Contributions to Statistics Joachim Kunert Christine H. Müller Editors Anthony C. Atkinson mODa 11 - Advances in Model- Oriented Design and Analysis Proceedings of the 11th International Workshop in Model-Oriented Design and Analysis held in Hamminkeln, Germany, June 12-17, 2016 Contributions to Statistics Moreinformationaboutthisseriesathttp://www.springer.com/series/2912 Joachim Kunert • Christine H. MuRller • Anthony C. Atkinson Editors mODa 11 - Advances in Model-Oriented Design and Analysis Proceedings of the 11th International Workshop in Model-Oriented Design and Analysis held in Hamminkeln, Germany, June 12-17, 2016 123 Editors JoachimKunert ChristineH.MuRller DepartmentofStatistics DepartmentofStatistics TechnischeUniversitaRtDortmund TechnischeUniversitaRtDortmund Dortmund,Germany Dortmund,Germany AnthonyC.Atkinson DepartmentofStatistics LondonSchoolofEconomics London,UnitedKingdom ISSN1431-1968 ContributionstoStatistics ISBN978-3-319-31264-4 ISBN978-3-319-31266-8 (eBook) DOI10.1007/978-3-319-31266-8 LibraryofCongressControlNumber:2016940826 ©SpringerInternationalPublishingSwitzerland2016 Thisworkissubjecttocopyright.AllrightsarereservedbythePublisher,whetherthewholeorpartof thematerialisconcerned,specificallytherightsoftranslation,reprinting,reuseofillustrations,recitation, broadcasting,reproductiononmicrofilmsorinanyotherphysicalway,andtransmissionorinformation storageandretrieval,electronicadaptation,computersoftware,orbysimilarordissimilarmethodology nowknownorhereafterdeveloped. Theuseofgeneraldescriptivenames,registerednames,trademarks,servicemarks,etc.inthispublication doesnotimply,evenintheabsenceofaspecificstatement,thatsuchnamesareexemptfromtherelevant protectivelawsandregulationsandthereforefreeforgeneraluse. Thepublisher,theauthorsandtheeditorsaresafetoassumethattheadviceandinformationinthisbook arebelievedtobetrueandaccurateatthedateofpublication.Neitherthepublishernortheauthorsor theeditorsgiveawarranty,expressorimplied,withrespecttothematerialcontainedhereinorforany errorsoromissionsthatmayhavebeenmade. Printedonacid-freepaper ThisSpringerimprintispublishedbySpringerNature TheregisteredcompanyisSpringerInternationalPublishingAGSwitzerland Preface This volume contains articles based on presentations at the 11th workshop on model-oriented data analysis and optimum design (mODa) in Hamminkeln- Dingden, Germany, during June 2016. The 11th workshop was organized by the Departmentof Statistics of the TU Dortmundand supported by the Collaborative ResearchCenter“Statisticalmodelingofnonlineardynamicprocesses”(SFB823) oftheGermanResearchFoundation(DFG). The mODa series of workshopsfocuses on nonstandarddesign of experiments andrelatedanalysisofdata.Themainobjectivesare: • To promote new advanced research areas as well as collaboration between academiaandindustry. • Whenever possible, to provide financial support for research in the area of experimentaldesignandrelatedtopics. • Togivejuniorresearcherstheopportunityofestablishingpersonalcontactsand workingtogetherwithleadingresearchers. • Tobringtogetherscientistsfromdifferentstatisticalschools–particularempha- sisisgiventotheinclusionofscientistsfromCentralandEasternEurope. The mODa series of workshops started at the Wartburg near Eisenach in the formerGDR in 1987and hascontinuedas a tri-annualseries of conferences.The locationsanddatesoftheformerconferencesareasfollows: • mODa1:Eisenach,formerGDR,1987, • mODa2:St.Kyrik,Bulgaria,1990, • mODa3:Peterhof,Russia,1992, • mODa4:Spetses,Greece,1995, • mODa5:Luminy,France,1998, • mODa6:Puchberg/Schneeberg,Austria,2001, • mODa7:Heeze,TheNetherlands,2004, • mODa8:Almagro,Spain,2007, • mODa9:Bertinoro,Italy,2010, • mODa10:ŁagówLubuski,Poland,2013. v vi Preface The articles in this volume provide an overview of current topics in research on experimentaldesign.Thetopicscoveredbythepapersare: • designsfortreatmentcombinations(Atkinson;Druilhet;GrömpingandBailey), • randomisation(Bailey;Ghiglietti;ShaoandRosenberger), • computerexperiments(CurtisandMaruri-Aguilar;Ginsbourger,Baccou,Cheva- lierandPerales), • designs for nonlinear regression and generalized linear models (Amo-Salas, Jiménez-Alcázar and López-Fidalgo;Burclová and Pázman;Cheng, Majumdar andYang;Mielke;RadloffandSchwabe), • designs for dependent data (Deldossi, Osmetti and Tommasi; Gauthier and Pronzato;PrusandSchwabe), • designsforfunctionaldata(Aletti,MayandTommasi;ZangandGroßmann), • adaptive and sequential designs (Borrotti and Pievatolo; Hainy, Drovandi and McGree;Knapp;Lane,WangandFlournoy), • designsfor specialfields of application(Bischoff;FedorovandXue; Graßhoff, HollingandSchwabe;Pepelyshev,StaroselskiyandZhigljavsky), • foundationsofexperimentaldesign(MüllerandWynn;Zhigljavsky,Golyandina andGillard). In this time of Big Data, it is often not emphasized in public discourse that experimentaldesignremainsextremelyimportant.ThemODaseriesofworkshops wishes to raise public awareness of the continuing importance of experimental design. In particular, the papers from various fields of application show that experimental design is not a mathematical plaything, but is of direct use in the sciences. Since the first workshop in Eisenach, optimal design for varioussituations has been at the heart of the research covered by mODa. Sequential design is another long-standing topic in the mODa series. It is clear that computer experiments, designsfor dependentdata, and functionaldata become increasinglyfeasible. For causalinferenceinparticular,old-fashionedmethodslikerandomization,blinding, andorthogonalityoffactorsremainindispensable.Inadditiontotheimportanceof theresearchcoveredhere,wethinkthatthearticlesinthisvolumeshowthebeauty ofmathematicalstatistics,whichshouldnotbeforgotten. For the editors, it was a pleasurereading these research results. We would like to thank the authors for submitting such nice work and for providingrevisions in time, wherevera revision was necessary.Last, but notleast, we want to thankthe refereeswhoprovidedthoughtfulandconstructivereviewsintime,helpingtomake thisvolumeafineadditiontoanystatistician’sbookshelves. Dortmund,Germany ChristineMüller JoachimKunert AnthonyAtkinson Contents OnApplyingOptimalDesignofExperimentswhenFunctional ObservationsOccur.............................................................. 1 GiacomoAletti,CaterinaMay,andChiaraTommasi OptimalDesignsforImplicitModels .......................................... 11 Mariano Amo-Salas, Alfonso Jiménez-Alcázar, andJesúsLópez-Fidalgo OptimumExperimentswithSetsofTreatmentCombinations ............. 19 AnthonyC.Atkinson DesignKeysforMultiphaseExperiments..................................... 27 R.A.Bailey OnDesignsforRecursiveLeastSquaresResidualstoDetect Alternatives....................................................................... 37 WolfgangBischoff A Multi-objective Bayesian Sequential Design Based on ParetoOptimality................................................................ 47 MatteoBorrottiandAntonioPievatolo OptimumDesignviaI-DivergenceforStableEstimationin GeneralizedRegressionModels ................................................ 55 KatarínaBurclováandAndrejPázman OnMultiple-ObjectiveNonlinearOptimalDesigns.......................... 63 QianshunCheng,DibyenMajumdar,andMinYang DesignforSmoothModelsoverComplexRegions........................... 71 PeterCurtisandHugoMaruri-Aguilar PKL-Optimality Criterion in Copula Models forEfficacy-ToxicityResponse.................................................. 79 LauraDeldossi,SilviaAngelaOsmetti,andChiaraTommasi vii viii Contents EfficientCircularCross-overDesignsforModelswithInteraction........ 87 PierreDruilhet SurvivalModelswithCensoringDrivenbyRandomEnrollment.......... 95 ValeriiV.FedorovandXiaoqiangXue OptimalDesignforPredictioninRandomFieldModelsvia CovarianceKernelExpansions................................................. 103 BertrandGauthierandLucPronzato AsymptoticPropertiesofanAdaptiveRandomlyReinforced UrnModel......................................................................... 113 AndreaGhiglietti DesignofComputerExperimentsUsingCompetingDistances BetweenSet-ValuedInputs...................................................... 123 David Ginsbourger, Jean Baccou, Clément Chevalier, andFrédéricPerales OptimalDesignfortheRaschPoisson-GammaModel ...................... 133 UlrikeGraßhoff,HeinzHolling,andRainerSchwabe RegularFractionsofFactorialArrays......................................... 143 UlrikeGrömpingandR.A.Bailey Likelihood-Free Extensions for Bayesian Sequentially DesignedExperiments........................................................... 153 MarkusHainy,ChristopherC.Drovandi,andJamesM.McGree AConfidenceIntervalApproachinSelf-DesigningClinicalTrials......... 163 GuidoKnapp ConditionalInferenceinTwo-StageAdaptiveExperimentsvia theBootstrap ..................................................................... 173 AdamLane,HaiYingWang,andNancyFlournoy Study Designs for the Estimation of the Hill Parameter inSigmoidalResponseModels.................................................. 183 TobiasMielke ControlledVersus“Random”Experiments:APrinciple.................... 191 WernerG.MüllerandHenryP.Wynn AdaptiveDesignsforOptimizingOnlineAdvertisementCampaigns...... 199 AndreyPepelyshev,YuriStaroselskiy,andAnatolyZhigljavsky Interpolationand Extrapolationin Random Coefficient RegressionModels:OptimalDesignforPrediction.......................... 209 MarynaPrusandRainerSchwabe Contents ix Invariance and Equivariance in Experimental Design forNonlinearModels ............................................................ 217 MartinRadloffandRainerSchwabe PropertiesoftheRandomBlockDesignforClinicalTrials ................. 225 HuiShaoandWilliamF.Rosenberger FunctionalDataAnalysisinDesignedExperiments ......................... 235 BairuZhangandHeikoGroßmann AnalysisandDesignintheProblemofVectorDeconvolution .............. 243 AnatolyZhigljavsky,NinaGolyandina,andJonathanGillard Index............................................................................... 253