多层统计分析模型：SAS与应用

Chapter1Introduction

1.1Conceptualframeworkofmultilevelmodeling

1.2Hierarchicallystructureddata

1.3Variablesinmultileveldata

1.4Analyticalproblemswithmultileveldata

1.5Advantagesandlimitationsofmultilevelmodeling

1.6Computersoftwareformultilevelmodeling

Chapter2BasicsofLinearMultilevelModels

2.1Intraclasscorrelationcoefficient(ICC)

2.2Formulationoftwo-levelmultilevelmodels

2.3Modelassumptions

2.4Fixedandrandomregressioncoefficients

2.5Cross-levelinteractions

2.6Measurementcentering

2.7Modelestimation

2.8Modelfit,hypothesistesting,andmodelcomparisons

2.8.1Modelfit

2.8.2Hypothesistesting

2.8.3Modelcomparisons

2.9Explainedlevel-1andlevel-2variances

2.10Stepsforbuildingmultilevelmodels

2.11Higher-levelmultilevelmodels

Chapter3ApplicationofTwo-levelLinearMultilevelModels

3.1Data

3.2Emptymodel

3.3Predictingbetween-groupvariation

3.4Predictingwithin-groupvariation

3.5Testingrandomlevel-1slopes

3.6Across-levelinteractions

3.7Otherissuesinmodeldevelopment

Chapter4ApplicationofMultilevelModelingtoLongitudinalData

4.1Featuresoflongitudinaldata

4.2Limitationsoftraditionalapproachesformodelinglongitudinaldata

4.3Advantagesofmultilevelmodelingforlongitudinaldata

4.4Formulationofgrowthmodels

4.5Datadescriptionandmanipulation

4.6Lineargrowthmodels

4.6.1Theshapeofaverageoutcomechangeovertime

4.6.2Randominterceptgrowthmodels

4.6.3Randominterceptandslopegrowthmodels

4.6.4Interceptandslopeasoutcomes

4.6.5Controllingforindividualbackgroundvariablesinmodels

4.6.6Codingtimescore

4.6.7Residualvariance/covariancestructures

4.6.8Time-varyingcovariates

4.7Curvilineargrowthmodels

4.7.1Polynomialgrowthmodel

4.7.2Dealingwithcollinearityinhigherorderpolynomialgrowthmodel

4.7.3Piecewise(linearspline)growthmodel

Chapter5MultilevelModelsforDiscreteOutcomeMeasures

5.1Introductiontogeneralizedlinearmixedmodels

5.1.1Generalizedlinearmodels

5.1.2Generalizedlinearmixedmodels

5.2SASProceduresformultilevelmodelingwithdiscreteoutcomes

5.3Multilevelmodelsforbinaryoutcomes

5.3.1Logisticregressionmodels

5.3.2Probitmodels

5.3.3Unobservedlatentvariablesandobservedbinaryoutcomemeasures

5.3.4Multilevellogisticregressionmodels

5.3.5Applicationofmultilevellogisticregressionmodels

5.3.6Applicationofmultilevellogitmodelstolongitudinaldata

5.4Multilevelmodelsforordinaloutcomes

5.4.1Cumulativelogitmodels

5.4.2Multilevelcumulativelogitmodels

5.5Multilevelmodelsfornominaloutcomes

5.5.1Multinomiallogitmodels

5.5.2Multilevelmultinomiallogitmodels

5.5.3Applicationofmultilevelmultinomiallogitmodels

5.6Multilevelmodelsforcountoutcomes

5.6.1Poissonregressionmodels

5.6.2Poissonregressionwithover-dispersionandanegativebinomialmodel

5.6.3MultilevelPoissonandnegativebinomialmodels

5.6.4ApplicationofmultilevelPoissonandnegativebinomialmodels

Chapter6OtherApplicationsofMultilevelModelingandRelatedIssues

6.1Multilevelzero-inflatedmodelsforcountdatawithextrazeros

6.1.1Fixed-effectZIPmodel

6.1.2Randomeffectzero-inflatedPoisson(RE-ZIP)models

6.1.3Randomeffectzero-inflatednegativebinomial(RE-ZINB)models

6.1.4ApplicationofRE-ZIPandRE-ZINBmodels

6.2Mixed-effectmixed-distributionmodelsforsemi-continuousoutcomes

6.2.1Mixed-effectsmixeddistributionmodel

6.2.2ApplicationoftheMixed-Effectmixeddistributionmodel

6.3Bootstrapmultilevelmodeling

6.3.1Nonparametricresidualbootstrapmultilevelmodeling

6.3.2Parametricresidualbootstrapmultilevelmodeling

6.3.3Applicationofnonparametricresidualbootstrapmultilevelmodeling

6.4Group-basedmodelsforlongitudinaldataanalysis

6.4.1Introductiontogroup-basedmodel

6.4.2Group-basedlogitmodel

6.4.3Group-basedzero-inflatedPoisson(ZIP)model

6.4.4Group-basedcensorednormalmodels

6.5Missingvaluesissue

6.5.1Missingdatamechanismsandtheirimplications

6.5.2Handlingmissingdatainlongitudinaldataanalyses

6.6Statisticalpowerandsamplesizeformultilevelmodeling

6.6.1Samplesizeestimationfortwo-leveldesigns

6.6.2Samplesizeestimationforlongitudinaldataanalysis

Reference

　　本书是国内第一本系统介绍各种多层模型的教学和科研参考书。书中采用国际通用的著名统计软件SAS来演示各种多层模型的应用，结合具体的实例，由浅入深地逐步介绍如何使用不同的SAS程序，如ProcMIXED，ProcNLMIXED和ProcGLIMMIX，来进行各种多层资料的模型分析。　　本书可作为综合性大学，医学院、财经大学，师范院校等相应专业的研究生或本科生教材，也可供实际应用工作者参考。　　MultilevelModels:AppficationsUsingSASiswritteninnontechnicalterms,focusesonthemethodsandapplicationsofvariousmultilevelmodels,includinglinermultilevelmodels,multilevellogisticregressionmodels,multilevelPoissonregressionmodels,multilevelnegativebinomialmodels,aswellassomecutting-edgeapplications,suchasmultilevelzero-inflatedPoisson(ZIP)model,randomeffectzero-inflatednegativebinomialmodel(RE-ZINB),mixed-effectmixed-distributionmodels,bootstrappingmultilevelmodels,andgroup-basedtrajectorymodels.Readerswilllearntobuildandapplymultilevelmodelsforhierarchicallystructuredcross-sectionaldataandlongitudinaldatausingtheinternationallydistributedsoftwarepackageStatisticsAnalysisSystem(SAS).DetailedSASsyntaxandoutputareprovidedformodelapplications,providingstudents,researchscientistsanddataanalystswithreadytemplatesfortheirapplications.