帕梅拉-欣德勒-管理研究方法（英文版·第13版）Schindler-CH-14-Accessible

14-1Business

Research

MethodsThirteenth

Edition

Pamela

S.

SchindlerChapter

14STAGE

4:

HYPOTHESIS

TESTING?2019

McGraw-Hill

Education.

All

rights

reserved.

Authorized

only

for

instructor

use

in

the

classroom.

Noreproduction

or

further

distribution

permitted

without

the

prior

written

consent

of

McGraw-Hill

Education.Learning

Objectives14-2?

2019

McGraw-Hill

Education.Understand

.

.

.The

nature

and

logic

of

hypothesis

testing.A

statistically

significant

difference

The

six-step

hypothesis

testing

procedure.The

differences

between

parametric

andnonparametric

tests

and

when

to

use

each.The

factors

that

influence

the

selection

of

anappropriate

test

of

statistical

significance.How

to

interpret

the

various

test

statistics.Research

Thought

Leader14-3?

2019

McGraw-Hill

Education.“A

fact

is

a

simple

statement

that

everyone

believes.It

is

innocent,

unless

found

guilty.

A

hypothesis

is

anovel

suggestion

that

no

one

wants

to

believe.

It

isguilty,

until

found

effective.”Edward

Teller,

theoretical

physicist,“father

of

the

hydrogen

bomb”(1908–2003)Hypothesis

Testing14-4?

2019

McGraw-Hill

Education.Induction

ReasoningDeduction

ReasoningReasoning

and

Hypotheses

(1

of

2)Inductions

are

an

inferential

leap

from

theevidence

presented.=14-5?

2019

McGraw-Hill

Education.Reasoning

and

Hypotheses

(2

of

2)Deductions

are

only

as

good

as

the

premises

onwhich

they

are

based.≠14-6?

2019

McGraw-Hill

Education.Statistical

Procedures14-7?

2019

McGraw-Hill

Education.Inferential

StatisticsDescriptive

StatisticsHypothesis

Testing

and

the

ResearchProcessJump

to

long

description14-8?

2019

McGraw-Hill

Education.The

“Ah-Ha”

MomentWhen

researchers

siftthrough

the

chaos

andfind

what

matters.14-9?

2019

McGraw-Hill

Education.Approaches

to

Hypothesis

Testing14-10?

2019

McGraw-Hill

Education.Classical

statistics

Objective

view

ofprobability

Establishedhypothesis

is

rejectedor

fails

to

be

rejected

Analysis

based

onsample

dataBayesian

statistics

Extension

of

classicalapproach

Analysis

based

onsample

data

Also

considersestablished

subjectiveprobability

estimatesSignificance

&

HypothesesStatistical

SignificancePractical

Significance14-11?

2019

McGraw-Hill

Education.Null

vs.

Alternative

HypothesesNullH0:=60

mpgH0:<60

mpgH0:>60

mpgAlternativeHA:HA:HA:≠

60

mpg>

60

mpg<

60

mpg14-12?

2019

McGraw-Hill

Education.Two-Tailed

Test

of

SignificanceJump

to

long

description14-13?

2019

McGraw-Hill

Education.One-Tailed

Test

of

SignificanceJump

to

long

description14-14?

2019

McGraw-Hill

Education.Decision

Rule14-15?

2019

McGraw-Hill

Education.Takeno

corrective

action

if

the

analysis

shows

thatone

cannot

reject

the

null

hypothesis.Statistical

DecisionsJump

to

long

description?

2019

McGraw-Hill

Education.14-16Probability

of

Making

a

Type

I

Error

(1

of

2)Jump

to

long

description?

2019

McGraw-Hill

Education.14-17Probability

of

Making

A

Type

I

Error

(2

of

2)Jump

to

long

description?

2019

McGraw-Hill

Education.14-18Critical

Values14-19?

2019

McGraw-Hill

Education.Factors

Affecting

Probability

ofCommitting

a

Error14-20?

2019

McGraw-Hill

Education.True

value

of

parameterAlpha

level

selectedOne

or

two-tailed

test

usedSample

standard

deviationSample

sizeProbability

of

Making

A

Type

II

ErrorJump

to

long

description14-21?

2019

McGraw-Hill

Education.Statistical

Testing

Procedures14-22?

2019

McGraw-Hill

Education.StagesState

null

hypothesisChoose

statistical

testSelect

level

of

significanceCompute

difference

valueObtain

critical

test

valueInterpret

the

testTests

of

Significance14-23?

2019

McGraw-Hill

Education.ParametricNonparametricAssumptions

for

Using

ParametricTestsIndependent

observationsNormal

distributionEqual

variancesInterval

or

ratio

scalesJump

to

long

description14-24?

2019

McGraw-Hill

Education.Probability

Plot

(1

of

3)14-25?

2019

McGraw-Hill

Education.Probability

Plot

(2

of

3)14-26?

2019

McGraw-Hill

Education.Probability

Plot

(3

of

3)Jump

to

long

description14-27?

2019

McGraw-Hill

Education.Advantages

of

Nonparametric

Tests

Easy

to

understand

anduseUsable

with

nominal

data

Appropriate

for

ordinaldata

Appropriate

for

non-normal

populationdistributions14-28?

2019

McGraw-Hill

Education.How

to

Select

a

TestHow

many

samples

areinvolved?If

>

2

Are

the

individual

casesindependent

or

related?

Is

the

measurement

nominal,

ordinal,

interval,or

ratio?14-29?

2019

McGraw-Hill

Education.Recommended

Statistical

Techniques14-30?

2019

McGraw-Hill

Education.MeasurementScaleOne-SampleCaseTwo-SampleTests:

RelatedSamplesTwo-Sample

Tests:IndependentSamplesk-Sample

Tests:RelatedSamplesk-Sample

Tests:IndependentSamplesNominalOrdinalInterval

andRatioBinomialx2

one-sampletestKolmogorov-Smirnov

one-sample

testRuns

testt-testZ

testMcNemarSign

testWilcoxonmatched-pairstestt-test

for

pairedsamplesFisher

exact

testx2

two-samples

testMedian

testMann-Whitney

UKolmogorov-SmirnovWald-Wolfowitzt-testZ

testCochran

QFriedmantwo-wayANOVARepeated-measuresANOVAx2

for

k

samplesMedianextensionKruskal-Wallisone-way

ANOVAOne-wayANOVAn-way

ANOVAQuestions

Answeredby

One-SampleTests

Differencebetweenobserved

and

expectedfrequencies?

Differencebetweenobserved

and

expectedproportions?

Significant

differencebetween

some

measure

ofcentral

tendency

and

thepopulation

parameter?14-31?

2019

McGraw-Hill

Education.Parametric

Tests14-32?

2019

McGraw-Hill

Education.Z-testt-testOne-Sample

t-Test

Example14-33?

2019

McGraw-Hill

Education.Null

Statistical

testSignificance

levelCalculated

valueCritical

test

valueHo:

=

50

mpgt-test.05,

n=1001.7861.66

(from

Appendix

C,Exhibit

C-2)One

Sample

Chi-Square

Test

ExampleLivingArrangementIntend

to

JoinNumberInterviewedPercent(no.interviewed/200)ExpectedFrequencies(percent

x

60)Dorm/fraternity16904527Apartment/roominghouse,

nearby13402012Apartment/roominghouse,

distant16402012Live

athome1530159Total602001006014-34?

2019

McGraw-Hill

Education.One-Sample

Chi-Square

Example14-35?

2019

McGraw-Hill

Education.Null

Statistical

testSignificance

levelCalculated

valueCritical

test

valueHo:

0

=

EOne-sample

chi-square.059.897.82(fromAppendixC,Exhibit

C-3)Two-Sample

Parametric

Tests14-36?

2019

McGraw-Hill

Education.Two-Sample

t-Test

Example

(1

of

2)14-37?

2019

McGraw-Hill

Education.A

GroupB

GroupAverage

hourlysalesX1

=$1,500X2

=

$1,300Standarddeviations1

=

225s2

=

251Two-Sample

t-Test

Example

(2

of

2)14-38?

2019

McGraw-Hill

Education.Null

Statistical

testSignificance

levelCalculated

valueCritical

test

valueHo:

A

sales

=

B

salest-test.05

(one-tailed)1.97,

d.f.

=

201.725(fromAppendixC,Exhibit

C-2)Two-Sample

Nonparametric

Tests:

Chi-Square14-39?

2019

McGraw-Hill

Education.Cell

DesignationCountExpected

ValuesOn-the-Job-Accident:YesOn-the-Job-Accident:

NoRow

TotalSmokerHeavy

SmokerModerateNonsmoker1,11,212,4168.247.752,192,26157.737.273,1133,2223518.0316.97Column

Total343266Two-Sample

Chi-Square

Example14-40?

2019

McGraw-Hill

Education.NullStatistical

testSignificance

levelCalculated

valueCritical

test

valueThere

is

no

difference

indistribution

channel

for

agecategories.Chi-square.056.86,

d.f.

=

25.99(from

Appendix

C,

Exhibit

C-3)SPSS

Cross-Tabulation

ProcedureJump

to

long

description14-41?

2019

McGraw-Hill

Education.Two-Related-Samples

Tests14-42?

2019

McGraw-Hill

Education.ParametricNonparametricSales

Data

for

Paired-Samples

t-Test14-43?

2019

McGraw-Hill

Education.SalesSalesCompanyYear

2Year

1Difference

DD2GM126932123505342711744329GE5457449662491224127744Exxon8665678944771259474944IBM6271059512319210227204Ford9614692300384614971716AT&T3611235173939881721Mobil502204811121094447881DuPont350993242726326927424Sears5379449975381914584761AmocoTotal23966207793187ΣD

=

3578110156969ΣD

=

157364693Paired-Samples

t-Test

Example14-44?

2019

McGraw-Hill

Education.NullStatistical

testSignificance

levelCalculated

valueCritical

test

valueYear

1

sales

=Year

2

salesPaired

sample

t-test.016.28,

d.f.

=

93.25(from

Appendix

C,

Exhibit

C-2)SPSS

Output

for

Paired-Samples

t-TestJump

to

long

description14-45?

2019

McGraw-Hill

Education.Related

Samples

Nonparametric

Tests:McNemar

Test

(1

of

2)14-46?

2019

McGraw-Hill

Education.BeforeAfter

FavorFavorDo

Not

FavorAfter

Do

NotFavorA

CB

DRelated

Samples

Nonparametric

Tests:McNemar

Test

(2

of

2)14-47?

2019

McGraw-Hill

Education.BeforeAfter

Do

NotFavorAfter

FavorFavorA=10B=90Do

Not

FavorC=60D=40k-Independent-Samples

Tests:

ANOVA14-48?

2019

McGraw-Hill

Education.Tests

the

null

hypothesis

that

the

means

of

three

ormore

populations

are

equal.One-way:

Uses

a

single-factor,

fixed-effects

modelto

compare

the

effects

of

a

treatment

or

factor

ona

continuous

dependent

variable.ANOVA

Example

(1

of

2)14-49?

2019

McGraw-Hill

Education.Model

SummarySourced.f.Sum

of

SquaresMean

SquareF

Valuep

ValueModel

(airline)211644.0335822.01728.3040.0001Residual

(error)5711724.550205.694Total5923368.583Means

TableCountMeanStd.

Dev.Std.

ErrorLufthansa2038.95014.0063.132Malaysia

Airlines2058.90015.0893.374Cathay

Pacific2072.90013.9023.108All

data

are

hypotheticalANOVA

Example

(2

of

2)14-50?

2019

McGraw-Hill

Education.NullStatistical

testSignificance

levelCalculated

valueA1

=

A2

=

A3ANOVA

and

F

ratio.0528.304,

d.f.

=

2,

57Critical

test

value3.16(from

Appendix

C,

Exhibit

C-9)Post

Hoc:

Scheffe’s

S

MultipleComparison

Procedure14-51?

2019

McGraw-Hill

Education.VersesDiffCrit.

Diff.p

ValueLufthansaMalaysia

Airlines19,95011.400.0002Cathay

Pacific33.95011.400.0001Malaysia

AirlinesCathay

Pacific14.00011.400.0122Multiple

Comparison

Procedures14-52?

2019

McGraw-Hill

Education.TestComplexComparisonsPairwiseComparisonsEqual

n’sOnlyUnequal

n’sEqualVariancesAssumedUnequalVariancesNot

AssumedFisher

LSDXXXBonferroniXXXTukey

HSDXXXTukey-KramerXXXGames-HowellXXXTamhane

T2XXXScheffé

SXXXXBrown-XXXXForsytheNewman-KeulsXXDuncanXXDunnet’s

T3XDunnet’s

CXANOVA

PlotsJump

to

long

description14-53?

2019

McGraw-Hill

Education.Jump

to

long

descriptionTwo-Way

ANOVA

Example14-54?

2019

McGraw-Hill

Education.Model

SummarySource

d.f.Sum

of

SquaresMean

SquareF

Valuep

ValueAirline211644.0335822.01739.1780.0001Seat

selection13182.8173182.81721.4180.0001Airline

by

seat

selection2517.033258.5171.7400.1853Residual548024.700148.606Means

Table

Effect:

Airline

by

Seat

SelectionCountMeanStd.

Dev.Std.

ErrorLufthansa

economy1035.60012.1403.839Lufthansa

business1042.30015.5504.917Malaysia

Airlines

economy1048.50012.5013.953Malaysia

Airlines

business1069.3009.1662.898Cathay

Pacific

economy1064.80013.0374.123Cathay

Pacific

businessAll

data

are

hypothetica1081.0009.6033.037lTwo-way

Analysis

of

Variance

PlotsJump

to

long

description14-55?

2019

McGraw-Hill

Education.k-Related-Samples

Tests14-56?

2019

McGraw-Hill

Education.More

than

two

levels

in

grouping

factorObservations

are

matchedData

are

interval

or

ratio?

2019

McGraw-Hill

Education.14-57Summary

Tables

for

Repeated-Measures

ANOVAModel

SummarySourced.f.Sum

of

SquaresMean

SquareF

Valuep

ValueAirline23552735.5017763.77567.1990.0001Subject

(group)5715067.650264.345Ratings1625.633625.63314.3180.0004Ratings

byair22061.7171030.85823.5920.0001Ratings

bysubj572490.65043.696Means

Tableby

AirlineCountMeanStd.

Dev.Std.

ErrorRating

1,

Lufthansa2038.95014.0063.132Rating

1,

Malaysia2058.90015.0893.374AirlinesRating

1,

Cathay

Pacific2072.90013.9023.108Rating

2,

Lufthansa2032.4008.2681.849Rating

2,

Malaysia2072.25010.5722.364AirlinesMeans

Table

Effect:

RatingsRating

2,

Cathay

Pacific

20

79.80011.2652.519CountMeanStd.

Dev.Std.

ErrorRating

16056.91719.9022.569Rating

26061.48323.2082.996All

data

are

hypothetical.Repeated

Measures

ANOVA

PlotJump

to

long

description14-58?

2019

McGraw-Hill

Education.Key

Terms

(1

of

2)14-59?

2019

McGraw-Hill

Education.a

priori

contrastsAlternative

hypothesisAnalysis

of

variance

(ANOVA)Bayesian

statisticsChi-square

testClassical

statisticsCritical

valueF

ratioInferential

statisticsK-independent-samples

testsK-related-samples

testsLevel

of

significanceMean

square

Multiple

comparison

tests(range

tests)Nonparametric

testsNormal

probability

plotNull

hypothesisObserved

significance

levelOne-sample

testsOne-tailed

testKey

Terms

(2

of

2)14-60?

2019

McGraw-Hill

Education.p

valueParametric

testsPower

of

the

testPractical

significanceRegion

of

acceptanceRegion

of

rejectionStatistical

significancet

distributionTrialst-test

Two-independent-samplestestsTwo-related-samples

testsTwo-tailed

testType

I

errorType

II

errorZ

distributionZ

testPhoto

Attributions14-61?

2019

McGraw-Hill

Education.SlideSource5?walenga/123RF5?MedicalRF.com6?CostinT/Getty

Images6?Image

Source,

all

rights

reserved.9?Caiaimage/Glow

Images11?

McGraw-Hill

Education/Mark

Dierker12?

McGraw-Hill

Education/Mark

Dierker24?

Anatolii

Babii

/

AlamySlide

Source28?

Shutterstock

/

Anton

Gepolov29?

Image

Source/Takahiro

Igarashi31?

Monalyn

Gracia/Fancy/age

fotostock

RF53?

Blend

Images

/

Alamy

Stock

Photo62?

VStock

/

Alamy

Stock

Photo63?

Janne

Tervonen

/

Alamy64?

Jukeboxhero/iStock/Getty

Images65Purestock/SuperStock13-62Chapter

14STAGE

4:

HYPOTHESIS

TESTINGDISCUSSION

OPPORTUNITIES?2019

McGraw-Hill

Education.

All

rights

reserved.

Authorized

only

for

instructor

use

in

the

classroom.

Noreproduction

or

further

distribution

permitted

without

the

prior

written

consent

of

McGraw-Hill

Education.Snapshot:

Testing

a

Hypothesis

forTroy-Bilt?

What

elements

in

TV

advertisingbuild

recall

and

consideration?

Those

who

are

market

ready

incategory

process

TV

adsdifferently

than

those

who

aren’tmarket

ready.Test

and

control

groups

used.

Test

ads

embedded

in

DIYHouse

Crashers.14-63?

2019

McGraw-Hill

Education.Snapshot:

Testing

a

Hypothesis

ofUnrealistic

Drug

use

in

MoviesTop

200

rental

movies

during

2

years.

Content

analysis

of

drinking,smoking,illicit

drugs,

prescriptions

and

OLCdrugs.Trained

coders.

Prevalence

of

use,

frequency

of

use,percentage

of

characters

using.14-64?

2019

McGraw-Hill

Education.Snapshot:

A/B

Testing

Comes

of

Age

Different

direct

mail

piecesdifferent

sent

to

samplegroups

Change

return

address

onenvelopeChange

offerChange

color

of

paper14-65?

2019

McGraw-Hill

Education.Snapshot:

A/B

Testing

Comes

of

Age

Does

web

page

designalter

click

behavior?

Are

we

being

testedwheneverwe

visit

a

webpage?Test

one

or

multiple

factors.

What

hypothesis

coulddrive

A/B

tests?14-66?

2019

McGraw-Hill

Education.Appendix

of

Image

Long

Descriptions14-67?

2019

McGraw-Hill

Education.Hypothesis

Testing

and

the

ResearchProcess

Long

Description14-68?

2019

McGraw-Hill

Education.The

flow

starts

with

investigative

questions,

then

flows

toFormulate

Preliminary

Hypotheses,

then

to

PreliminaryAnalysis

Plan

(which

includes

Refine

Hypothesis

and

DataVisualization),

then

to

Measurement

Questions.The

flow

thengoes

to

Research

Design

(which

includes

Data

DesignCollection

and

Sampling

Design,

which

lead

to

the

InstrumentDevelopment).

This

then

flows

to

Data

Collection

andPreparation,

then

possibly

to

Revise

Hypotheses,

thenHypothesis

Testing,

and

finally

to

Data

Analysis

&Interpretation.Jump

to

imageTwo-Tailed

Test

of

Significance

LongDescription14-69?

2019

McGraw-Hill

Education.The

graph

has

a

bell

curve

marked

at

Z=-1.96

and

Z=1.96.The

tails

of

the

bell

curve

outside

these

two

boundaries

eachhave

an

area

of

.025,

and

it

is

in

these

two

regions

where

H-naught

is

rejected.

In

the

region

between

these

twoboundaries,

you

do

not

reject

H-naught.Jump

to

imageOne-Tailed

Test

of

Significance

LongDescription14-70?

2019

McGraw-Hill

Education.The

graph

has

a

bell

curve

marked

at

Z=1.645.

Thearea

to

the

right

of

this

boundary

is

.05,

and

it

is

inthis

region

where

H-naught

is

rejected.

In

theregion

to

the

left

of

this

boundary,

you

do

not

rejectH-naught.Jump

to

imageStatistical

Decisions

Long

Description14-71?

2019

McGraw-Hill

Education.The

graphic

shows

a

pair

of

2

by

2

squares,

in

these

are

thefollowing

information.

1)

Correct

Decision:

Accept

H0

when

H0

is

true;

Power

of

test:

Probability

=

1–alpha;

Innocent

ofcrime,

found

not

guilty.

2)

Type

II

Error:

Accept

H0

when

Ha

is

true;

Power

of

test:

Probability

=

beta;

guilty

of

crime,unjustly

acquitted.

3)

Type

I

Error:

Accept

Ha

when

H0

istrue;

Significance

level:

Probability

=

alpha;

Innocent,unjustly

convicted.

4)

Correct

Decision:

Accept

Ha

when

Hais

true;

Power

of

test:

Probability

=

1–beta;

Guilty,

justlyconvicted.Jump

to

imageProbability

of

Making

A

Type

I

ErrorLong

Description

(1

of

2)14-72?

2019

McGraw-Hill

Education.The

graph

involves

a

two-tailed

test.

A

bell

curve

hasmarked:mu=50,

x-bar-sub-c-sub-1=46.08,

x-bar-sub-c-sub-2

=

53.92.Between

these

two

points

is

the

region

of

acceptance,

andoutside

these

two

points

are

the

regions

of

rejection.Jump

to

imageProbability

of

Making

A

Type

I

ErrorLong

Description

(2

of

2)Jump

to

image14-73?

2019

McGraw-Hill

Education.The

graph

involves

a

one-tailed

test.

A

bell

graph

hasmarked:mu=50,

x-bar-sub-c=53.29.

To

the

left

of

this

point

isthe

region

of

acceptance,

and

to

the

right

is

the

region

ofrejection,

alpha

=.05.Probability

of

Making

A

Type

II

ErrorLong

DescriptionJump

to

image14-74?

2019

McGraw-Hill

Education.The

graph

consists

of

two

overlapping

bell

curves

-

the

one

on

the

left

labeled

alpha,

and

the

one

on

the

right

labeledbeta.

On

the

alpha

graph,

the

line

that

marks

alpha

=

.05

isextended

up

to

the

beta

graph.

The

portion

of

the

graph

tothe

left

of

this

line

is

"don"t

reject

H-naught,"

and

to

the

ri"reject

H-naught."

1

minus

alpha

=

.95

is

marked

on

the

leftof

this

line,

as

well.

The

left

bell

curve

has

mu

=

50,

the

righbell

curve

hasmu=54,and

the

point

where

alpha

=

.05

is53.29.

To

the

left

of

this

point

beta

=

.36,

and

to

the

right

isminus

beta

=

.64.Assumptions

for

Using

ParametricTests

Long

DescriptionJump

to

image14-75?

2019

McGraw-Hill

Education.One

is

holding

a

computer

tablet

pointing

to

data

on

thescreen

while

the

other

one

is

writing

on

a

chart.Probability

Plot

(3

of

3)

LongDescriptionJump

to

image14-76?

2019

McGraw-Hill

Education.The

first

is

a

normal

plot

of

sample,

where

the

points

closelyfollow

a

curved

trend.

The

second

is

a

detrended

normal

plotof

sample,

where

the

points

follow

a

curve

that

is

not

marked.SPSS

Cross-Tabulation

ProcedureLong

DescriptionJump

to

image14-77?

2019

McGraw-Hill

Education.The

title

is

Income

by

Possession

of

MBA.

Across

the

top

isYes

or

No,

and

down

the

side

is

High

1

and

Low

2.

The

cellsare:

30

30

Row

Total

60;

1030

Row

Total

40;

Column

Total

40

60;

Total

100.

The

table

below

has

column

headers:

Chi-Square,

Value,

D.F.,

Significance.

Rows

are:

Pearson,

6.25,1,

.01242;

Continuity

Correction

5.25174,

1,

.02192;Likelihood

Ratio

6.43786,

1,

.01117;

Mantel-Haenszel6.18750,

1,

.01287.

Minimum

ExpectedFrequency:

16.000SPSS

Output

for

Paired-Samples

t-TestLong