8/18/2019 Chapter 10 Elementary Statistics
1/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 1
!ecture "lides
Elementary Statistics Eleenth Edition
and the #riola "tatistics "eries
$y %ario &. #riola
8/18/2019 Chapter 10 Elementary Statistics
2/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 2
Chapter 10Correlation and Regression
10-1 Review and Preview
10-2 Correlation
10-3 Regression
10-4 Variation and Prediction Intervals
10-5 Multiple Regression10- Modeling
8/18/2019 Chapter 10 Elementary Statistics
3/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 3
!ection 10-1
Review and Preview
8/18/2019 Chapter 10 Elementary Statistics
4/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 4
ReviewIn Chapter " we presented #ethods $or #a%ing
in$erences $ro# two sa#ples& In !ection "-4 weconsidered two dependent sa#ples' with eachvalue o$ one sa#ple so#ehow paired with avalue $ro# the other sa#ple& In !ection "-4 we
considered the di$$erences (etween the pairedvalues' and we illustrated the use o$ h)pothesistests $or clai#s a(out the population o$di$$erences& *e also illustrated the construction
o$ con$idence interval esti#ates o$ the #ean o$all such di$$erences& In this chapter we againconsider paired sa#ple data' (ut the o(+ective is$unda#entall) di$$erent $ro# that o$ !ection "-4&
8/18/2019 Chapter 10 Elementary Statistics
5/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 5
Preview
In this chapter we introduce #ethods $ordeter#ining whether a correlation' orassociation' (etween two varia(les e,ists andwhether the correlation is linear& or linear
correlations' we can identi$) an e.uation that(est $its the data and we can use that e.uationto predict the value o$ one varia(le given thevalue o$ the other varia(le& In this chapter' we
also present #ethods $or anal)/ingdi$$erences (etween predicted values andactual values&
8/18/2019 Chapter 10 Elementary Statistics
6/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 6
Preview
In addition' we consider #ethods $oridenti$)ing linear e.uations $or correlationsa#ong three or #ore varia(les& *e concludethe chapter with so#e (asic #ethods $or
developing a #athe#atical #odel that can (eused to descri(e nonlinear correlations(etween two varia(les&
8/18/2019 Chapter 10 Elementary Statistics
7/138Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 7
!ection 10-2Correlation
8/18/2019 Chapter 10 Elementary Statistics
8/138Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 8
e) Concept
In part 1 o$ this section introduces the linear
correlation coe$$icient r ' which is a nu#erical#easure o$ the strength o$ the relationship
(etween two varia(les representing.uantitative data&
sing paired sa#ple data so#eti#es called
(ivariate data' we $ind the value o$ r usuall)using technolog)' then we use that value toconclude that there is or is not a linearcorrelation (etween the two varia(les&
8/18/2019 Chapter 10 Elementary Statistics
9/138Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 9
e) Concept
In this section we consider onl) linearrelationships' which #eans that whengraphed' the points appro,i#ate a straight-
line pattern&In Part 2' we discuss #ethods o$ h)pothesistesting $or correlation&
8/18/2019 Chapter 10 Elementary Statistics
10/138Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 10
Part 1 asic Concepts o$ Correlation
8/18/2019 Chapter 10 Elementary Statistics
11/138Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 11
6e$inition
7 correlation e,ists (etween twovaria(les when the values o$ one
are so#ehow associated with thevalues o$ the other in so#e wa)&
8/18/2019 Chapter 10 Elementary Statistics
12/138Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 12
6e$inition
8he linear correlation coe$$icient r #easures the strength o$ the linearrelationship (etween the paired
.uantitative x- and y-values in a sa#ple&
8/18/2019 Chapter 10 Elementary Statistics
13/138Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 13
9,ploring the 6ata
*e can o$ten see a relationship (etween twovaria(les () constructing a scatterplot&
igure 10-2 $ollowing shows scatterplots withdi$$erent characteristics&
8/18/2019 Chapter 10 Elementary Statistics
14/138Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 14
!catterplots o$ Paired 6ata
igure 10-2
8/18/2019 Chapter 10 Elementary Statistics
15/138Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 15
!catterplots o$ Paired 6ata
igure 10-2
8/18/2019 Chapter 10 Elementary Statistics
16/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 16
!catterplots o$ Paired 6ata
igure 10-2
8/18/2019 Chapter 10 Elementary Statistics
17/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 17
Re.uire#ents
1& 8he sa#ple o$ paired x, y data is a si#plerando# sa#ple o$ .uantitative data&
2& Visual e,a#ination o$ the scatterplot #ust
con$ir# that the points appro,i#ate a straight-line pattern&
3& 8he outliers #ust (e re#oved i$ the) are%nown to (e errors& 8he e$$ects o$ an) other
outliers should (e considered () calculating r with and without the outliers included&
8/18/2019 Chapter 10 Elementary Statistics
18/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 18
:otation $or the;inear Correlation Coe$$icient
n < nu#(er o$ pairs o$ sa#ple data
Σ denotes the addition o$ the ite#sindicated&
Σ x denotes the su# o$ all x -values&
Σ x 2 indicates that each x -value should (es.uared and then those s.uares added&
Σ x 2 indicates that the x -values should (eadded and then the total s.uared&
8/18/2019 Chapter 10 Elementary Statistics
19/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 19
:otation $or the;inear Correlation Coe$$icient
Σ xy indicates that each x -value should (e $irst#ultiplied () its corresponding y-value&7$ter o(taining all such products' $indtheir su#&
r < linear correlation coe$$icient $or sa#pledata&
ρ < linear correlation coe$$icient $orpopulation data&
8/18/2019 Chapter 10 Elementary Statistics
20/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 20
or#ula 10-1
nΣ
xy – (Σ x )(Σ y)
n(Σ
x 2) – (Σ
x )2 n(Σ
y2) – (Σ
y)2r '
8he linear correlation coe$$icient r #easures thestrength o$ a linear relationship (etween thepaired values in a sa#ple&
Co#puter so$tware or calculators can co#pute r
or#ula
8/18/2019 Chapter 10 Elementary Statistics
21/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 21
Interpreting r
sing 8a(le 7- I$ the a(solute value o$ theco#puted value o$ r ' denoted =r =' e,ceeds thevalue in 8a(le 7-' conclude that there is a linearcorrelation& >therwise' there is not su$$icient
evidence to support the conclusion o$ a linearcorrelation&
sing !o$tware I$ the co#puted P -value is less
than or e.ual to the signi$icance level' concludethat there is a linear correlation& >therwise' thereis not su$$icient evidence to support theconclusion o$ a linear correlation&
8/18/2019 Chapter 10 Elementary Statistics
22/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 22
Caution
now that the #ethods o$ this sectionappl) to a linear correlation& I$ )ouconclude that there does not appear to
(e linear correlation' %now that it ispossi(le that there #ight (e so#e otherassociation that is not linear&
8/18/2019 Chapter 10 Elementary Statistics
23/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 23
Round to three deci#al placesso that it can (e co#pared to
critical values in 8a(le 7-&
se calculator or co#puter i$
possi(le&
Rounding the ;inearCorrelation Coe$$icient r
8/18/2019 Chapter 10 Elementary Statistics
24/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 24
Properties o$ the;inear Correlation Coe$$icient r
1& ?1 ≤ r ≤ 1
2& i$ all values o$ either varia(le are converted to adi$$erent scale' the value o$ r does not change&
3& 8he value o$ r is not a$$ected () the choice o$ x and y& Interchange all x- and y-values and thevalue o$ r will not change&
4& r #easures strength o$ a linear relationship&5& r is ver) sensitive to outliers' the) can
dra#aticall) a$$ect its value&
8/18/2019 Chapter 10 Elementary Statistics
25/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 25
9,a#ple
8he paired pi//a@su(wa) $are costs $ro#
8a(le 10-1 are shown here in 8a(le 10-2& seco#puter so$tware with these paired sa#plevalues to $ind the value o$ the linear
correlation coe$$icient r $or the paired
sa#ple data&
Re.uire#ents are satis$ied si#ple rando#
sa#ple o$ .uantitative dataA Minita(scatterplot appro,i#ates a straight lineAscatterplot shows no outliers - see ne,t slide
8/18/2019 Chapter 10 Elementary Statistics
26/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 26
9,a#ple
sing so$tware or a calculator' r is
auto#aticall) calculated
8/18/2019 Chapter 10 Elementary Statistics
27/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 27
Interpreting the ;inearCorrelation Coe$$icient r
*e can (ase our interpretation andconclusion a(out correlation on a P -valueo(tained $ro# co#puter so$tware or a criticalvalue $ro# 8a(le 7-&
8/18/2019 Chapter 10 Elementary Statistics
28/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 28
Interpreting the ;inearCorrelation Coe$$icient r
sing Co#puter !o$tware to Interpret r
I$ the co#puted P -value is less than or e.ualto the signi$icance level' conclude that thereis a linear correlation&>therwise' there is not su$$icient evidence to
support the conclusion o$ a linear correlation&
8/18/2019 Chapter 10 Elementary Statistics
29/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 29
Interpreting the ;inearCorrelation Coe$$icient r
sing 8a(le 7- to Interpret r
I$ =r = e,ceeds the value in 8a(le 7-' concludethat there is a linear correlation&>therwise' there is not su$$icient evidence tosupport the conclusion o$ a linear correlation&
8/18/2019 Chapter 10 Elementary Statistics
30/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 30
Interpreting the ;inearCorrelation Coe$$icient r
Critical Values $ro# 8a(le 7- and the
Co#puted Value o$ r
8/18/2019 Chapter 10 Elementary Statistics
31/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 31
sing a 0&05 signi$icance level' interpret the
value o$ r < 0&11B $ound using the 2 pairs o$weights o$ discarded paper and glass listedin 6ata !et 22 in 7ppendi, & *hen the
paired data are used with co#puterso$tware' the P -value is $ound to (e 0&34& Isthere su$$icient evidence to support a clai#o$ a linear correlation (etween the weights
o$ discarded paper and glass
9,a#ple
8/18/2019 Chapter 10 Elementary Statistics
32/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 32
Re.uire#ents are satis$ied si#ple rando#sa#ple o$ .uantitative dataA scatterplotappro,i#ates a straight lineA no outliers
9,a#ple
sing !o$tware to Interpret r
8he P -value o(tained $ro# so$tware is 0&34&ecause the P -value is not less than or
e.ual to 0&05' we conclude that there is notsu$$icient evidence to support a clai# o$ alinear correlation (etween weights o$discarded paper and glass&
8/18/2019 Chapter 10 Elementary Statistics
33/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 33
9,a#ple
sing 8a(le 7- to Interpret r
I$ we re$er to 8a(le 7- with n < 2 pairs o$sa#ple data' we o(tain the critical value o$0&254 appro,i#atel) $or α < 0&05& ecause =0&11B= does not e,ceed the value o$ 0&254$ro# 8a(le 7-' we conclude that there is notsu$$icient evidence to support a clai# o$ a
linear correlation (etween weights o$discarded paper and glass&
8/18/2019 Chapter 10 Elementary Statistics
34/138
8/18/2019 Chapter 10 Elementary Statistics
35/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 35
sing the pi//a su(wa) $are costs in 8a(le10-2' we have $ound that the linear
correlation coe$$icient is r < 0&"DD& *hatproportion o$ the variation in the su(wa)$are can (e e,plained () the variation in the
costs o$ a slice o$ pi//a*ith r < 0&"DD' we get r 2 < 0&"B&
*e conclude that 0&"B or a(out "DE o$ the
variation in the cost o$ a su(wa) $ares can (ee,plained () the linear relationship (etween thecosts o$ pi//a and su(wa) $ares& 8his i#plies thata(out 2E o$ the variation in costs o$ su(wa) $ares
cannot (e e,plained () the costs o$ pi//a&
9,a#ple
8/18/2019 Chapter 10 Elementary Statistics
36/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 36
Co##on 9rrorsInvolving Correlation
1& Causation It is wrong to conclude that
correlation i#plies causalit)&
2& 7verages 7verages suppress individualvariation and #a) in$late the correlation coe$$icient&
3& ;inearit) 8here #a) (e so#e relationship
(etween x and y even when there is no linearcorrelation&
8/18/2019 Chapter 10 Elementary Statistics
37/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 37
Caution
now that correlation does not
i#pl) causalit)&
8/18/2019 Chapter 10 Elementary Statistics
38/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 38
Part 2 or#al F)pothesis 8est
8/18/2019 Chapter 10 Elementary Statistics
39/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 39
or#al F)pothesis 8est
*e wish to deter#ine whether thereis a signi$icant linear correlation(etween two varia(les&
F th i 8 t $ C l ti
8/18/2019 Chapter 10 Elementary Statistics
40/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 40
F)pothesis 8est $or Correlation:otation
n < nu#(er o$ pairs o$ sa#ple data
r < linear correlation coe$$icient $or a sample o$ paired data
ρ < linear correlation coe$$icient $or a population o$ paired data
F th i 8 t $ C l ti
8/18/2019 Chapter 10 Elementary Statistics
41/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 41
F)pothesis 8est $or CorrelationRe.uire#ents
1& 8he sa#ple o$ paired x, y data is a si#plerando# sa#ple o$ .uantitative data&
2& Visual e,a#ination o$ the scatterplot #ust
con$ir# that the points appro,i#ate astraight-line pattern&
3& 8he outliers #ust (e re#oved i$ the) are
%nown to (e errors& 8he e$$ects o$ an)other outliers should (e considered ()calculating r with and without the outliersincluded&
8/18/2019 Chapter 10 Elementary Statistics
42/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 42
F)pothesis 8est $or CorrelationF)potheses
H 0 ρ
8/18/2019 Chapter 10 Elementary Statistics
43/138
9,a#ple
8/18/2019 Chapter 10 Elementary Statistics
44/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 44
9,a#ple
se the paired pi//a su(wa) $are data in 8a(le
10-2 to test the clai# that there is a linearcorrelation (etween the costs o$ a slice o$pi//a and the su(wa) $ares& se a 0&05signi$icance level&
Re.uire#ents are satis$ied as in the earliere,a#ple&
H 0 ρ
8/18/2019 Chapter 10 Elementary Statistics
45/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 45
9,a#ple
8he test statistic is r < 0&"DD $ro# an earlier
9,a#ple& 8he critical value o$ r < 0&D11 is$ound in 8a(le 7- with n < and α < 0&05&ecause =0&"DD= G 0&D11' we re+ect H 0 r < 0&
Re+ecting no linear correlationJ indicates
that there is a linear correlation&
*e conclude that there is su$$icient evidenceto support the clai# o$ a linear correlation(etween costs o$ a slice o$ pi//a and su(wa)$ares&
8/18/2019 Chapter 10 Elementary Statistics
46/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 46
F)pothesis 8est $or CorrelationP -Value $ro# a t 8est
H 0 ρ
8/18/2019 Chapter 10 Elementary Statistics
47/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 47
F)pothesis 8est $or CorrelationConclusion
I$ the P -value is less than or e.ual to thesigni$icance level' re+ect H 0 and conclude that thereis su$$icient evidence to support the clai# o$ alinear correlation&
I$ the P -value is greater than the signi$icancelevel' $ail to re+ect H 0 and conclude that there
is not su$$icient evidence to support the clai#
o$ a linear correlation&
P -value se co#puter so$tware or use 8a(le7-3 with n ? 2 degrees o$ $reedo# to $ind theP -value corresponding to the test statistic t &
9,a#ple
8/18/2019 Chapter 10 Elementary Statistics
48/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 48
9,a#ple
se the paired pi//a su(wa) $are data in 8a(le
10-2 and use the P -value #ethod to test theclai# that there is a linear correlation (etweenthe costs o$ a slice o$ pi//a and the su(wa)$ares& se a 0&05 signi$icance level&
Re.uire#ents are satis$ied as in the earliere,a#ple&
H 0 ρ
8/18/2019 Chapter 10 Elementary Statistics
49/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 49
9,a#ple
8he linear correlation coe$$icient is r < 0&"DD
$ro# an earlier 9,a#ple and n < si, pairso$ data' so the test statistic is
*ith d$ < 4' 8a(le 7- )ields a P -value that isless than 0&01&
Co#puter so$tware generates a test statistic o$t < 12&"2 and P -value o$ 0&00022&
t =r
1− r 2
n − 2
=0.988
1− 0.9882
6 − 2
= 12.793
9,a#ple
8/18/2019 Chapter 10 Elementary Statistics
50/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 50
9,a#ple
sing either #ethod' the P -value is less
than the signi$icance level o$ 0&05 so were+ect H 0 ρ < 0&
*e conclude that there is su$$icient evidenceto support the clai# o$ a linear correlation(etween costs o$ a slice o$ pi//a and su(wa)$ares&
> 8 il d 8 t
8/18/2019 Chapter 10 Elementary Statistics
51/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 51
>ne-8ailed 8ests
>ne-tailed tests can occur with a clai# o$ a
positive linear correlation or a clai# o$ a negativelinear correlation& In such cases' the h)potheseswill (e as shown here&
or these one-tailed tests' the P -value #ethodcan (e used as in earlier chapters&
R
8/18/2019 Chapter 10 Elementary Statistics
52/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 52
Recap
In this section' we have discussed Correlation&
8he linear correlation coe$$icient r &
Re.uire#ents' notation and $or#ula $or r &
Interpreting r &
or#al h)pothesis testing&
8/18/2019 Chapter 10 Elementary Statistics
53/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 53
!ection 10-3
Regression
C t
8/18/2019 Chapter 10 Elementary Statistics
54/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 54
e) Concept
In part 1 o$ this section we $ind the e.uation o$the straight line that (est $its the paired sa#pledata& 8hat e.uation alge(raicall) descri(es therelationship (etween two varia(les&
8he (est-$itting straight line is called aregression line and its e.uation is called theregression e.uation&
In part 2' we discuss #arginal change'in$luential points' and residual plots as tools$or anal)/ing correlation and regressionresults&
8/18/2019 Chapter 10 Elementary Statistics
55/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 55
Part 1 asic Concepts o$ Regression
R i
8/18/2019 Chapter 10 Elementary Statistics
56/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 56
Regression
8he t)pical e.uation o$ a straight line
y < mx K b is e,pressed in the $or# y < b0 K b1 x ' where b0 is the y-intercept and b1
is the slope&
L
8he regression e.uation e,presses arelationship (etween x called the
e,planator) varia(le' predictor varia(le or
independent varia(le' and y called the
response varia(le or dependent varia(le&
L
6 $i iti
8/18/2019 Chapter 10 Elementary Statistics
57/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 57
6e$initions
Regression 9.uationiven a collection o$ paired data' the regressione.uation
Regression ;ine
8he graph o$ the regression e.uation is calledthe regression line or line o$ (est $it' or least s.uares line&
y < b0 K b1 x L
alge(raicall) descri(es the relationship (etween the two varia(les&
: t ti $
8/18/2019 Chapter 10 Elementary Statistics
58/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 58
:otation $orRegression 9.uation
y-intercept o$regression e.uation
!lope o$ regressione.uation
9.uation o$ theregression line
PopulationPara#eter
!a#ple!tatistic
β0 b0
β1 b1
y < β 0 K β1 x y < b0 K b1 x L
R i t
8/18/2019 Chapter 10 Elementary Statistics
59/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 59
Re.uire#ents
1& 8he sa#ple o$ paired x , y data is arando# sa#ple o$ .uantitative data&
2& Visual e,a#ination o$ the scatterplot
shows that the points appro,i#ate astraight-line pattern&
3& 7n) outliers #ust (e re#oved i$ the) are%nown to (e errors& Consider the e$$ectso$ an) outliers that are not %nown errors&
or#ulas $or b and b
8/18/2019 Chapter 10 Elementary Statistics
60/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 60
or#ulas $or b0 and b1
or#ula 10-3 slope
y-interceptor#ula 10-4
calculators or co#puters canco#pute these values
b0 = y − b1 x
b1
= r s
y
s x
!pecial Propert)
8/18/2019 Chapter 10 Elementary Statistics
61/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 61
8he regression line $its the
sa#ple points (est&
!pecial Propert)
Rounding the y intercept b
8/18/2019 Chapter 10 Elementary Statistics
62/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 62
Rounding the y-intercept b0 and the !lope b1
Round to three signi$icant digits&
I$ )ou use the $or#ulas 10-3 and 10-4'do not round inter#ediate values&
8/18/2019 Chapter 10 Elementary Statistics
63/138
9,a#ple
8/18/2019 Chapter 10 Elementary Statistics
64/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 64
9,a#ple
Re.uire#ents are satis$ied si#ple rando#
sa#pleA scatterplot appro,i#ates a straightlineA no outliers
Fere are results $ro# $our di$$erent technologiestechnologies
9,a#ple
8/18/2019 Chapter 10 Elementary Statistics
65/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 65
9,a#ple
7ll o$ these technologies show that the
regression e.uation can (e e,pressed asy < 0&034 K0&"45 x ' where y is the predictedcost o$ a su(wa) $are and x is the cost o$ aslice o$ pi//a&
*e should %now that the regression e.uation isan esti#ate o$ the true regression e.uation&8his esti#ate is (ased on one particular set o$sa#ple data' (ut another sa#ple drawn $ro#
the sa#e population would pro(a(l) lead to aslightl) di$$erent e.uation&
L L
9,a#ple
8/18/2019 Chapter 10 Elementary Statistics
66/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 66
9,a#ple
raph the regression e.uation
$ro# the preceding 9,a#ple on thescatterplot o$ the pi//a@su(wa) $are data ande,a#ine the graph to su(+ectivel) deter#inehow well the regression line $its the data&
ˆ 0.0346 0.945= + y x
>n the ne,t slide is the Minita( displa) o$ the
scatterplot with the graph o$ the regression lineincluded& *e can see that the regression line$its the data .uite well&
9,a#ple
8/18/2019 Chapter 10 Elementary Statistics
67/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 67
9,a#ple
sing the Regression 9.uation
8/18/2019 Chapter 10 Elementary Statistics
68/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 68
1& se the regression e.uation $or predictionsonl) i$ the graph o$ the regression line on thescatterplot con$ir#s that the regression line
$its the points reasona(l) well&
sing the Regression 9.uation$or Predictions
2& se the regression e.uation $or predictionsonl) i$ the linear correlation coe$$icient r indicates that there is a linear correlation
(etween the two varia(les as descri(ed in!ection 10-2&
sing the Regression 9.uation
8/18/2019 Chapter 10 Elementary Statistics
69/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 69
3& se the regression line $or predictions onl) i$the data do not go #uch (e)ond the scope o$the availa(le sa#ple data& Predicting too $ar(e)ond the scope o$ the availa(le sa#pledata is called extrapolation' and it couldresult in (ad predictions&
sing the Regression 9.uation$or Predictions
4& I$ the regression e.uation does not appear to(e use$ul $or #a%ing predictions' the (estpredicted value o$ a varia(le is its pointesti#ate' which is its sa#ple #ean&
!trateg) $or Predicting Values o$ Y
8/18/2019 Chapter 10 Elementary Statistics
70/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 70
!trateg) $or Predicting Values o$ Y
sing the Regression 9.uation
8/18/2019 Chapter 10 Elementary Statistics
71/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 71
I$ the regression e.uation is not a good#odel' the (est predicted value o$ y is si#pl)y ' the #ean o$ the y values&
Re#e#(er' this strateg) applies to linearpatterns o$ points in a scatterplot&
I$ the scatterplot shows a pattern that is not a
straight-line pattern' other #ethods appl)' asdescri(ed in !ection 10-&
sing the Regression 9.uation$or Predictions
L
8/18/2019 Chapter 10 Elementary Statistics
72/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 72
Part 2 e)ond the asics o$ Regression
6e$initions
8/18/2019 Chapter 10 Elementary Statistics
73/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 73
6e$initions
In wor%ing with two varia(les related ()a regression e.uation' the #arginalchange in a varia(le is the a#ount that
it changes when the other varia(lechanges () e,actl) one unit& 8he slopeb1 in the regression e.uation representsthe #arginal change in y that occurs
when x changes () one unit&
6e$initions
8/18/2019 Chapter 10 Elementary Statistics
74/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 74
6e$initions
In a scatterplot' an outlier is a pointl)ing $ar awa) $ro# the other datapoints&
Paired sa#ple data #a) include one or#ore in$luential points' which are
points that strongl) a$$ect the graph o$the regression line&
9,a#ple
8/18/2019 Chapter 10 Elementary Statistics
75/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 75
9,a#ple
Consider the pi//a su(wa) $are data $ro# the
Chapter Pro(le#& 8he scatterplot located tothe le$t on the ne,t slide shows theregression line& I$ we include this additionalpair o$ data x < 2&00'y < ?20&00 pi//a is still
N2&00 per slice' (ut the su(wa) $are is N?20&00which #eans that people are paid N20 to ridethe su(wa)' this additional point would (e anin$luential point (ecause the graph o$ the
regression line would change considera(l)'as shown () the regression line located tothe right&
9,a#ple
8/18/2019 Chapter 10 Elementary Statistics
76/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 76
9,a#ple
9,a#ple
8/18/2019 Chapter 10 Elementary Statistics
77/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 77
9,a#ple
Co#pare the two graphs and )ou will see
clearl) that the addition o$ that one pair o$values has a ver) dra#atic e$$ect on theregression line' so that additional point is anin$luential point& 8he additional point is also
an outlier (ecause it is $ar $ro# the otherpoints&
6e$inition
8/18/2019 Chapter 10 Elementary Statistics
78/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 78
or a pair o$ sa#ple x and y values' theresidual is the di$$erence (etween theobserved sa#ple value o$ y and the y -value that is predicted () using theregression e.uation& 8hat is'
6e$inition
residual < o(served y ? predicted y = y – y L
Residuals
8/18/2019 Chapter 10 Elementary Statistics
79/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 79
Residuals
8/18/2019 Chapter 10 Elementary Statistics
80/138
6e$initions
8/18/2019 Chapter 10 Elementary Statistics
81/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 81
7 residual plot is a scatterplot o$ the x ' y values a$ter each o$ they -coordinate values has (een replaced
() the residual value y ? y where y denotes the predicted value o$ y & 8hatis' a residual plot is a graph o$ thepoints x ' y ? y &
6e$initions
L L
L
Residual Plot 7nal)sis
8/18/2019 Chapter 10 Elementary Statistics
82/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 82
Residual Plot 7nal)sis
*hen anal)/ing a residual plot' loo% $or apattern in the wa) the points are con$igured'and use these criteria
8he residual plot should not have an o(viouspattern that is not a straight-line pattern&
8he residual plot should not (eco#e thic%er
or thinner when viewed $ro# le$t to right&
Residuals Plot - Pi//a@!u(wa)
8/18/2019 Chapter 10 Elementary Statistics
83/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 83
Residuals Plot Pi//a@!u(wa)
Residual Plots
8/18/2019 Chapter 10 Elementary Statistics
84/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 84
Residual Plots
Residual Plots
8/18/2019 Chapter 10 Elementary Statistics
85/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 85
Residual Plots
Residual Plots
8/18/2019 Chapter 10 Elementary Statistics
86/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 86
Residual Plots
Co#plete Regression 7nal)sis
8/18/2019 Chapter 10 Elementary Statistics
87/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 87
Co#plete Regression 7nal)sis
1& Construct a scatterplot and veri$) that thepattern o$ the points is appro,i#atel) astraight-line pattern without outliers& I$there are outliers' consider their e$$ects ()
co#paring results that include the outliersto results that e,clude the outliers&
2& Construct a residual plot and veri$) that
there is no pattern other than a straight-line pattern and also veri$) that theresidual plot does not (eco#e thic%er orthinner&
Co#plete Regression 7nal)sis
8/18/2019 Chapter 10 Elementary Statistics
88/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 88
Co#plete Regression 7nal)sis
3& se a histogra# and@or nor#al .uantileplot to con$ir# that the values o$ theresiduals have a distri(ution that isappro,i#atel) nor#al&
4& Consider an) e$$ects o$ a pattern over ti#e&
Recap
8/18/2019 Chapter 10 Elementary Statistics
89/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 89
Recap
In this section we have discussed
8he (asic concepts o$ regression&
Rounding rules&
sing the regression e.uation $orpredictions&
Interpreting the regression e.uation&
>utliers Residuals and least-s.uares&
Residual plots&
8/18/2019 Chapter 10 Elementary Statistics
90/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 90
!ection 10-4
Variation and PredictionIntervals
e) Concept
8/18/2019 Chapter 10 Elementary Statistics
91/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 91
e) Concept
In this section we present a #ethod $orconstructing a prediction interval' which is an
interval esti#ate o$ a predicted value o$ y&
8/18/2019 Chapter 10 Elementary Statistics
92/138
6e$inition
8/18/2019 Chapter 10 Elementary Statistics
93/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 93
8he total deviation o$ x , y is thevertical distance y – y' which is thedistance (etween the point x , y andthe hori/ontal line passing through thesa#ple #ean y&
6e$inition
8/18/2019 Chapter 10 Elementary Statistics
94/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 94
8he e,plained deviation is the verticaldistance y ? y' which is the distance
(etween the predicted y-value and thehori/ontal line passing through thesa#ple #ean y.
L
6e$inition
8/18/2019 Chapter 10 Elementary Statistics
95/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 95
8he une,plained deviation is thevertical distance y ? y' which is thevertical distance (etween the point
x , y and the regression line& 8hedistance y ? y is also called a residual' as de$ined in !ection 10-3&
L
L
ne,plained' 9,plained' and 8otal 6eviation
8/18/2019 Chapter 10 Elementary Statistics
96/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 96
igure 10-B
p ' p '
Relationships
8/18/2019 Chapter 10 Elementary Statistics
97/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 97
total deviation < e,plained deviation K une,plained deviation
y - y < y - y K ( y - yL L
total variation < e,plained variation K une,plained variation
Σ
y - y 2
<Σ
y - y2
KΣ
( y - y)2L L
or#ula 10-5
p
6e$inition
8/18/2019 Chapter 10 Elementary Statistics
98/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 98
r 2 <e,plained variation&
total variation
8he value o$ r 2
is the proportion o$ thevariation in y that is e,plained () the linearrelationship (etween x and y&
Coe$$icient o$ deter#inationis the a#ount o$ the variation in y thatis e,plained () the regression line&
6e$inition
8/18/2019 Chapter 10 Elementary Statistics
99/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 99
7 prediction interval' is an interval
esti#ate o$ a predicted value o$ y &
6e$inition
8/18/2019 Chapter 10 Elementary Statistics
100/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 100
8he standard error o$ esti#ate' denoted
() se is a #easure o$ the di$$erences or
distances (etween the o(served
sa#ple y-values and the predicted
values y that are o(tained using the
regression e.uation&
(
8/18/2019 Chapter 10 Elementary Statistics
101/138
9,a#ple
8/18/2019 Chapter 10 Elementary Statistics
102/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 102
se or#ula 10- to $ind the standard error o$esti#ate se $or the paired pi//a@su(wa) $are
data listed in 8a(le 10-1in the Chapter Pro(le#&n = 6
Σ
y2 = 9.2175
Σ y = 6.35
Σ
xy = 9.4575
b0 = 0.034560171
b1 = 0.94502138
se <n - 2
Σ
y
2
- b0 Σ
y - b1 Σ
xy
se <
6 – 2
9.2175 – (0.034560171)(6.35) – (0.94502138)(9.4575)
se < 0&122"DB00 < 0&123
Prediction Interval $or an
8/18/2019 Chapter 10 Elementary Statistics
103/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 103
y - E O y O y K E L L
Individual y
where
E < tα
2 se n(Σ x 2) ? (Σ x )
2
n( x 0 ? x )2
1 K K1n
x 0 represents the given value o$ x
tα
2 has n ? 2 degrees o$ $reedo#
9,a#ple
8/18/2019 Chapter 10 Elementary Statistics
104/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 104
E < tα
2 seK
n(Σ x 2) – (Σ x )2
n( x 0 ? x )2
1 + 1 n
E = (2.776)(0.12298700)
6(9.77) – (6.50)2
6(2.25 – 1.0833333)2
1 + 1 6
E = (2.776)(0.12298700)(1.2905606) = 0.441
or the paired pi//a@su(wa) $are costs $ro# theChapter Pro(le#' we have $ound that $or a pi//a
cost o$ N2&25' the (est predicted cost o$ a su(wa)$are is N2&1& Construct a "5E prediction interval$or the cost o$ a su(wa) $are' given that a slice o$pi//a costs N2&25 so that , < 2&25&
K
9,a#ple
8/18/2019 Chapter 10 Elementary Statistics
105/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 105
Construct the con$idence interval&
y – E < y < y + E
2.16 – 0.441
8/18/2019 Chapter 10 Elementary Statistics
106/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 106
In this section we have discussed
9,plained and une,plained variation&
Coe$$icient o$ deter#ination&
!tandard error esti#ate&
Prediction intervals&
8/18/2019 Chapter 10 Elementary Statistics
107/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 107
!ection 10-5Multiple Regression
e) Concept
8/18/2019 Chapter 10 Elementary Statistics
108/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 108
8his section presents a #ethod $or anal)/ing alinear relationship involving #ore than two varia(les&
*e $ocus on three %e) ele#ents1& 8he #ultiple regression e.uation&
2& 8he values o$ the ad+usted R2.
3& 8he P -value&
8/18/2019 Chapter 10 Elementary Statistics
109/138
6e$inition
8/18/2019 Chapter 10 Elementary Statistics
110/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 110
7 #ultiple regression e.uation e,presses alinear relationship (etween a response varia(le
y and two or #ore predictor varia(les x 1, x 2, x 3 .
. . , x k
8he general $or# o$ the #ultiple regressione.uation o(tained $ro# sa#ple data is
y = b0 + b1 x 1 + b2 x 2 + . . . + bk x k .L
:otation
8/18/2019 Chapter 10 Elementary Statistics
111/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 111
y = b0 + b1 x 1+ b2 x 2+ b3 x 3 +. . .+ bk x k eneral $or# o$ the #ultiple regression e.uation
n < sa#ple si/e
k < nu#(er o$ predictor varia(les
y < predicted value o$ y
x 1, x 2, x 3 . . . , x k are the predictorvaria(les
L
L
:otation - cont
8/18/2019 Chapter 10 Elementary Statistics
112/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 112
ß 0, ß 1, ß 2, . . . , ß k are the para#eters $or the#ultiple regression e.uation
y = ß 0 + ß 1 x 1+ ß 2 x 2+…+ ß k x k
b0, b1, b2, . . . , bk are the sample estimates
o$ the para#eters ß 0, ß 1, ß 2, . . . , ß k
8/18/2019 Chapter 10 Elementary Statistics
113/138
9,a#ple
8/18/2019 Chapter 10 Elementary Statistics
114/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 114
8a(le 10- includes a rando#
sa#ple o$ heights o$ #others'$athers' and their daughters(ased on data $ro# the :ationalFealth and :utrition
9,a#ination& ind the #ultipleregression e.uation in which theresponse ) varia(le is theheight o$ a daughter and the
predictor , varia(les are theheight o$ the #other and heighto$ the $ather&
9,a#ple
8/18/2019 Chapter 10 Elementary Statistics
115/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 115
8he Minita( results are shown here
9,a#ple
8/18/2019 Chapter 10 Elementary Statistics
116/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 116
ro# the displa)' we see that the #ultiple
regression e.uation isFeight < B&5 K B&0BMother K 0&14ather
sing our notation presented earlier in this
section' we could write this e.uation as
y < B&5 K 0&B0B x 1 K 0&14 x 2
where ) is the predicted height o$ a daughter' x 1 is the height o$ the #other' and x 2 is the
height o$ the $ather&
L
L
6e$inition
8/18/2019 Chapter 10 Elementary Statistics
117/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 117
8he #ultiple coe$$icient o$ deter#ination R2
is a #easure o$ how well the #ultipleregression e.uation $its the sa#ple data&
8he ad+usted coe$$icient o$ deter#ination is the #ultiple coe$$icient o$ deter#ination R2
#odi$ied to account $or the nu#(er o$
varia(les and the sa#ple si/e&
7d+usted Coe$$icient o$6 t i ti
8/18/2019 Chapter 10 Elementary Statistics
118/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 118
6eter#ination
7d+usted R2 = 1 – (n – 1)
[n – (k + 1)](1– R
2)
or#ula 10-B
where n < sa#ple si/e
k < nu#(er o$ predictor x varia(les
P -Value
8/18/2019 Chapter 10 Elementary Statistics
119/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 119
8he P -value is a #easure o$ theoverall signi$icance o$ the #ultipleregression e.uation& ;i%e thead+usted R 2' this P -value is a good
#easure o$ how well the e.uation $itsthe sa#ple data&
P -Value
8/18/2019 Chapter 10 Elementary Statistics
120/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 120
8he displa)ed Minita( P -value o$ 0&000
rounded to three deci#al places is s#all'indicating that the #ultiple regressione.uation has good overall signi$icance and isusa(le $or predictions& 8hat is' it #a%es sense
to predict heights o$ daughters (ased onheights o$ #others and $athers& 8he value o$0&000 results $ro# a test o$ the null h)pothesisthat β 1 < β 2 < 0& Re+ection o$ β 1 < β2 < 0 i#plies
that at least one o$ β1 and β 2 is not 0' indicatingthat this regression e.uation is e$$ective inpredicting heights o$ daughters&
inding the est MultipleR i 9 ti
8/18/2019 Chapter 10 Elementary Statistics
121/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 121
Regression 9.uation
1& se co##on sense and practical considerations toinclude or e,clude varia(les&
2& Consider the P -value& !elect an e.uation havingoverall signi$icance' as deter#ined () the P -value$ound in the co#puter displa)&
inding the est MultipleR i 9 ti
8/18/2019 Chapter 10 Elementary Statistics
122/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 122
Regression 9.uation
3& Consider e.uations with high values o$ ad+usted R2
and tr) to include onl) a $ew varia(les&
I$ an additional predictor varia(le is included' thevalue o$ ad+usted R2 does not increase () asu(stantial a#ount&
or a given nu#(er o$ predictor x varia(les'select the e.uation with the largest value o$ad+usted R2&
In weeding out predictor x varia(les that donthave #uch o$ an e$$ect on the response yvaria(le' it #ight (e help$ul to $ind the linearcorrelation coe$$icient r $or each o$ the pairedvaria(les (eing considered&
8/18/2019 Chapter 10 Elementary Statistics
123/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 123
Part 2 6u##) Varia(les and
;ogistic 9.uations
6u##) Varia(le
8/18/2019 Chapter 10 Elementary Statistics
124/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 124
Man) applications involve a dichoto#ousvaria(le which has onl) two possi(le discretevalues such as #ale@$e#ale' dead@alive' etc&&7 co##on procedure is to represent the two
possi(le discrete values () 0 and 1' where 0represents $ailureJ and 1 represents success&
7 dichoto#ous varia(le with the two values 0and 1 is called a du##) varia(le&
;ogistic Regression
8/18/2019 Chapter 10 Elementary Statistics
125/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 125
*e can use the #ethods o$ this section i$the du##) varia(le is the predictor varia(le&
I$ the du##) varia(le is the response
varia(le we need to use a #ethod %nown aslogistic regression&
7s the na#e i#plies logistic regression
involves the use o$ natural logarith#s& 8histe,t (oo% does not include detailedprocedures $or using logistic regression&
8/18/2019 Chapter 10 Elementary Statistics
126/138
8/18/2019 Chapter 10 Elementary Statistics
127/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 127
!ection 10-Modeling
e) Concept
8/18/2019 Chapter 10 Elementary Statistics
128/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 128
8his section introduces so#e (asic conceptso$ developing a #athe#atical #odel' which isa $unction that $itsJ or descri(es real-world
data&
nli%e !ection 10-3' we will not (e restrictedto a #odel that #ust (e linear&
8I-D3@D4 Plus eneric Models
8/18/2019 Chapter 10 Elementary Statistics
129/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 129
;inear y = a + bx
Quadratic y = ax 2 + bx + c
;ogarith#ic y = a + b ln x 9,ponential y = ab x
Power y = ax b
8/18/2019 Chapter 10 Elementary Statistics
130/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 130
8he slides that $ollow illustrate the graphso$ so#e co##on #odels displa)ed on a
8I-D3@D4 Plus Calculator
8/18/2019 Chapter 10 Elementary Statistics
131/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 131
8/18/2019 Chapter 10 Elementary Statistics
132/138
8/18/2019 Chapter 10 Elementary Statistics
133/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 133
8/18/2019 Chapter 10 Elementary Statistics
134/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 134
8/18/2019 Chapter 10 Elementary Statistics
135/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 135
6evelop#ent o$ a oodMathe#atical Model
8/18/2019 Chapter 10 Elementary Statistics
136/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 136
Mathe#atical Model
;oo% $or a Pattern in the raph 9,a#inethe graph o$ the plotted points andco#pare the (asic pattern to the %nowngeneric graphs o$ a linear $unction&
ind and Co#pare Values o$ R 2 !elect$unctions that result in larger values o$ R 2'(ecause such larger values correspond to$unctions that (etter $it the o(served
points&8hin% se co##on sense& 6ont use a
#odel that leads to predicted values %nownto (e totall) unrealistic&
I#portant Point
8/18/2019 Chapter 10 Elementary Statistics
137/138
Copyright © 2010, 2007, 2004 Pearson Education, Inc. All Rights Resered. 10.1 - 137
p
8he (est choice o$ a #odeldepends on the set o$ data (eing
anal)/ed and re.uires an e,ercise in +udg#ent' not +ust co#putation&J
Recap
8/18/2019 Chapter 10 Elementary Statistics
138/138
In this section we have discussed
8he concept o$ #athe#atical #odeling&
raphs $ro# a 8I-D3@D4 Plus calculator& Rules $or developing a good #athe#atical#odel&