版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领
文档简介
Lecture13
MultipleClassificationAnalysis(MCA)1ThisLectureCoversTheMCAastheequivalentofamultipleregressionanalysisMCAadaptedtologisticregression2
Multipleclassificationanalysis(MCA)isusedtoexaminetheeffectofeachindependentvariableonthedependentvariablewhilecontrollingfortheeffectsoftheotherindependentvariables,whenthedependentvariableisanquantitativevariableandtheindependentvariablesarecategorical.3
MCAismosteasilyexplainedasmultipleregressionwithdummyvariables.Thus,thedependentvariableisaquantitativevariable,andtheindependentvariablesarecategoricalvariables,representedbydummyvariables.4
MCAwithonecategoricalpredictorvariableisequivalenttoone-wayanalysisofvariance;MCAwithtwocategoricalpredictorvariablesisequivalenttotwo-wayanalysisofvariance;andsoon.5ControlvariablesmaybeaddedtotheMCAmodel.Whenquantitativevariables,inadditiontocategoricalvariables,areincludedamongthepredictorvariables,MCAisequivalenttowhatiscommonlycalledanalysisofcovariance.6
MCAcanbeextendedbeyondmultipleregression,forexample,itcanbeextendedtologisticregressionandCoxregression.7MCAspecifiesthelinearmodelasfollows:8MCAisastatisticaloptionwiththeANOVAprocedure9AnexampleToexaminetheeffectsofplaceofresidence,ethnicityandeducationlevelonwomen’sageatfirstmarriageANOVAVARIABLES=afmBYuandr(1,2)ethnic(1,2)educat(1,5)/MAXORDERSNONE/STATISTICSMEANMCA/METHODEXPERIM.10111213TheMCAoutputshowstheestimated(orpredicted)meansofthedependentvariableforeachcategoryoftheexplanatoryvariables,unadjustedandadjustedfortheeffectsoftheotherexplanatoryvariablesinthemodel.Italsoshowstheunadjustedandadjusteddeviationsfromthegrandmean.Therefore,thepredictedmeanminusthedeviationforeachcategoryisalwaysthesameandisequaltothegrandmean.
predictedmean–deviation=grandmeanThe“deviationsadjustedforfactors”areequivalenttotheb1,b2,…bncoefficientsaftercontrollingfortheeffectoftheotherexplanatoryvariables.14Eachcombinationofthebcoefficientswillgivetheestimatedvalueofthedependentvariableforarespondentwiththecorrespondingcharacteristics.15Acomparisonoftheunadjustedandadjustedmeansforeachcategoryoftheindependentvariablesshowswhathappenswhenanadjustmentismadefortheeffectsoftheothervariables.Thelargertherangeinthedeviationsamongthecategoriesofeachexplanatoryvariable,thegreaterthesignificanceofthatfactorinaffectingthedependentvariable.16Ifwerunmultipleregression1718
Multipleclassificationanalysisisanextensionofmultipleregressionthatallowsustouseregressioncoefficientstopredictthemeanvaluecontrollingfortheeffectofotherpredictorsinthemodel.MCAissimplyawayofsolvingtheregressionequation.
19UnadjustedmeansUnadjustedmeanscanbeobtainedfromsimplelinearregression:“unadjusted”means“withoutcontrols”:20AdjustedmeansAdjustedmeansareobtainedfrommultiplelinearregression:“adjusted”means“withcontrols”:21AdjustedmeansWhenwecalculatemeanageatfirstmarriagebyresidence,ethnicityandeducationarethecontrolvariables;Whenwecalculatemeanageatfirstmarriagebyethnicity,residenceandeducationarethecontrolvariables;Whenwecalculatemeanageatfirstmarriagebyeducation,residenceandethnicityarethecontrolvariables.22AdjustedmeansIngeneral,whenweconsideradjustedmeansforonepredictorvariable,alltheothervariablesarethecontrolvariables.Statisticalcontrolsareintroducedbyholdingthecontrolvariablesconstantattheirmeanvalues.23HoldingthecontrolvariablesconstantattheirmeanvaluesForacontinuousvariable,suchasage,themeanvalueseemsanappropriatemeasure.Inessencewearecontrollingfortheaverageexperienceorperson.However,theuseofameanforadichotomous‘dummy’variableseemsartificialaseachindividualcanhaveavalueof0or1butnoonecanhaveavalueinbetween0and1.However,ifwethinkofthemeanofadichotomousvariableasaproportionitsusemakesmoresense.Herewearecontrollingfortheproportionofmales(forsex)ortheproportionofpeoplewithcollegeeducationetc.
24TosolvetheregressionequationusingExcelThefollowingexamplepresentsawayofsettingupanMCAcalculationtableinExcel.
Weneedthecoefficientsoftheregressionmodel.ThesecanbecopiedstraightfromtheSPSSoutputintoExcel.ThenweneedtogeneratethemeanvalueofeachindependentvariableandcopythesevaluesintoExcel.Wegivetheconstantavalueof1inthemeancolumnsothatwhenwemultiplythecolumnsitremainsattheoriginalvalue.
2526IncludingintervalindependentvariablesinMCAIntervalscalevariablescanbeincludedinMCAascovariates.Theirroleinthemodelcanbeviewedintwowaysconceptually:
●asacontrolvariable
●asanotherexplanatoryvariable27SPSSSyntax28AnexampleToexaminetheeffectsofplaceofresidence,ethnicityandeducationlevelonwomen’sageatfirstmarriage,whencontrollingforwomen’sageANOVAVARIABLES=afmBYuandr(1,2)ethnic(1,2)educat(1,5)WITHage/COVARIATESWITH/MAXORDERSNONE/STATISTICSMEANMCAREG/METHODEXPERIM.29303132333435Thethreeequationsarethesame:36MCAAdaptedto
LogisticRegressionWhenMCAisadaptedtologisticregression,bothunadjustedandadjustedvaluesoftheresponsevariablecanbecalculated,justasinordinaryMCA.37Theunadjustedvaluesarebasedonlogisticregressionthatincorporateonepredictorvariableatatime,andtheadjustedvaluesarebasedonthecompletemodelincludingallpredictorvariablessimultaneously.38UnfortunatelySPSSdoesnotincludeMCAprogramsforlogisticregression,wehavetoconstructtheMCAtablesfromtheunderlyinglogisticregressions.39AnIllustrativeExampleToillustratehowMCAcanbeadaptedtologisticregression,weconsidertheabortionexample:theabortionuseeffectofage,pregnancy,residence,ethnicity,andeducation.40VariablesP:estimatedprobabilityofabortionuseAgeisclassifiedintotwobroadgroups:15-34and35-49Numberofpregnanciesinthreecategories:1,2,and3andoverResidence,ethnicity,andeducationcategorizedasformerly4142Whencalculatingthevalueoffromthefittedregressionforeachcategoryofaparticularindependentvariable,dummyvaluesofthatvariablearesettobecombinationsofonesandzeroswhileallothervariablesarecontrolledbyholdingthemconstantattheirmeanvalues.
43AdjustedPExponentiationofthevaluesofyieldstheadjustedvaluesof()
AdjustedvaluesofParecalculatedas:44AnimportantpointUnlikeOLSregressioninwhichsubstitutionofmeanvaluesfortheindependentvariablesalwaysyieldsthemeanvalueofthedependentvariable,thisisgenerallynotthecaseforlogisticregression.45Inlinearregression,substitutionofmeanvaluesfortheindependentvariablesalwaysyieldsthemeanvalueofthedependentvariable46Thisiscalculatedfromtheregressionequation47OverallmeanageatfirstmarriagecalculatedfromthesampleThisisdirectlycalculatedfromthesample48However,thevalueofoverallPproducedfromthelogisticregressionequationisgenerallynot
温馨提示
- 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
- 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
- 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
- 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
- 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
- 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
- 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。
最新文档
- 鹰潭职业技术学院《音乐教师素质课程》2024-2025学年第二学期期末试卷
- 轻工制造行业市场前景及投资研究报告:供应链出海产业趋势
- 硬质合金精加工工测试验证竞赛考核试卷含答案
- 气体深冷分离工班组考核评优考核试卷含答案
- 四年级简便运算100道练习题(含答案)
- 内画工安全理论强化考核试卷含答案
- 聚乙烯装置操作工风险识别考核试卷含答案
- 纬编工安全知识水平考核试卷含答案
- 压电石英晶体配料装釜工岗前安全演练考核试卷含答案
- 甲醛装置操作工成果转化强化考核试卷含答案
- 2025年贵州省普通高中学业水平合格性考试模拟(四)历史试题(含答案)
- GB/T 45732-2025再生资源回收利用体系回收站点建设规范
- CJ/T 120-2016给水涂塑复合钢管
- 痰液粘稠度护理
- 广西南宁市2025届高三下学期第二次适应性考试化学试题(原卷版+解析版)
- 核电子学试题及答案
- 【初中 语文】第15课《青春之光》课件-2024-2025学年统编版语文七年级下册
- 高校大学物理绪论课件
- 生产周报工作总结
- 2025年黑龙江省高职单招《语文》备考重点试题库(含真题)
- 国网福建省电力限公司2025年高校毕业生(第二批)招聘高频重点提升(共500题)附带答案详解
评论
0/150
提交评论