103'!%@!>I'!4VF01'A!A'M20>2%#4!@1%&!

"" " #"9!

%&'(')&')*$*,.2#-'&$'"#$5)&'(.*#$6#'G##.$'"#$'#&'$&'(')&')*$(.5$- 4$).',$($*,8.'$
#" "
34560)&,.&,620)7(+*,-2!
0F4'! !"!0#A! "#"!01'!0M'103'4*!>I'!R'#>10C!S2&2>!KI'%1'&!0CC%J4!
.5(-5$#--,-&9$$
!"
7'*'30("#'($,!)"*(5"4(!,#(!8,#9':(.4)(,)(,(!"0)1(;"#$%&'#!'(
%!F4'!=%1&0C!X%A'C4!>%!0??1%B2&0>'!>I'21!40&?C2#3!
## $
%!" $ !"% & $ % $
-.%'$<),(!$'>)$0!6#2!"#$!&)<-./!)!,)$/'!()*+,':!;&'!,)$/'$!%&'!
" $%%!"$ $ !"% &
>12GF>2%#49!KI0>!C'0M'4!>J%!>I2#34!@%1!F4!>%!@23F1'!%F>!@%1!'0$IY!
()*+,'=!%&'!(&#$%'$!%&'!-.%'$<),:!
/).2#.25#(+.25.'5#5)81.+1%2;##<,.+4(#+,)#9).&+7#%"#+,)#=><?#1+#
"
-('),$)&$*(11#5$($*%&'(')&')*$6#*(8&#$)'$(1&,$*,8.'&$'"#$.8/6#-$,+$#&')/('#5$
;&'!$'02?%-#.!-.!%&'!*)$/-.!#"!'$$#$!)?&-'<'0!>-%&!)!,)$/'$!
18-comparisions
F$'4!>I'!>04T!%@!$I%%42#3!0!40&?C2#3!A24>12GF>2%#!>%!$I%%42#3!0!
• When the levels of one factor (diet) are associated with the levels of another factor (location), we say that these two factors are
(-5$#--,-&$'"('$&#0(-('#$($'#&'$&'(')&')*$+-,/$($.811$"70,'"#&)&9$!"#$#&')/('#5$
()*+,'!-(!.#%!+$#+#$%-#.),!%#!%&'!-.?$')('!-.!%&'!()*+,'!(-@':!;&'!
0#!0#A!0!4>0#A01A!A'M20>2%#9! confounded. Randomization eliminates confounding, simplifying the interpretation of data.
• " we reject H0 if the p-value of the test is less than 0.05.
(-5$#--,-$,+$'"#$5)++#-#.*#$6#'G##.$(2#-(3#&$,+$).5#0#.5#.'$&(/01#&$)&$
/,*9%#("$('**"*(&'!*',0'0(<%)8(=-:!A(!)!$'(2,%=!"#$!-.(%).?'=!%#!?2%!
• The estimated standard error of the difference between averages of independent samples is
'!&'0#!%@!>I'!40&?C2#3!A24>12GF>2%#!24!>I'!?%?FC0>2%#!
%&'!*)$/-.!#"!'$$#$!-.!&),"=!6#2!.''0!B!%-*'(!)(!*).6!?)('(:!C#(%(=!
$$% $%%
0&'>'1*!'2>I'1!-!%1! # 9!/0$I!?010&'>'1!C2'4!0>!>I'!$'#>'1!%@!2>4!
$%"!" $ $ !"% # # '
&#>'<'$=!%6+-?),,6!$-('!-.!+$#+#$%-#.!%#!-1(>":(5"43&(06'#&(?()%/'0( $
&?C2#3!A24>12GF>2%#9!Z%1! !" &
!"*!>I'!&'0#!%@!>I'!40&?C2#3! $ & %
)(!*2?&!%#!(,-?'!%&'!*)$/-.!#"!'$$#$!-.!&),":!
• Conditon for doing t-test for se.
>12GF>2%#!24!-*!>I'!?%?FC0>2%#!?1%?%1>2%#9!!Z%1!
o SRS condition (Simple Random #"*!>I'!&'0#!%@!
(#:&!**(-#*+!"#32&4;$%#;"!-'#'(-(&!/#";?"/&!5*"#@.A%-<#@0A#*2#<!"*!-';!"+#*+(# Sample)
40&?C2#3!A24>12GF>2%#!24!#9! o Similar variances
o No evident lurking
'&2;5"8#B2&#*+!"#(C%45$()#!*,"#42&(#;"(3;$#*2#:&!*(#*+!"#32&4;$%#%"$
factor

" " • Two-sample t-tests and two-sample

o Nearly normal
'!4>0#A01A!A'M20>2%#!%@!>I'!40&?C2#3!A24>12GF>2%#!24!>I'!
7"<(-,*9'(,(0,/6-'(&"(<'(#''&@(A(0%/6-'(,#0<'*(%0(B/"*'(%0( $'%Bothconfidence
$(% intervals for the difference between two means allow us to compare results
$%"!" $
#A01A!'11%19!Z%1!0#!0M'103'!%@!2#A'?'#A'#>!%G4'1M0>2%#4*!!"
.'))'*:C(.4)(&,),(!"0)()%/'(,#&(/"#'51(7"<(/4!8(%0('#"498@(
obtained
'
from two
( # # "
samples. ' $ rely on standard error of the difference between two sample averages and the use
procedures
of a t-model for the sampling & distribution.
&( Experiments provide the ideal data for such comparisons. In an experiment, subjects
'>I'1!0!?1%?%1>2%#!%1!#%>*!>I'!@%1&FC0!@%1!>I'!4>0#A01A!'11%1!24!
D'$"*'(5"4(.'9%#()"(!"--'!)(&,),:(%)30(,(9""&(%&',()"(E#"<(<8')8'*(
& from a sample are randomly 'assigned !"#$%&'()*&+*#,&-*./012#
to treatment groups defined by levels of the experimental factor. This randomization
avoids confounding that introduces the possibility of lurking factors.
05*!!
%&'!()*+,'!(-@'!6#2!?).!)""#$0!>-,,!1'!)0'D2)%'!"#$!>&)%!6#2!
$0183$).$'"#$#&')/('#&$+-,/$!(61#$>P%@=$'"#$&'(.5(-5$#--,-$)&$
16-confidence interval
'
!!"#$'!"#$'%%&!'(!)*+!,-+.!/012!34,53221*6!(*5!
>).%!%#!,')$.!)1#2%!%&'!+#+2,)%-#.:!
%& $'"$&'() % & %( ! $%'(*% !!16/*!/03!(*57+-8!
*5!/03!2/869859!355*5:!)*+!.3/!/03!2/869859!355*5!(*5!,5*,*5/1*62;!
$%"!""' $• !"( # #
E"!6#2!F.#>!%&'!.''0'0!*)$/-.!#"!'$$#$=!6#2!?).!'(%-*)%'!%&'! ' $D#E8EF#52;-<"$
((
I'!#F&'10>%1*!'!24!>I'![Q!%@!0!42#3C'!%G4'1M0>2%#*!0#A!2!24!>I'! (!
#!"& ##
#'!'00,*5(0,/6-'(0%F'1(2)30("#-5(,(94'00(.'!,40'(5"4(<"#3)(E#"<(
%&'(')&')*$+,-$'"#$'#&'$,+$- !" L$#!!"$%!% $7#0#!"#*+;"$
#$
&?C'!42\'9!]I'#!0M'1032#3!2#A2$0>%14!80!.^5!AF&&D!M0120GC'<!
2.%-,!6#2!/'%!%&'!()*+,':!G)*+,'!(-@'!?),?2,)%-#.(!)$'!$)$',6!'H)?%:!
4 > K
!
%
" •• A confidence
;#!/'%!)!+)$%-?2,)$!*)$/-.!#"!'$$#$=!(#,<'!"#$!-!-.!%&'!"#$*2,)3!
"!" $ !"(interval # $ %provides ! a range
+'&% !"34# parameter based on the data in a sample.
$ % of plausible values for a population
"#-./'*%0&/"#&1%*2*%'-*"0&
" ## '
!# #
#
"# !( $',$$
!" $%"!"" ' $ !"($ # ('()
" !
"
<!2,3=18-!53-8/1*6201,!>*162! !"
#$ ! #
869!(&!?6=3!)*+!@6*A! !" ! !":!)*+!@6*A!
#$
•
• Confidence intervals provide #
;&'!.'?'(()$6!()*+,'!(-@'!0'+'.0(!#.! :!I'!?)..#%!'(%-*)%'!
a range #! of a population. The coverage (or confidence level) of a
of plausible values for a parameter
!%&'("&)*"(+,*"-*.+&/01()/2"+("&)+&"3*&4**1"&)*",*+1"+15"6+-/+17*"
confidence interval is the probability that this procedure produces an interval that includes the parameter. Most often,
>-%&!2!1'?)2('!>'!&)<'!%#!?&##('!-!60.,)0!?#,,'?%-./!%&'!()*+,':!E"!
(!8!B356*+--1!5869*7!C8518D-3!*:!A0353!E85"*%!F!'"#$'%&!
confidence intervals& have coverage 0.95 and are known as 95% confidence intervals. The margin of error is the half-length of
'()%&
"
>'!&)<'!.#!-0')!#"!! # the 95% confidence interval. The one-sample z-interval for the mean of a population is   y ± 2 s/   n. This confidence interval
=!-%!>#2,0!1'!)!/##0!-0')!%#!#1%)-.!)!(*),,!
#$"!12!/03!7386!*(!8!=*-+76!*(!6+7D352!=*939!82!#!38=0!/173!/03!
presumes a large sample. This same interval applies to proportions of large samples, with s2 estimated as ˆ p (1- ˆ p ). The
()*+,'!%#!'(%-*)%'!"#:!A!'+(-.*/.012!?#,,'?%(!)!(*),,!()*+,'!#"=!
endpoints of a confidence interval should " be rounded to presentation precision by applying the 3-to-30 Rule to the standard
C36/!*(!16/3532/!08,,362!869!=*939!82!G!*/035A123&!!<!=*-+76!*(!
error.
()6=!JK!%#!LK!?)('(!%#!'(%-*)%'!
• Difference 0.4191 #:! t Ratio 0.879117
691=8/*52!*(!86!3C36/!-1@3!/012!12!=8--39!8!+,--.&/0120345&!!H03!
o Std Err Dif 0.4768 Prob > |t| 0.3802
2#()8'(06'!%,-(!,0'("$(6*"6"*)%"#0:(<'(&"#3)(#''&(,(6%-")(0,/6-'1(
o Upper CL Dif 1.3583
0.8,1"/(1'&"58,39"":0-3!-1@3-)!/03!6873!=*732!(5*7!/03!(8=/!/08/!
o Lower CL Dif -0.5200
;&'!$',)%-#.(&-+!1'%>''.!#o!).0!8!),,#>(!6#2!%#!'(%-*)%'!-!
03!9+77)!C8518D-3!12!9351C39!(5*7!86*/035:!7*53!+23(+--)!=*939!
Confidence 0.95
o
>-%&#2%!F.#>-./!).6%&-./!)1#2%!8:!E"!6#2!,##F!)%!.)%-#.),!
*-+76&!!'6!/012!3487,-3:!/03!9+77)!C8518D-3!1691=8/32!A03/035!
o 18. Which of the following is INcorrect?
04*+'50:(5"43--(#")%!'()8,)(/"0)("$()8'/(8,+'(,."4)(G:GHH(6'"6-'1(
03!=+2/*735!8==3,/2!86!8,,-1=8/1*6;! o
o
A. The mean difference of about 0.4 is statistically insignificant at the 5% level.
B. The null hypothesis that the population means are the same cannot be rejected at the 5% significance level.
I"43--(,-0"(#")%!'(68*,0'0(-%E'(BJ8'0'(*'04-)0(,*'(,!!4*,)'()"(K(L(
62&F!#! 1(!8,,-1=8/1*6!12!53/+5639!
o C. The values between –0.52 and 1.36 cannot be rejected as possible population mean differences at the 5%
significance level, and this holds true in particular for the value zero.
6"%#)01C(J8'(KLM(%0()8'(/,*9%#("$('**"*1(2)()4*#0("4)()8,)()8'0'()<"(
62&F!G! 1(!8,,-1=8/1*6!12!6*/!53/+5639!
o The value of |t| says that the observed mean difference 0.419 is 0.879 standarderror estimates away from zero.
+$#+'$%-'(=!J=JKK!$'(+#.0'.%(!).0!)!L9!*)$/-.!#"!'$$#$=!?#*'!
03!7386! &$!12!/03!,5*,*5/1*6!*(!/1732!/08/!8!6
o The p-value says that the!12!#:!A08/!A3!
2
probability of observing a value of |t| > 0.879 in other datasets is about 0.38, assuming
the null hypothesis of equal population means is correct.
%#/'%&'$:! #$
2+8--)!-8D3-! !"&!H*!233!/08/!
• The central&$!!F! !"theorem
limit#$ states that means across datasets are&$
:!A51/3!/03!6+7358/*5!*(! !82!/03!
ever more normally distributed as N → ∞.
• The standard deviation of proportions across datasets shrinks at the rate 1/N ½
!
+7!*(!/03!62&!!<!7*53!2+5,51216.!/016.!08,,362!/*& ' &!H03!2+7!*(! ! $
!"3!4#
03!2I+8539!93C18/1*62!8D*+/!
"
* Complex Bayes &$!217,-1(132!-1@3!/012;!
formula: reversing conditioning
% %

'# '#
P(A|B)*P(B)$
" "& & &$
P(B|A) $ $
$
="-----------------------------------
% & ( & #$ !" "
(
P(A|B)*P(B) + P(A|not B)*P(not" B)
(%" (%"

" numerator = P(A and B)$

Proof:
denominator
'#
% = P(A
%& $ '#
!"B) +(P(A and"&
and#$
$
!" !
not #$
B) $
&( % % & ( %"

!"$ ( %" #"& #$

% !% & %" # #$ !"$ $

03!=*+6/!7#!12!/03!6+7D35!*(!*632:!2*!7 #$ !"!F!7#&!'(!A3!,-+.!16!/012!
4,53221*6!(*5!7#:!A3!.3/!
" %

' #& ( & &$$$ % %!"& #$!"# #$!"$ ( %#$"&

!"# #$!"$
$

(%" "
# #$
% %#$"&
!" !"$!"& #$( !" #$
!"# !
 ! ! 52)/-2;+/2.3!.0!/'(!/()/!)/,/2)/24!
!""#"$%&'('"')*'%+#,"*'%)#-%(#,)./"#$%$! $% !2)!
!""#"$%&'('"')*'%+#,"*'%)#-%(#,)./"#$%$!
! ! &! , )!
! -$ +!!
!"#$%&#'(&)&*"+&,)-."%&*/'&"!&'(&!()*!(++,-*.!/0!#12!34!"!5!1267!8&'*!9:))*;&! $% # ) (! '! & *!
)(&*!,4!<=(>?!&'*;!&'*!=),/(/@+@&0!,4!(!<(>=+*!,4!"11!>*<<(A*<!-@&'!+*<<!
&'(;!"#B!<=(>!@<! !0+-1-2+-2*% !
! "!"#$%&'($)*)+%&%,,(,# $ ! " "#% !" -./0#&'(!3+>>!'@8./'()2)!/(>>)!+)!-:!;+/!),@)!3./'23<!,;.+/!,A!!B.
!
'! "#& !" "8-.8.-/2.3:!C3.D23<!.3(!8,-,6(/(-!2)!(3.+<'A!E21(3!":!@.
-./0 & " *!
$ !)! FG!.0! % *% ,!
!"!"#$%&"'"H"$"IJ#A!&',/!5.()!3./!D.-C!0.-!/'(!6(,3:!'.
(! "1/& "2K3.D23<!
3$ "1/& "2 3$ +!
-!5.()!3./!268>@! ,A!B.-!/()/)!.0!-:!D(!3((5!/.!()/26
!
'! -./0 & -.45
/'(!5,/,A! *!
F igure 15-6. Sa m$pling %
!)!%distribution for !" under H0,! . !
(! -.451/& -.452 3/-- +!
! ()*+*%&",-"./0#.1&"2).#2*3"+*45,2*",
!"#$%&'()*+"&,*#&-,.&'/+&#-"&,0%"+1"2&%/345"&3"/6& #"!;@!/'(!)/,35,-5!5(12,/
!&'()!*+,-!./(!
- ! "% % &4./ #
),68>(:!$A!!L'(3!D(!()/26,/(!
/01,./()'2(3!-(45!"67!8/(!).4534+3!(++,+!1+,9'3()!./(!):4&(7!!8/(! ,!0-.6!/'(!5,/,!/.!0235!/'(!)/,
$ -.-/6 ! 6*"5,/*5"#$"6#$)","5.6*+"2,&*"7$%8!
)4-1&(!-(45!453!./(!/01,./()'2(3!-(45!&'(!
z-statistic +
8C<!:<:(+D!$!<&(;.<!4,)!(!<&(;.().!;,)>(+!)(;.,>!E()@(/+*2!F'*! !
!"#$%&$'()'*)&"+&)+&%&$+&$,)*-'.) +,"%$%# $ !
3!%4+/%+'% $ # / &
0'/1)12-2#(&3/4"%&*'/4&#5"&6/'4)-&#)1-"&/'&%/*#+)'"78&95"'"$%&/6-(&)&0+&$.%&"1)+&%(1%-1)"--'-)'*)&"+&)+&%&$+&$,
3!A!&'(!/-+(!)/,35,-5!(--.-!/',/! /.!52)/23<+2)'!2/!0-.6!/'(!8.8+>,/2.3!*+,3/2/@!FGH $% IM,9'#
!
"2%B!9'(;9*!4,)!(!4(+<*!=,<@&@E*!*)),)!@4!"!5!12672!
5.()!3./!-(*+2-(!),68>(! #"$ "
2,4&:"7;<8"#&"$)*"&$,-=,+="*++.+".>"$)*"$*&$"&$,$#&$#2?"@+#$$*-"
# /
!
()/26,/()A!U-.5+4()!,!"0+-1-2+-2*A!
!"#$%&#'(&)6/#5"'&35/23"&*/'&"D!,;*!&'(&!@<!<>(++*)!/:&!<&@++!@;!&'*!)*A@,;!
#$A!&'(!()/26,/(5!)/,35,-5!(--.-!
& 2 '
5*$$*+&:"7$%8"#=*-$#>#*&",-"*&$#A,$*"$),$"&1/&$#$1$*&"$!0.-!
! ,A!G1
<=*9@4@*.!/0!# 1 2!34!"!5!1266D!&'*!<=(>$4@+&*)@;A!<,4&-()*!@<!/*&&*)!&'(;!&'*!
/',/!+)()!),68>(!()/26,/()A! 345 $ 3//
)+;)/2/+/2.3:!D(!4.3/23+(!/.!+)(!/'(!=.-6,>!?.5(>!0.-!/'(!) !
# % 779:
9:))*;&!<0<&*>D!/:&!;,&!A,,.!*;,:A'!&,!=(0!4,)!@&<*+42!F'*!9'(;9*!4,)!(!
U-.5+4()!,%!0+-1-2+-2*A! 52)/-2;+/2.3!+3>())!#!2)!)6,>>!H#!N!9OIA!
6572 483
t-statistic
4(+<*!=,<@&@E*!*)),)!@;9)*(<*<D! ).4534+3!(++,+)!414+.7!8/(!+4.',!,*!./(!3(9'4.',5!,*!./(!.().!).4.').':!*+,-!
B*+*%&").6"$)*"$*&$"6.+C&"#-"$)#&"*D,A45*?"()*"A*,-".>"$)*
'! "
-./0 & "6 *! '! *!
!.,!./(!#$%&'(%#)!).4534+3!(++,+!')!;5,<5!4)!4!!0+-1-2+-2*!=,+%-0"1-2#>7!
52)/-2;+/2.3!,/!/'(!;.+35,-@!.0!&
-./0 & -.44 O!2)!P#OO!8(-!6.3/'!,35!/'
!)!% % ,!$ ! )!% % ,!
8/(!%$).4.').':!:,?5.)!./(!5?-@(+!,*!#$%&'(%#)!).4534+3!(++,+)!./4.!
)/,35,-5!(--.-!$9'#!F"GHI9'JKL"M"NK?OP"4*+"A.-$)?"()#&"E+,
(! "1/& "2 3$ +! (!! -.441/& -.442 3/-- +!
)(14+4.(!./(!.().!).4.').':! #"!*+,-!"67!A,+!&4+B(!)4-1&()!=*!7&89:;&."&,0#/(6&
),68>23<!52)/-2;+/2.3Q!P#RS!>2()!).!0,-!,D,@!0-.6!/'(!),68>
- ! " % % &/.7 # !
4!+$94&?(!*,+!./(!%$).4.').':!4)!<(!3'3!*,+!./(!,$).4.').':7!
52)/-2;+/2.3!+35(-!& O!/',/!@.+!4,33./!(1(3!)((!/'(!,-(,!/',
$ -.-50 /'(!"$1,>+(T!
8/(!/?B(!%$).4.').':!.(&&)!?)!./4.!./(!-,9(!.,!C(59(+!)/,?&3!@(!1+,*'.4@&(7!!
F'*!9'(;9*!4,)!(!4(+<*!=,<@&@E*!*)),)!@<!<>(++D!/:&!A),-@;A2!F'@<!4@A:)*!
!
8/')!)4-1&(!1+,9'3()!).+,5B!(9'3(5:(!./4.!49(+4B(!+(5.)!4+(!/'B/(+!./45!
<',-<!&'*!<(>=+@;A!.@<&)@/:&@,;!,4! !"
"#! 4,)!&')**!E(+:*<!,4!"G!1267D!1266D!
D#667!E*!"6!(%&#+*";&#-"+"$%&/53,%#&6,&<-/6<"&,'&)"##(6)&/&%/345"&.(#-&
!
(;.!1261D!&'*!/,:;.()0!,4!&'*!;:++!'0=,&'*<@<2!
49(+4B(!+(5.!D#%F!,+!-,+(!1(+!-,5./7!G!+(5.!,*!D#66!')!:,--,5!*,+!,5(!
1+,1(+.0H!@?.!5,.!*,+!./(!-(45!,*!4!)4-1&(!,*!%I#!1+,1(+.'()7!
!
3,441"56%7)'0+2.'.%8'+-%#(%1%9'1)%
! !
-.+/0(%&.*!+(1('#%#1! "!!
2/00!34+.%3#$&$! "6J!"!=&"6! "6J!"!7&"6!
50%#1*(%&6#!34+.%3#$&$! "4J!"!K!"6! "4J!"!L!"6!
7#$%!$%(%&$%&8!
! #"!
F igure 15-2. Sa mpling distributions for p = 0.20, 0.22, or 0.24.
9%(*)(1)!#11.1! $#= !" >!M!$>?*!
F'*!9'(;9*!4,)!(!4(+<*!=,<@&@E*!*)),)!@<!&'*!<'(.*.!()*(!,4!&'*!<(>=+@;A!
:#;#8%!"<!&=! +$94&?(!L!&! +$94&?(!L!&!
.@<&)@/:&@,;!@;!H@A:)*!"#$62!F'*!()*(!A),-<!(<!"!(==),(9'*<!&'*! !
.1! = #"!$!"6>N$#=
/,:;.()0!,4!&'*!)*A@,;!(++,-*.!/0!# ! !" >!K!,&!! = !$!"6>N$#= !" >!L!$,&!!
1D!&'*!/)*(I$*E*;!E(+:*!12612!34!"!5! #"
:7;:<&#5"'"$%&)&-)'="&35)63"&*/'&="##26=&)&%)40-"&#5)#&*//-%&#5"&#"6#)#2,"&
.*9@<@,;!):+*!@;&,!)*J*9&@;A!#12!34!"!5!1261D!&'*!9'(;9*!4,)!(!4(+<*!=,<@&@E*!
8F0=*!3?!*)),)!@<! ! ! ! !
! ! ! ! "#$%! "#$"%!