Springer Texts in Statistics
e e e
SerieseEditors:
G.eCasella
S.eFienberg
I.eOlkin
Foreotheretitlesepublishedeinethiseseries,egoetoehttp://www.sp
ringer.com/series/417
,Peter D. Hoff
e e
A First Course in Bayesian Statisti
e e e e e
cal Methods
e
13 e
,PetereD.eHoffe Departme
nteofeStatisticseUniversity
eofeWashingtoneSeattleeW
Ae98195-4322eUSA
ISSNe 1431-875X
ISBNe 978-0-387-92299-7 e-ISBNe 978-0-387-92407-
6eDOIe 10.1007/978-0-387-92407-6
SpringereDordrechteHeidelbergeLondoneNeweYork
LibraryeofeCongresseControleNumber:e2009929120
⃝e ce SpringereScience+BusinesseMedia,eLLC e2009
Allerightsereserved.eThiseworkemayenotebeetranslatedeorecopiedeinewholeeoreinepartewithoutetheewrittene p
ermissioneofetheepublishere(SpringereScience+BusinesseMedia,eLLC,e233eSpringeStreet,eNeweYork,e NY
10013,eUSA),eexcepteforebriefeexcerptseineconnectionewithereviewseorescholarlyeanalysis.eUseeine conne
ctionewitheanyeformeofeinformationestorageeanderetrieval,eelectroniceadaptation,ecomputeresoftware,e oreb
yesimilareoredissimilaremethodologyenoweknowneorehereafteredevelopedeiseforbidden.
Theeuseeinethisepublicationeofetradeenames,etrademarks,eserviceemarks,eandesimilareterms,eeveneifetheyear
ee noteidentifiedeasesuch,eisenotetoebeetakeneaseaneexpressioneofeopinioneasetoewhethereorenotetheyeareesubjec
tetoe proprietaryerights.
Printedeoneacid-freeepaper
SpringereiseparteofeSpringereScience+BusinesseMediae(www.springer.com)
, Preface
Thisebookeoriginatedefromeaeseteofelectureenoteseforeaeone-quarteregraduate-
elevelecourseetaughteatetheeUniversityeofeWashington.eTheepurposeeofetheecours
eeisetoefamiliarizeetheestudentsewithetheebasiceconceptseofeBayesianetheoryeand
e toequicklyegetethemeperformingetheireownedataeanalyseseusingeBayesianecom-
eputationaletools.eTheeaudienceeforethisecourseeincludesenon-
statisticsegraduateestudentsewhoedidewelleinetheiredepartment’segraduate-
leveleintroductoryestatis-
eticsecourses eandewhoealsoehaveeaneinteresteinestatistics.eAdditionally,efirst-
eandesecond-
yearestatisticsegraduateestudentsehaveefoundethisecourseetoebeeaeusefuleintrodu
ctionetoestatisticalemodeling.eLikeetheecourse,ethisebookeiseintendedetoe beeaeself
-
containedeandecompacteintroductionetoetheemaineconceptseofeBayesianetheorye
andepractice.eByetheeendeofetheetext,ereaderseshouldehaveetheeabilityetoeunderst
andeandeimplementetheebasicetoolseofeBayesianestatisticalemethodseforetheireo
wnedataeanalysisepurposes.eTheetexteisenoteintendedeaseaecomprehen-
esiveehandbookeforeadvancedestatisticaleresearchers,ealthougheiteisehopedethatet
hiselatterecategoryeofereadersecouldeuseethisebookeaseaequickeintroductionetoeBa
yesianemethodseandeaseaepreparationeforemoreecomprehensiveeandedetailedest
udies.
Computing
MonteeCarloesummarieseofeposterioredistributionseplayeaneimportanteroleeine t
hee waye datae analysese aree presentede ine thise text.e Mye experiencee hase beene
thateonceeaestudenteunderstandsetheebasiceideaeofeposterioresampling,etheireda
taeanalysesequicklyebecomeemoreecreativeeandemeaningful,eusingerelevantepost
eriorepredictiveedistributionseandeinterestingefunctionseofeparameters.eTheeope
n-sourceeRestatisticalecomputingeenvironmenteprovidesesufficientefunction-
ealityetoemakeeMonteeCarloeestimationeveryeeasyeforeaelargeenumbereofestatis-
eticale models,e ande examplee R-
codee ise providede throughoute thee text.e Muche ofetheeexampleecodeecanebeerune
“aseis”eineR,eandeessentiallyealleofeitecanebeeruneafteredownloadingetheerelevante
datasetsefrometheecompanionewebsiteeforethisebook.
e e e
SerieseEditors:
G.eCasella
S.eFienberg
I.eOlkin
Foreotheretitlesepublishedeinethiseseries,egoetoehttp://www.sp
ringer.com/series/417
,Peter D. Hoff
e e
A First Course in Bayesian Statisti
e e e e e
cal Methods
e
13 e
,PetereD.eHoffe Departme
nteofeStatisticseUniversity
eofeWashingtoneSeattleeW
Ae98195-4322eUSA
ISSNe 1431-875X
ISBNe 978-0-387-92299-7 e-ISBNe 978-0-387-92407-
6eDOIe 10.1007/978-0-387-92407-6
SpringereDordrechteHeidelbergeLondoneNeweYork
LibraryeofeCongresseControleNumber:e2009929120
⃝e ce SpringereScience+BusinesseMedia,eLLC e2009
Allerightsereserved.eThiseworkemayenotebeetranslatedeorecopiedeinewholeeoreinepartewithoutetheewrittene p
ermissioneofetheepublishere(SpringereScience+BusinesseMedia,eLLC,e233eSpringeStreet,eNeweYork,e NY
10013,eUSA),eexcepteforebriefeexcerptseineconnectionewithereviewseorescholarlyeanalysis.eUseeine conne
ctionewitheanyeformeofeinformationestorageeanderetrieval,eelectroniceadaptation,ecomputeresoftware,e oreb
yesimilareoredissimilaremethodologyenoweknowneorehereafteredevelopedeiseforbidden.
Theeuseeinethisepublicationeofetradeenames,etrademarks,eserviceemarks,eandesimilareterms,eeveneifetheyear
ee noteidentifiedeasesuch,eisenotetoebeetakeneaseaneexpressioneofeopinioneasetoewhethereorenotetheyeareesubjec
tetoe proprietaryerights.
Printedeoneacid-freeepaper
SpringereiseparteofeSpringereScience+BusinesseMediae(www.springer.com)
, Preface
Thisebookeoriginatedefromeaeseteofelectureenoteseforeaeone-quarteregraduate-
elevelecourseetaughteatetheeUniversityeofeWashington.eTheepurposeeofetheecours
eeisetoefamiliarizeetheestudentsewithetheebasiceconceptseofeBayesianetheoryeand
e toequicklyegetethemeperformingetheireownedataeanalyseseusingeBayesianecom-
eputationaletools.eTheeaudienceeforethisecourseeincludesenon-
statisticsegraduateestudentsewhoedidewelleinetheiredepartment’segraduate-
leveleintroductoryestatis-
eticsecourses eandewhoealsoehaveeaneinteresteinestatistics.eAdditionally,efirst-
eandesecond-
yearestatisticsegraduateestudentsehaveefoundethisecourseetoebeeaeusefuleintrodu
ctionetoestatisticalemodeling.eLikeetheecourse,ethisebookeiseintendedetoe beeaeself
-
containedeandecompacteintroductionetoetheemaineconceptseofeBayesianetheorye
andepractice.eByetheeendeofetheetext,ereaderseshouldehaveetheeabilityetoeunderst
andeandeimplementetheebasicetoolseofeBayesianestatisticalemethodseforetheireo
wnedataeanalysisepurposes.eTheetexteisenoteintendedeaseaecomprehen-
esiveehandbookeforeadvancedestatisticaleresearchers,ealthougheiteisehopedethatet
hiselatterecategoryeofereadersecouldeuseethisebookeaseaequickeintroductionetoeBa
yesianemethodseandeaseaepreparationeforemoreecomprehensiveeandedetailedest
udies.
Computing
MonteeCarloesummarieseofeposterioredistributionseplayeaneimportanteroleeine t
hee waye datae analysese aree presentede ine thise text.e Mye experiencee hase beene
thateonceeaestudenteunderstandsetheebasiceideaeofeposterioresampling,etheireda
taeanalysesequicklyebecomeemoreecreativeeandemeaningful,eusingerelevantepost
eriorepredictiveedistributionseandeinterestingefunctionseofeparameters.eTheeope
n-sourceeRestatisticalecomputingeenvironmenteprovidesesufficientefunction-
ealityetoemakeeMonteeCarloeestimationeveryeeasyeforeaelargeenumbereofestatis-
eticale models,e ande examplee R-
codee ise providede throughoute thee text.e Muche ofetheeexampleecodeecanebeerune
“aseis”eineR,eandeessentiallyealleofeitecanebeeruneafteredownloadingetheerelevante
datasetsefrometheecompanionewebsiteeforethisebook.
