Вы находитесь на странице: 1из 10

Page1 of10

Print|CloseWindow
Note:Largeimagesandtablesonthispagemaynecessitateprintinginlandscapemode.

Basic&ClinicalBiostatistics>Chapter1.IntroductiontoMedicalResearch>

INTRODUCTIONTOMEDICALRESEARCH:INTRODUCTION
Thegoalofthistextistoprovideyouwiththetoolsandskillsyouneedtobeasmartuserandconsumer
ofmedicalstatistics.Thisgoalhasguidedusintheselectionofmaterialandinthepresentationof
information.Thischapteroutlinesthereasonsphysicians,medicalstudents,andothersinthehealthcare
fieldshouldknowbiostatistics.Italsodescribeshowthebookisorganized,whatyoucanexpecttofindin
eachchapter,andhowyoucanuseitmostprofitably.

THESCOPEOFBIOSTATISTICS&EPIDEMIOLOGY
Theword"statistics"hasseveralmeanings:dataornumbers,theprocessofanalyzingthedata,andthe
descriptionofafieldofstudy.ItderivesfromtheLatinwordstatus,meaning"mannerofstanding"or
"position."Statisticswerefirstusedbytaxassessorstocollectinformationfordeterminingassetsand
assessingtaxesanunfortunatebeginningandonetheprofessionhasnotentirelyliveddown.
Everyoneisfamiliarwiththestatisticsusedinbaseballandothersports,suchasabaseballplayer's
battingaverage,abowler'sgamepointaverage,andabasketballplayer'sfreethrowpercentage.In
medicine,someofthestatisticsmostoftenencounteredarecalledmeans,standarddeviations,
proportions,andrates.Workingwithstatisticsinvolvesusingstatisticalmethodsthatsummarizedata(to
obtain,forexample,meansandstandarddeviations)andusingstatisticalprocedurestoreachcertain
conclusionsthatcanbeappliedtopatientcareorpublichealthplanning.Thesubjectareaofstatisticsis
thesetofallthestatisticalmethodsandproceduresusedbythosewhoworkwithstatistics.The
applicationofstatisticsisbroadindeedandincludesbusiness,marketing,economics,agriculture,
education,psychology,sociology,anthropology,andbiology,inadditiontoourspecialinterest,medicine
andotherhealthcaredisciplines.Hereweusethetermsbiostatisticsandbiometricstorefertothe
applicationofstatisticsinthehealthrelatedfields.
Althoughthefocusofthistextisbiostatistics,sometopicsrelatedtoepidemiologyareincludedaswell.
Thesetopicsandothersspecifictoepidemiologyarediscussedinmoredetailinthecompanionbook,
MedicalEpidemiology(Greenberg,2000).Theterm"epidemiology"referstothestudyofhealthandillness
inhumanpopulations,or,moreprecisely,tothepatternsofhealthordiseaseandthefactorsthat
influencethesepatternsitisbasedontheGreekwordsfor"upon"(epi)and"people"(demos).Once
knowledgeoftheepidemiologyofadiseaseisavailable,itisusedtounderstandthecauseofthedisease,
determinepublichealthpolicy,andplantreatment.Theapplicationofpopulationbasedinformationto
decisionmakingaboutindividualpatientsisoftenreferredtoasclinicalepidemiologyand,more
recently,evidencebasedmedicine.Thetoolsandmethodsofbiostatisticsareanintegralpartofthese
disciplines.

BIOSTATISTICSINMEDICINE
Cliniciansmustevaluateandusenewinformationthroughouttheirlives.Theskillsyoulearninthistext

Page2 of10

willassistinthisprocessbecausetheyconcernmodernknowledgeacquisitionmethods.Inthefollowing
subsections,welistthemostimportantreasonsforlearningbiostatistics.(Themostwidelyapplicable
reasonsarementionedfirst.)
EvaluatingtheLiterature
Readingtheliteraturebeginsearlyinthetrainingofhealthprofessionalsandcontinuesthroughouttheir
careers.Theymustunderstandbiostatisticstodecidewhethertheycanbelievetheresultspresentedinthe
literature.Journaleditorstrytoscreenoutarticlesthatareimproperlydesignedoranalyzed,butfewhave
formalstatisticaltrainingandtheynaturallyfocusonthecontentoftheresearchratherthanthemethod.
Investigatorsforlarge,expensivestudiesalmostalwaysconsultstatisticiansforassistanceinproject
designanddataanalysis,especiallyresearchfundedbytheNationalInstitutesofHealthandothernational
agenciesandfoundations.Eventhenitisimportanttobeawareofpossibleshortcomingsinthewaya
studyisdesignedandcarriedout.Insmallerresearchprojects,investigatorsconsultwithstatisticiansless
frequently,eitherbecausetheyareunawareoftheneedforstatisticalassistanceorbecausethe
biostatisticalresourcesarenotreadilyavailableoraffordable.Theavailabilityofeasytousecomputer
programstoperformstatisticalanalysishasbeenimportantinpromotingtheuseofmorecomplex
methods.Thissameaccessibility,however,enablespeoplewithoutthetrainingorexpertiseinstatistical
methodologytoreportcomplicatedanalyseswhentheyarenotalwaysappropriate.
Theproblemswithstudiesinthemedicalliteraturehavebeenamplydocumented,andwesuggestyou
examinesomeofthefollowingreferencesformoreinformation.Oneofthemostcomprehensivereports
wasbyWilliamsonandcolleagues(1992)whoreviewed28articlesthatexaminedthescientificadequacy
ofstudydesigns,datacollection,andstatisticalmethodsinmorethan4200publishedmedicalstudies.The
reportsassesseddrugtrialsandsurgical,therapeutic,anddiagnosticprocedurespublishedinmorethan30
journals,manyofthemwellknownandprestigious(eg,BritishMedicalJournal,JournaloftheAmerican
MedicalAssociation,NewEnglandJournalofMedicine,CanadianMedicalAssociationJournal,andLancet).
Williamsonandcolleaguesdeterminedthatonlyabout20%of4235researchreportsmettheassessors'
criteriaforvalidity.Eightoftheassessmentarticleshadalsolookedattherelationshipbetweenthe
frequencyofpositivefindingsandtheadequacyofthemethodsusedinresearchreportsandfoundthat
approximately80%ofinadequatelydesignedstudiesreportedpositivefindings,whereasonly25%of
adequatelydesignedstudieshadpositivefindings.Thus,evidenceindicatesthatpositivefindingsare
reportedmoreofteninpoorlyconductedstudiesthaninwellconductedones.
Otherarticlesindicatethattheproblemswithreportedstudieshavenotimprovedsubstantially.Mllner
andcolleagues(2002)reviewed34commonlyreadmedicaljournalsandfoundinadequatereportingof
methodstocontrolforconfounding(otherfactorsthatmayaffecttheoutcome).Mossandcolleagues
(2003)reportedthat40%ofpulmonaryandcriticalcarearticlesthatusedlogisticregression(discussedin
Chapter10)maynothaveusedthemethodappropriately.TheproblemisnotlimitedtotheEnglish
speakingliterature.Asexamples,Hayran(2002)reportedthat56%ofthearticlesintheTurkishliterature
usedimproperorinadequatestatistics,andSkovlund(1998)evaluatedcancerresearcharticlespublished
byNorwegianauthorsandfoundthat64%hadunsuitablestatisticalmethods.Severaljournals,including
theNewEnglandJournalofMedicine,theJournaloftheAmericanMedicalAssociation,theCanadian
MedicalAssociationJournal,andtheBritishMedicalJournal,havecarriedseriesofarticlesonstudydesign

Page3 of10

andstatisticalmethods.Someofthearticleshavebeenpublishedinaseparatemonograph(eg,Bailarand
Mosteller,1992Greenhalgh,1997b).
Journalshavealsopublishedanumberofarticlesthatsuggesthowpractitionerscouldbetterreporttheir
researchfindings.Weagreewithmanyoftheserecommendations,butwefirmlybelievethatwe,as
readers,mustassumetheresponsibilityfordeterminingwhethertheresultsofapublishedstudyarevalid.
Ourdevelopmentofthisbookhasbeenguidedbythestudydesignsandstatisticalmethodsfound
primarilyinthemedicalliterature,andwehaveselectedtopicstoprovidetheskillsneededtodetermine
whetherastudyisvalidandshouldbebelieved.Chapter13focusesspecificallyonhowtoreadthe
medicalliteratureandprovideschecklistsforflawsinstudiesandproblemsinanalysis.
ApplyingStudyResultstoPatientCare
Applyingtheresultsofresearchtopatientcareisthemajorreasonpracticingcliniciansreadthemedical
literature.Theywanttoknowwhichdiagnosticproceduresarebest,whichmethodsoftreatmentare
optimal,andhowthetreatmentregimenshouldbedesignedandimplemented.Ofcourse,theyalsoread
journalstostayawareanduptodateinmedicineingeneralaswellasintheirspecificareaofinterest.
Chapters3and12discusstheapplicationoftechniquesofevidencebasedmedicinetodecisionsaboutthe
careofindividualpatients.Greenhalgh(2002)discusseswaystointegratequalitativeresearchinto
evidencebasedmedicine.
InterpretingVitalStatistics
Physiciansmustbeabletointerpretvitalstatisticsinordertodiagnoseandtreatpatientseffectively.Vital
statisticsarebasedondatacollectedfromtheongoingrecordingofvitalevents,suchasbirthsanddeaths.
Abasicunderstandingofhowvitalstatisticsaredetermined,whattheymean,andhowtheyareused
facilitatestheiruse.Chapter3providesinformationonthesestatistics.
UnderstandingEpidemiologicProblems
Practitionersmustunderstandepidemiologicproblemsbecausethisinformationhelpsthemmake
diagnosesanddevelopmanagementplansforpatients.Epidemiologicdatarevealtheprevalenceofa
disease,itsvariationbyseasonoftheyearandbygeographiclocation,anditsrelationtocertainrisk
factors.Inaddition,epidemiologyhelpsusunderstandhownewlyidentifiedvirusesandotherinfectious
agentsspread.Thisinformationhelpssocietymakeinformeddecisionsaboutthedeploymentofhealth
resources,forexample,whetheracommunityshouldbeginasurveillanceprogram,whetherascreening
programiswarrantedandcanbedesignedtobeefficientandeffective,andwhethercommunityresources
shouldbeusedforspecifichealthproblems.Describingandusingdatainmakingdecisionsarehighlighted
inChapters3and12.
InterpretingInformationAboutDrugsandEquipment
Physicianscontinuallyevaluateinformationaboutdrugsandmedicalinstrumentsandequipment.This
materialmaybeprovidedbycompanyrepresentatives,sentthroughthemail,orpublishedinjournals.
Becauseofthehighcostofdevelopingdrugsandmedicalinstruments,companiesdoalltheycanto
recouptheirinvestments.Toselltheirproducts,acompanymustconvincephysiciansthatitsproductsare
betterthanthoseofitscompetitors.Tomakeitspoints,itusesgraphs,charts,andtheresultsofstudies
comparingitsproductswithothersonthemarket.Everychapterinthistextisrelatedtotheskillsneeded
toevaluatethesematerials,butChapters2,3,and13areespeciallyrelevant.

Page4 of10

UsingDiagnosticProcedures
Identifyingthecorrectdiagnosticproceduretouseisanecessityinmakingdecisionsaboutpatientcare.In
additiontoknowingtheprevalenceofagivendisease,physiciansmustbeawareofthesensitivityofa
diagnostictestindetectingthediseasewhenitispresentandthefrequencywithwhichthetestcorrectly
indicatesnodiseaseinawellperson.Thesecharacteristicsarecalledthesensitivityandspecificityofa
diagnostictest.InformationinChapters4and12relatesparticularlytoskillsforinterpretingdiagnostic
tests.
BeingInformed
Keepingabreastofcurrenttrendsandbeingcriticalaboutdataaremoregeneralskillsandonesthatare
difficulttomeasure.Theseskillsarealsonoteasyforanyonetoacquirebecausemanyresponsibilities
competeforaprofessional'stime.Oneofthebyproductsofworkingthroughthistextisaheightened
awarenessofthemanythreatstothevalidityofinformation,thatis,theimportanceofbeingalertfor
statementsthatdonotseemquiteright.
AppraisingGuidelines
Thenumberofguidelinesfordiagnosisandtreatmenthasincreasedgreatlyinrecentyears.Practitioners
cautionthatguidelinesshouldnotbeaccepteduncriticallyalthoughsomearebasedonmedicalevidence,
manyrepresentthecollectiveopinionofexperts.AreviewofguidelinesinlungcancerbyHarpoleand
colleagues(2003)foundthatlessthan30%wereevidencebasedandconcludedthattheguidelinesmay
reflectclinicalpracticeinsteadofqualitydata.
EvaluatingStudyProtocolsandArticles
Physiciansandothersinthehealthfieldwhoareassociatedwithuniversities,medicalschools,ormajor
clinicsareoftencalledontoevaluatematerialsubmittedforpublicationinmedicaljournalsandtodecide
whetheritshouldbepublished.Healthpractitioners,ofcourse,havetheexpertisetoevaluatethecontent
ofaprotocolorarticle,buttheyoftenfeeluncomfortableaboutcritiquingthedesignandstatistical
methodsofastudy.Nostudy,howeverimportant,willprovidevalidinformationaboutthepracticeof
medicineandfutureresearchunlessitisproperlydesignedandanalyzed.Carefulattentiontotheconcepts
coveredinthistextwillprovidephysicianswithmanyoftheskillsnecessaryforevaluatingthedesignof
studies.
ParticipatinginorDirectingResearchProjects
Cliniciansparticipatinginresearchwillfindknowledgeaboutbiostatisticsandresearchmethods
indispensable.Residentsinallspecialtiesaswellasotherhealthcaretraineesareexpectedtoshow
evidenceofscholarlyactivity,andthisoftentakestheformofaresearchproject.Thecomprehensive
coverageoftopicsinthistextshouldprovidemostofthemwiththeinformationtheyneedtobeactive
participantsinallaspectsofresearch.

THEDESIGNOFTHISBOOK
Weconsiderthistexttobebothbasicandclinicalbecauseweemphasizeboththebasicconceptsof
biostatisticsandtheuseoftheseconceptsinclinicaldecisionmaking.Wehavedesignedacomprehensive
textcoveringthetraditionaltopicsinbiostatisticsplusthequantitativemethodsofepidemiologyusedin
research.Forexample,weincludecommonlyusedwaystoanalyzesurvivaldatainChapter9illustrations
ofcomputeranalysesinchaptersinwhichtheyareappropriate,becauseresearcherstodayusecomputers

Page5 of10

tocalculatestatisticsandapplicationsoftheresultsofstudiestothediagnosisofspecificdiseasesandthe
careofindividualpatients,sometimesreferredtoasmedicaldecisionmakingorevidencebasedmedicine.
Wehaveaddedanumberofnewtopicstothiseditionandhaveextendedourcoverageofsomeothers.
Wehaveenhancedourdiscussionofevidencebasedmedicineandcontinuetoemphasizetheimportant
conceptofthenumberofpatientsthatneedtobetreatedwithagiveninterventioninordertopreventone
undesirableoutcome(numberneededtotreat).Thelikelihoodratio,oftenpresentedinmanyjournal
articlesthatevaluatediagnosticproceduresoranalyzeriskfactors,iscoveredinmoredetail.Wecontinue
tousecomputersoftwaretoillustratethediscussionofthenumberofsubjectsneededindifferenttypesof
studies(power).Wehaveincreasedourcoverageoftheincreasinglyimportantmultivariatemethods,
especiallylogisticregressionandtheCoxproportionalhazardmodel.Thechaptersonsurvival
methodsandanalysisofvariancehavebeenrevised.Duetoseveralsuggestions,wehaveincludedan
entirelynewchapteronsurveyresearchthatwethinkwillbeofinteresttomanyofourreaders.
Ourapproachdeemphasizescalculationsandusescomputerprogramstoillustratetheresultsofstatistical
tests.Inmostchapters,weincludethecalculationsofsomestatisticalprocedures,primarilybecausewe
wishtoillustratethelogicbehindthetests,notbecausewebelieveyouneedtobeabletoperformthe
calculationsyourself.Someexercisesinvolvecalculationsbecausewehavefoundthatsomestudentswish
toworkthroughafewproblemsindetailsoastounderstandtheproceduresbetter.Themajorfocusof
thetext,however,isontheinterpretationanduseofresearchmethods.
Awordregardingtheaccuracyofthecalculationsisinorder.Manyexamplesandexercisesrequireseveral
steps.Theaccuracyofthefinalanswerdependsonthenumberofsignificantdecimalplacestowhich
figuresareextendedateachstepofthecalculationwegenerallyextendthemtotwoorthreeplaces.
Calculatorsandcomputers,however,useagreaternumberofsignificantdecimalplacesateachstepand
oftenyieldananswerdifferentfromthatobtainedusingonlytwoorthreesignificantdigits.Thedifference
usuallywillbesmall,butdonotbeconcernedifyourcalculationsvaryslightlyfromours.
Theexamplesusedaretakenfromstudiespublishedinthemedicalliterature.Occasionally,weusea
subsetofthedatatoillustrateamorecomplexprocedure.Inaddition,wesometimesfocusononlyone
aspectofthedataanalyzedinapublishedstudyinordertoillustrateaconceptorstatisticaltest.To
explaincertainconcepts,weoftenreproducetablesandgraphsastheyappearinapublishedstudy.These
reproductionsmaycontainsymbolsthatarenotdiscusseduntilalaterchapterinthisbook.Simplyignore
suchsymbolsforthetimebeing.Wechosetoworkwithpublishedstudiesfortworeasons:First,they
convincereadersoftherelevanceofstatisticalmethodsinmedicalresearchandsecond,theyprovidean
opportunitytolearnaboutsomeinterestingstudiesalongwiththestatistics.
Wehavealsomadeanefforttoprovideinsightsintothecoherencyofstatisticalmethods.Weoftenrefer
tobothpreviousandupcomingchapterstohelptieconceptstogetherandpointoutconnections.This
techniquerequiresustousedefinitionssomewhatdifferentlyfrommanyotherstatisticaltextsthatis,
termsareoftenusedwithinthecontextofadiscussionwithoutaprecisedefinition.Thedefinitionisgiven
later.Severalexamplesappearintheforegoingdiscussions(eg,vitalstatistics,means,standard
deviations,proportions,rates,validity).Webelievethatusingtermsproperlywithinseveralcontextshelps
thereaderlearncomplexideas,andmanyideasinstatisticsbecomeclearerwhenviewedfromdifferent

Page6 of10

perspectives.Sometermsaredefinedaswegoalong,butprovidingdefinitionsforeverytermwould
inhibitourabilitytopointouttheconnectionsbetweenideas.Toassistthereader,weuseboldfacefor
terms(thefirstfewtimestheyareused)thatappearintheGlossaryofstatisticalandepidemiologicterms
providedattheendofthebook.

THEORGANIZATIONOFTHISBOOK
Inadditiontothenewtopics,inthethirdeditionwereorganizedsomeofthechapterstocoincidemore
closelywiththewaywethinkaboutresearch.Forexample,manyofthesameprinciplesareinvolvedin
methodstosummarizeanddisplaydata,sothesearenowinthesamechapter(Chapter3)anddiscussed
inanintegratedmanner.Mostbiostatisticaltexts,ourpreviouseditionsincluded,dividethemethodsinto
chapters,with,forexample,achapteronttests,anotherchapteronchisquare( 2)tests,andso
forth.Inthisedition,weorganizethemethodstorelatetothekindofresearchquestionbeingasked.
Therefore,thereisachapteronanalyzingresearchquestionsinvolvingonegroupofsubjects(Chapter5),
anotherchapteronresearchquestionsinvolvingtwogroupsofsubjects(Chapter6),andyetanotheron
researchquestionsinvolvingmorethantwogroupsofsubjects(Chapter7).Webelievethisorganizationis
morelogical,andwehopeitfacilitatesthelearningprocess.
Eachchapterbeginswithtwocomponents:keyconceptsandanintroductiontotheexamples(presenting
problems)coveredinthechapter.Thekeyconceptsareintendedtohelpreadersorganizeandvisualize
theideastobediscussedandthentoidentifythepointatwhicheachisdiscussed.Attheconclusionof
eachchapterisasummarythatintegratesthestatisticalconceptswiththepresentingproblemsusedto
illustratethem.Whenflowchartsordiagramsareuseful,weincludethemtohelpexplainhowdifferent
proceduresarerelatedandwhentheyarerelevant.TheflowchartsaregroupedinAppendixCforeasy
reference.
Patientscometotheirhealthcareproviderswithvarioushealthproblems.Indescribingtheirpatients,
theseproviderscommonlysay,"Thepatientpresentswith..."or"Thepatient'spresentingproblemis..
."Weusethisterminologyinthistexttoemphasizethesimilaritybetweenmedicalpracticeandthe
researchproblemsdiscussedinthemedicalliterature.Almostallchaptersbeginwithpresentingproblems
thatdiscussstudiestakendirectlyfromthemedicalliteraturetheseresearchproblemsareusedto
illustratetheconceptsandmethodspresentedinthechapter.Inchaptersinwhichstatisticsarecalculated
(eg,themeaninChapter3)orstatisticalproceduresareexplained(eg,thettestinChapters5and6),
datafromthepresentingproblemsareusedinthecalculations.Wetrytoensurethatthemedicalcontent
ofthepresentingproblemsisstillvalidatthetimethistextispublished.Asadvancesaremadein
medicine,however,itispossiblethatsomeofthiscontentwillbecomeoutdatedwewillattempttocorrect
thatinthenextedition.
Wehaveattemptedtoselectpresentingproblemsthatrepresentabroadarrayofinterests,whilebeing
surethatthestudiesusethemethodswewanttodiscuss.Beginningwiththethirdedition,we
incorporateddatafromanumberofinvestigators,whogenerouslyagreedtosharetheirdatawithus,and
wehavecontinuedthatpracticeinthefourthedition.Furthermore,manyauthorsagreedtoletuspublish
asubsetoftheirdatawiththistext,thusprovidinguswithanumberofadvantages:Wecanuserealdata
inthestatisticalprograms,andreaderscanusethesamedatatoreinforceandextendtheirknowledgeof
themethodswediscuss.Sometimeswefocusononlyasmallpartoftheinformationpresentedinthe

Page7 of10

articleitself,butweusuallycommentontheirfindingsinthesummaryofthechapter.Wehavetriednot
tomisinterpretanyofthedataorreportedfindings,andwetakeresponsibilityforanyerrorswemayhave
committedindescribingtheirresearch.Werecommendthatourreadersobtainacopyoftheoriginal
publishedarticleanduseit,alongwithourdiscussionofthestudy,toenhancetheirclinicalknowledgeas
wellastheirstatisticalexpertise.
ThiseditioncontinuestheinclusionofactualdataandsoftwareontheCDROMthataccompaniesthistext.
[NotetoAccessLangeusers:thisdataandsoftwareisnotavailableonthewebsite.]Thedatasetsare
providedinseveraldifferentformatstomakeiteasytousethemforstatisticalanalysis.Thesoftwareis
thatdevelopedbyDr.JerryHintze:theNumberCruncherStatisticalSystem(NCSS),asetofstatistical
procedures,andthePowerAnalysisStatisticalSystem(PASS),acomputerprogramthatestimatesthe
samplesizesneededinastudy.TheroutinesfromtheNCSSandPASSsoftwarethatareillustratedinthis
textareincludedontheCD.Wehaveusedthissoftwareforanumberofyearsandfinditcomprehensive
andeasytouse.Manyoftheillustrationsofthestatisticalproceduresinthisbookwerefacilitatedbyusing
NCSS,andwehopeourreadersuseittoreplicatetheanalysesandtodotheexercises.Asanadded
benefit,theproceduresincludedonthisversionofNCSScanalsobeusedtoanalyzeotherdatasets.
Pleaserefertotheinstructions,immediatelyfollowingtheAcknowledgments,entitled"UsingtheCDROM"
forinformationonfindingthedocumentationforutilizingtheprogramsanddatasetsontheCDROM.
WehaveestablishedanInternetWebsitetoprovideyouwiththemostuptodateinformationand
additionalresourcesonbiostatistics.TheWebsiteaddressishttp://www.clinicalbiostatistics.com
Wehopeyouwillusethesite,andweinviteyoutosendcommentsorsuggestions,includingany
calculationerrorsyoumayfind,tousattheemailaddresspostedontheWebsite.Weappreciatethe
communicationsfromreadersofthepreviouseditionsithelpsustrytomakethenexteditionbetter.
Somestatisticaltestsarenotroutinelyavailableonmostcommercialpackages.Wehavedesigneda
spreadsheetprogramtoanalyzethesespecialinstances.ThesefilesareincludedinafolderontheCD
namedCalculations.Again,refertotheinstructionsimmediatelyfollowingtheAcknowledgmentsentitled
"UsingtheCDROM"formoreinformation.
Exercisesareprovidedwithallchapters(2,3,4,5,6,7,8,9,10,11,12,13)answersaregivenin
AppendixB,mostwithcompletesolutions.Weincludedifferentkindsofexercisestomeetthedifferent
needsofstudents.Someexercisescallforcalculatingastatisticorastatisticaltest.Somefocusonthe
presentingproblemsorotherpublishedstudiesandaskaboutthedesign(asinChapter2)oraboutthe
useofelementssuchascharts,graphs,tables,andstatisticalmethods.Occasionally,exercisesextenda
conceptdiscussedinthechapter.Thisadditionaldevelopmentisnotcriticalforallreaderstounderstand,
butitprovidesfurtherinsightsforthosewhoareinterested.Someexercisesrefertotopicsdiscussedin
previouschapterstoprovideremindersandreinforcements.
Again,wehighlyrecommendthatyouobtaincopiesoftheoriginalarticlesandreviewthemintheir
entirety.Ifyouareusingthisbookinanorganizedcourse,wesuggestyouformsmallgroupstodiscuss
thearticlesandexaminehowtheconceptscoveredinthebookaredealtwithbythevariousresearchers.
Finally,inresponsetoapleafromourownstudents,acollectionofmultiplechoicequestionsisgivenin
Chapter13thesequestionsprovideausefulposttestforstudentswhowanttobesuretheyhave

Page8 of10

masteredthematerialpresentedinthetext.
Thesymbolsusedinstatisticsaresometimesasourceofconfusion.Thesesymbolsarelistedonthe
insidebackcoverforreadyaccess.Whenmorethanonesymbolforthesameitemisencounteredinthe
medicalliterature,weusethemostcommononeandpointouttheothers.Also,aswementioned
previously,aglossaryofbiostatisticalandepidemiologictermsisprovidedattheendofthebook(after
Chapter13).
Symbol
=

equals

doesnotequal

<

lessthan
lessthanorequalto

>

greaterthan
greaterthanorequalto
squarerootofa

|a|

absolute(orpositive)valueofa

P(A)

probabilityofeventA

P(A|B)

probabilityofeventAgiventhateventBhashappened

H0

nullhypothesis

H1

alternativehypothesis

GreekletteralphaprobabilityoftypeIerror
GreekletterbetaprobabilityoftypeIIerroralso,populationvalueoftheslopeofthe
regressionline
Greekletterdeltapopulationmeandifferencebetweentwopairedmeasurements
Greekletterkappausedtodenoteanindexofagreementorreproducibility
Greekletterlambdausedtodenotetermsintheloglinearmodel
Greeklettermupopulationmean
Greekletterpipopulationproportion
Greekletterrhopopulationcorrelation
Greeklowercaselettersigmapopulationstandarddeviation
Greekuppercaselettersigmasymbolindicatingasum
CV

coefficientofvariation

df

degreeoffreedom

baseofnaturallogarithm(equalto2.718)

expectedfrequency

symbolfortheFtestanddistribution

hazardfunction

Page9 of10

LR

likelihoodratio

MSA

meansquareamonggroups

MSE

errormeansquare

samplesize

observedfrequency

samplecorrelation

r2

squaredcorrelation,calledofthecoefficientofdetermination

R2

squaredmultiplecorrelationinmultipleregression

SD

samplestandarddeviation

SE

standarderrorofthemean

sY.X

standarderroroftheestimateinregression

symbolforthetratio(thecriticalratiothatfollowsatdistribution)

independent(explanatory,predictor)variableinregression

symbolforthechisquaretest

xbarsamplemean
Y

dependent(outcome,response,criterion)variableinregression

Y'

predictedvalueofYregression

symbolforthezratio(thecriticalratiothatfollowsazorstandardnormaldistribution

ADDITIONALRESOURCES
Weprovidereferencestoothertextsandjournalarticlesforreaderswhowanttolearnmoreaboutatopic.
WiththegrowthoftheInternet,manyresourceshavebecomeeasilyavailableforlittleornocost.A
numberofstatisticalprogramsandresourcesareavailableontheInternet.Someoftheprogramsare
freeware,meaningthatanyonemayusethemfreeofchargeothers,calledshareware,chargearelatively
smallfeefortheiruse.Manyofthesoftwarevendorshavefreeproductsorsoftwareyoucandownloadand
useforarestrictedperiodoftime.
Wehavelistedafewexcellentsitesforinformationonbiostatistics.BecauseInternetaddresseschange
periodically,thefollowingsitesmaybedifferentbythetimeyoutrytoaccessthem.Wewillattemptto
monitorthesesitesandwillpostanychangesonourWebsitehttp://www.clinicalbiostatistics.com
StatisticalSolutionshttp://www.statsol.ie/hassimpleprogramstofindsamplesizesforcomparingtwo
meansortwoproportionsaswellascalculatingthestatisticaltestsforcomparingtwomeansortwo
proportions.
TheNCSShomepagehttp://www.ncss.comhaslinkstoafreeprobabilitycalculatoryoucanusetolookup
valuesinstatisticaldistributions(ratherthanusingthetablesinAppendixA).

Page10 of10

TheAmericanStatisticalAssociation(ASA)hasanumberofsectionswithaspecialemphasis,suchas
TeachingStatisticsintheHealthSciences,BiometricsSection,StatisticalEducation,andothers.Manyof
theseSectionhomepagescontainlinkstostatisticalresources.TheASAhomepageis
http://www.amstat.org
TheTeachingStatisticsintheHealthScienceshomepageishttp://www.bio.ri.ccf.org/ASA_TSHSandhas
linkstoteachingmaterials.
DartmouthUniversityhaslinkstotheimpressiveChanceDatabasehttp://www.dartmouth.edu/%
7Echance/index.htmlwhichcontainsmanyteachingresourcesand,inturn,manyusefullinkstoother
resources.
TheMedicalUniversityofSouthCarolinahaslinkstoalargenumberofevidencebasedmedicinesites,
includingitsownresourceshttp://www.musc.edu/dc/icrebm/index.html
TheInternationalStatisticalInstituteNewsletterVolume26,No.1(76)2002providesareviewoffree
statisticalinformationonthewebathttp://www.cbs.nl/isi/NLet02104.htm
CopyrightTheMcGrawHillCompanies.Allrightsreserved.
PrivacyNotice.AnyuseissubjecttotheTermsofUseandNotice.

Вам также может понравиться