &ei)htCa)eCse8 &ill tend to "eha'e ho#o)eneo sly &ith respect to losin) &ei)ht. 9-1/. n . 30 r.4 **+0 . 96231 **:B . 16441 **+ . 2263/4 **9 . 2263/4 ; 16441 ; 96231 . 116 044 =*9 .
116044 96231 . 13.31 =*+0 . . 36291./3 5/97537 3
F 5362037 . =*+0C=*9 . /1.3 0e,ect H0. p-'al e is 'ery s#all. $ot all of the fo r #ethods are eq ally effecti'e.
9-9
9-13.
**+0 . 36102 **9 . 106111 r . 2 ni . 20 for all i =*+0 . **+0C5 r ; 17 . 36102C3 . 16014.13 =*9 . **9C5n ; r7 . 106111C51/0 ; 27 . /9.11 F 5361127 . 14./3 > 2.3/ 5crit. point for . 0.017. +herefore6 re,ect H0. $ot all tapes are eq ally appealin). p-'al e is 'ery s#all. n1 . 32 n2 . 30 n3 . 32 n4 . 41 n .141 =*+0 . **+0C5 r ; 17 . 46133C3 . 16112.33 F 5361337 . =*+0C=*9 . 16112.33C412 . 3./3 5at . 0.017 2./3 J 3./3 J 3.92 5at . 0.017 Ae can re,ect H0 at . 0.01. +here is so#e e'idence that the fo r na#es are not all eq ally &ell li?ed. *oft&are pac?a)es: 3 ** soft&are . 336/41 ** co#p ter . 146121 ** int. . 226/99 **9 . 4346113 n . /0 *o rce soft&are co#p ter interaction error +otal Co#p ters 4
9-12.
9-19.
df 2 3 / 302 319
:oth #ain effects and the interactions are hi)hly si)nificant. 9./0. +reat#ent df . 5r-17 . 2 :loc? df . 34 +otal df . 224 +otal sa#ple siPe &as 221: 9rror df . 5n-175r-17 . 5347527 . 142 Critical 'al e of 45.016 26 1427 . 3.01326 &hich is less than 4 . 13./1. +he res lts are si)nificant.
9-10
9-/1. *o rce pet location interaction error +otal ** 226241 346111 316332 1146392 /426932 df 3 3 9 144 119 =* 36411 116113 36130.29 36249.99 F 1.93 2.99 0.92
+here are no interactions. +here are no pet #ain effects. 5 . 0.017 2./2 J 2.99 J 3.92 5 . 0.017 +h s there are location #ain effects at . 0.01. 9-/2. F-ratio . 4.1431 p-'al e . .0132 5 sin) a co#p ter7. At . 0.016 only )ro ps 1 and 3 are si)nificantly different fro# each other. @r ) )ro p is si)nificantly different fro# the $o. +reat#ent )ro p.
5% )S F F%riti%"# p!*"#u& 2 1601.56 4.54708749 3.123901138 0.0138 R&+&%t 72 352.21667 74
'(
Con(i'&n%& Int&r*"#s o( 8rou5 )&"ns #!o$p %on&i'ence nte!(al Dru6 24.16 ) 7.4824354 P#"%&bo 27.8 ) 7.4824354 N0!Tr&"t3&nt 39.48 ) 7.4824354
Tu4&- t&st (or 5"ir,is& %o35"rison o( 6rou5 3&"ns Dru6 r 3 P#"%&bo P#"%&bo n-r 72 N0!Tr&"t3&nt Si6 N0!Tr&"t3&nt q0 3.41 T 12.7994
9-/3.
a. :loc?in) 5repeated #eas res7 is #ore efficient as e'ery person is hisCher o&n control. 0ed ctions in errors. Bi#itationsH =ay"e carryo'er effects fro# trial to trial.
9-11
9-/4.
". **+0 . 446132 **9 . 1126/32 r.3 n. 30 =*+0 . 446132C2 . 22622/ =*9 . 1126/32C5297527 . 16942./2 F 526127 . 11.43. 0e,ect H0. n1 . n2 . n3 . 11 r . 3 A one-&ay A$%VA )i'es an F-'al e of 22.216 &hich is si)nificant e'en at J 0.0016 hence &e re,ect the hypothesis of no differences a#on) the three #odels. =*9 . 42.16 so at . 0.01 &e se the critical point q . 4.33 5closest to the req ired 'al e for df-s . 36 4276 )i'in) the + ?ey criterion T . q x GI . 124.33 x P . 121.40 x Z . 102.33 *o:
=*9 C ni
Msin) T . 3.236 &e re,ect the hypothesis of x GI . x Z and also x P . x Z 5at the 0.01 le'el of si)nificance76 " t not the x GI . x P hypothesis.
5% )S F F%riti%"# p!*"#u& 2 1068.8889 22.2082976 3.219938094 0.0000 R&+&%t 42 48.130159 44
'(
Con(i'&n%& Int&r*"#s o( 8rou5 )&"ns #!o$p %on&i'ence nte!(al 8I 124.733 ) 3.6149467 P9i##i5s 121.4 ) 3.6149467 :&nit9 108.733 ) 3.6149467
Tu4&- t&st (or 5"ir,is& %o35"rison o( 6rou5 3&"ns 8I r 3 P9i##i5s P9i##i5s n-r 42 :&nit9 Si6 Si6 q0 4.37 T 7.82789
:&nit9
9-/1.
**9 . 426223.923
9-12
9-/3.
9./2. 9./9.
A one-&ay A$%VA stron)ly re,ectin) H0. 4or the three le'els of *tore6 91F confidence inter'als are calc lated for #eans6 as sho&n6 &hich do not o'erlap at all. Case 11: 0atin) Aines 5+e#plate: A$%VA.8ls6 sheet: 1-Aay7 data:
n 1 2 3 4 5 6 7 8 9 10 11 11 C9"r' 89 88 89 78 80 86 87 88 88 89 88 10 13 11 )&r#ot C.7#"n% C.S"u* 91 81 92 88 81 89 99 81 89 90 82 9 91 81 92 88 78 90 88 79 91 89 80 93 90 83 91 87 81 97 88 88 85 86
17 @o not re,ect the n ll hypothesis6 there is no difference in the a'era)e ratin)s d e to the type of )rape.
ANOVA T"b#& Sour%& SS '( Between 411.617 3 Within 6545.63 41 Total 6957.24 44 5% )S F F%riti%"# p!*"#u& 137.21 0.8594 2.8327 0.4698 159.65
9-13
1. ANOVA
n 1 2 3 4 5 6 7 8 9 10 10 10 10 S%"n1 S%"n; S%"n< 16 13 18 15 18 19 12 13 15 15 15 14 16 18 19 15 14 16 15 15 17 14 15 14 12 14 15 14 16 17
ANOVA T"b#& Sour%& SS '( Between 20.6 2 Within 79.7 27 Total 100.3 29
)S 10.3 2.9519
0e,ect the n ll hypothesis of eq al n #"er of scans per #in te. 2. 0o&s . Cler?s6 col #ns . scanners
5% )S F F%riti%"# p!*"#u& 5.19167 2.1239 2.5787 0.0934 45.35 18.552 3.2043 0.0000R&+&%t 1.76667 0.7227 2.1521 0.6705 2.44444
ANOVA T"b#& Sour%& SS '( *ow 20.76667 4 %ol$+n 90.7 2 nte!action 14.13333 8 "!!o! 110 45 Total 235.6 59
0e,ect the n ll hypothesis of eq al n #"er of scans per #in te 5col #ns7 @o not re,ect the n ll hypothesis that the cler?s are eq ally efficient. +here are no interaction effects present.
9-14