Table Of ContentIntroductionto
QuantitativeDataAnalysis
intheBehavioraland
SocialSciences
Introduction to
Quantitative Data Analysis
in the Behavioral and
Social Sciences
Michael J. Albers
East Carolina University
Thiseditionfirstpublished2017
2017JohnWiley&Sons,Inc.
Allrightsreserved.Nopartofthispublicationmaybereproduced,storedinaretrievalsystem,or
transmitted,inanyformorbyanymeans,electronic,mechanical,photocopying,recordingor
otherwise,exceptaspermittedbylaw.Adviceonhowtoobtainpermissiontoreusematerialfrom
thistitleisavailableathttp://www.wiley.com/go/permissions.
TherightofMichaelJ.Alberstobeidentifiedastheauthor(s)ofthisworkhasbeenassertedin
accordancewithlaw.
RegisteredOffices
JohnWiley&Sons,Inc.,111RiverStreet,Hoboken,NJ07030,USA
EditorialOffice
111RiverStreet,Hoboken,NJ07030,USA
Fordetailsofourglobaleditorialoffices,customerservices,andmoreinformationaboutWiley
productsvisitusatwww.wiley.com.
Wileyalsopublishesitsbooksinavarietyofelectronicformatsandbyprint-on-demand.Some
contentthatappearsinstandardprintversionsofthisbookmaynotbeavailableinotherformats.
LimitofLiability/DisclaimerofWarranty
Thepublisherandtheauthorsmakenorepresentationsorwarrantieswithrespecttothe
accuracyorcompletenessofthecontentsofthisworkandspecificallydisclaimallwarranties;
includingwithoutlimitationanyimpliedwarrantiesoffitnessforaparticularpurpose.Thiswork
issoldwiththeunderstandingthatthepublisherisnotengagedinrenderingprofessional
services.Theadviceandstrategiescontainedhereinmaynotbesuitableforeverysituation.In
viewofon-goingresearch,equipmentmodifications,changesingovernmentalregulations,and
theconstantflowofinformationrelatingtotheuseofexperimentalreagents,equipment,and
devices,thereaderisurgedtoreviewandevaluatetheinformationprovidedinthepackageinsert
orinstructionsforeachchemical,pieceofequipment,reagent,ordevicefor,amongotherthings,
anychangesintheinstructionsorindicationofusageandforaddedwarningsandprecautions.
Thefactthatanorganizationorwebsiteisreferredtointhisworkasacitationand/orpotential
sourceoffurtherinformationdoesnotmeanthattheauthororthepublisherendorsesthe
informationtheorganizationorwebsitemayprovideorrecommendationsitmaymake.Further,
readersshouldbeawarethatwebsiteslistedinthisworkmayhavechangedordisappeared
betweenwhenthisworkswaswrittenandwhenitisread.Nowarrantymaybecreatedor
extendedbyanypromotionalstatementsforthiswork.Neitherthepublishernortheauthorshall
beliableforanydamagesarisingherefrom.
LibraryofCongressCataloguing-in-PublicationDataappliedfor.
Hardback:9781119290186
Coverimage:Magnilion/gettyimages
CoverdesignbyWiley
Setin10/12ptWarnockPro-RegularbyThomsonDigital,Noida,India
10 9 8 7 6 5 4 3 2 1
v
Contents
Preface ix
AbouttheCompanionWebsite xiii
1 Introduction 1
BasisofHowAllQuantitativeStatisticalBasedResearch 1
DataAnalysis,NotStatisticalAnalysis 3
QuantitativeVersusQualitativeResearch 8
WhattheBookCoversandWhatItDoesNotCover 9
BookStructure 10
References 11
PartI DataAnalysisApproaches 13
2 StatisticsTerminology 15
StatisticallyTestingaHypothesis 15
StatisticalSignificanceandp-Value 19
ConfidenceIntervals 26
EffectSize 27
StatisticalPowerofaTest 31
PracticalSignificanceVersusStatisticalSignificance 34
StatisticalIndependence 34
DegreesofFreedom 36
MeasuresofCentralTendency 37
PercentileandPercentileRank 41
CentralLimitTheorem 42
LawofLargeNumbers 44
References 48
vi Contents
3 AnalysisIssuesandPotentialPitfalls 49
EffectsofVariables 49
OutliersintheDataset 53
RelationshipsBetweenVariables 53
ASingleContradictoryExampleDoesNotInvalidateaStatistical
Relationship 60
References 62
4 GraphicallyRepresentingData 63
DataDistributions 63
BellCurves 64
SkewedCurves 68
BimodalDistributions 71
PoissonDistributions 75
BinomialDistribution 77
Histograms 79
ScatterPlots 80
BoxPlots 81
RangesofValuesandErrorBars 82
References 85
5 StatisticalTests 87
Inter-RaterReliability 87
RegressionModels 92
ParametricTests 93
NonparametricTests 95
One-TailedorTwo-TailedTests 96
TestsMustMakeSense 99
References 103
PartII DataAnalysisExamples 105
6 OverviewofDataAnalysisProcess 107
KnowHowtoAnalyzeItBeforeStartingtheStudy 107
PerformanExploratoryDataAnalysis 108
PerformtheStatisticalAnalysis 109
AnalyzetheResultsandDrawConclusions 110
WritingUptheStudy 111
References 112
Contents vii
7 AnalysisofaStudyonReadingandLightingLevels 113
LightingandReadingComprehension 113
KnowHowtheDataWillBeAnalyzedBeforeStartingthe
Study 113
PerformanExploratoryDataAnalysis 115
PerformanInferentialStatisticalAnalysis 122
Exercises 132
8 AnalysisofUsabilityofanE-CommerceSite 135
UsabilityofanE-CommerceSite 135
StudyOverview 135
KnowHowYouWillAnalyzetheDataBeforeStartingthe
Study 136
PerformanExploratoryDataAnalysis 138
PerformanInferentialStatisticalAnalysis 147
Follow-UpTests 151
PerformingFollow-UpTests 153
Exercises 157
Reference 158
9 AnalysisofEssayGrading 159
AnalysisofEssayGrading 159
ExploratoryDataAnalysis 160
InferentialStatisticalDataAnalysis 165
Exercises 173
Reference 175
10 SpecificAnalysisExamples 177
HandlingOutliersintheData 177
Floor/CeilingEffects 182
OrderEffects 183
DatafromStratifiedSampling 184
MissingData 184
NoisyData 186
TransformtheData 187
References 188
11 OtherTypesofDataAnalysis 189
Time-SeriesExperiment 189
AnalysisforDataClusters 192
Low-ProbabilityEvents 193
MetadataAnalysis 193
Reference 195
viii Contents
A ResearchTerminology 197
Independent,Dependent,andControlledVariables 197
BetweenSubjectsandWithinSubjects 199
ValidityandReliability 200
VariableTypes 201
TypeofData 201
IndependentMeasuresandRepeatedMeasures 203
VariationinDataCollection 205
Probability—What30%ChanceMeans 212
References 214
Index 215
ix
Preface
Thisbookstrivestobeanintroductiontoquantitativedataanalysisforstudents
whohavelittleornoprevioustrainingeitherinstatisticsorindataanalysis.It
does not attempt to cover all types of data analysis situations, but works to
impartthepropermindsetinperformingadataanalysis.Toooftentheproblem
withpoorlyanalyzedstudiesisnotthenumbercrunchingitself,butalackofthe
critical thinking process required to make sense of the statistical results. This
book works to provide some of that training.
Statisticsisatool.Knowinghowtoperformat-testoranANOVAissimilar
toknowinghowtousestylesandpagelayoutinWord.Justbecauseyouknow
howtousestylesdoesnotmakeyouawriter.Itwillnotevenmakeyouagood
layout person if you do not know when and why to apply those styles.
Likewise, statistics is not data analysis. Learning how to use a software
package to perform a t-test is relatively easy and quick for a student. But
knowingwhenandwhytoperformat-testisadifferent,andmorecomplex,
learningoutcome.Ihadastudent,whohadtakentwograduate-levelbusiness
statistics courses, remark when she turned in a statistics heavy report in a
writingclass:“Inthestatclasses,Ionlylearnedenoughtogetmethroughthe
testproblems.Ihavenoideahowtoanalyzethisdata.”Shehadlearnedhow
tocrunchnumbers,butnothowtoanalyzedata.Bluntly,shewastedhertime
and money in those two classes.
Theissueforresearchersinthesocialsciencesisnottolearnstatistics,butlearn
toanalyzedata.Thegoalisnottolearnhowtousethestatisticalteststocrunch
numbers, but to be able use those tests to interpret the data and draw valid
conclusions from it. There is a wide range of statistical tests relevant to data
analysis; some that every researcher should be able to perform and some that
requiretheadvice/helpofastatisticalexpert.Goodquantitativedataanalysisdoes
notrequireacomprehensiveknowledgeofstatistics,but,rather,knowingenough
toknowwhenitistimetoaskforhelpandwhatquestionstoask.
Everyquantitativeresearchstudy(essentiallybydefinition)collectssometype
ofdatathatmustthenbeanalyzedtohelpdrawthestudy’sconclusions.Agreat
study design is useless unless the data is properly analyzed. But teaching that
x Preface
data analysis to students is a difficult task. What I have found is that most
textbooks fall into one of these categories.
Researchmethodtextbooksthatexplainhowtocreateandexecute astudy,
buttypicallyareverylightonhowtoanalyzethedata.Theyareexcellenton
explainingmethodsofsettingupthestudyandcollectingthedata,butnoton
the methods to analyze it after it has been collected.
Statisticstextbooksthatexplainhowtoperformstatisticaltests.Thetestsare
explained in an acontextual manner and in rigorous statistical terms. Stu
dentslearnhowtoperformatest,but,fromaresearchstandpoint,theequally
importantquestionsofwhenandwhytoperformitgetshortshrift.Asdothe
questionsofhowtointerprettheresultsandhowtoconnectthoseresultsto
the research situation.
Thisbookdiffersfromtextbooksinthesetwocategoriesbecauseitfocuseson
teachinghowtoanalyzedatafromastudy,ratherthanhowtoperformastudy
orhowtoperformindividualstatisticaltests.Noticethatinthefirstsentenceof
apreviousparagraphIsaid“datathatmustthenbeanalyzedtohelpdrawthe
study’s conclusions.” The key word in the sentence is help versus give the
study’sconclusions.Theresultsofstatisticaltestsarenotthefinalconclusions
for research data analysis. The researcher must study the test results, apply
themtothesituationalcontext,andthendrawconclusionsthatmakesense(see
Figure 1.1 in Chapter 1). To support that process, this book works to place
statistical tests within the context of a data analysis problem and provide the
background to connect a specific type of data with the appropriate test. The
work is placed within long examplesand the entire process of data analysis is
coveredinacontextualizedmanner.Itlooksatthedataanalysisfromdifferent
viewpointsandusingdifferentteststoenableastudenttolearnhowandwhen
to apply different analysis methods.
Twomajorgoalsaretoteachwhatquestionstoaskduringallphasesofadata
analysisandhowtojudgetherelevanceofpotentialquestions.Itiseasytorun
statisticaltestsonallcombinationsofthedata,butmostofthosetestshaveno
relevance or validity regardless of the actual research question.
This book strives to explain the when, why, and what for, rather than the
buttonpushinghow-to.Thedataanalysischaptersofmanyresearchtextbooks
are little more than an explanation of various statistical tests. As a result,
studentscomeawaythinkingtheimportantquestionsareprocedural,suchas:
“How do I run a chi-squared test?” “What is the best procedure, a Kruskal–
WallistestorastandardANOVA?”and“Letmetellyouaboutmydata,andyou
cantell mewhat procedureto run.”(Rogers,2010, p.8). These arethewrong
questionstobeaskingatthebeginningofadataanalysis.Rather,studentsneed
to think along the lines of “what relationships do I need to understand?” and
“whataretheimportantpracticalissuesIneedtoworryabout?”Unfortunately,
mostdataanalysistextsgetthemlostisthetreesofindividualtestsandnever
explains where they are within a data analysis forest.