水迷宫重复测量数据的方差分析及其在SPSS中的实现

:(No.39830450)Correspondence:Prof.GuoqinJIN;Email:[e-mail]邱宏1,金国琴2,金如锋1,赵伟康31.,2012032.,2012033.,201203:介绍用SPSS11.0实现对水迷宫重复测量资料的方差分析,以期对采用重复测量设计方案的临床和基础科研工作者提供可借鉴的统计学分析方法:应用SPSS11.0中一般线性模型(generallinearmodel,GLM)的repeatedmeasures和multivariate过程对水迷宫重复测量数据进行重复测量方差分析和多元方差分析,并进行不同时间点和不同组间的两两比较:首先通过球形检验(Mauchlystestofsphericity)的结果判断重复测量数据之间是否存在相关性,如存在相关性(P0.05),宜进行多元方差分析,或采用GreenhouseGeisser的校正结果;通过计算个体间(betweensubject)变异,可分析处理因素有无效应;计算个体内(withinsubject)变异,可分析时间因素有无效应,时间与处理因素之间有无交互效应;使用重复测量数据多重比较配对的t检验法(Bonferroni法),可进行每个分组每个时间点上作用的两两比较;使用多元方差分析的方法,可进行每个时间点上每个分组之间作用的两两比较:对于重复测量数据的方差分析,宜采用一般线性模型的重复测量过程方法进行分析SPSS统计软件易于实现重复测量数据的方差分析:方差分析;统计学;SPSS;线性模型:R573;:A;:16721977(2007)01010105AnalysisofvarianceofrepeateddatameasuredbywatermazewithSPSSHongQIU1,GuoqinJIN2,RufengJIN1,WeikangZHAO31.DepartmentofPreventiveMedicineandMedicalStatistics,ShanghaiUniversityofTraditionalChineseMedicine,Shanghai201203,China2.DepartmentofofBiochemistry,ShanghaiUniversityofTraditionalChineseMedicine,Shanghai201203,China3.InstituteofGeriatrics,ShanghaiUniversityofTraditionalChineseMedicine,Shanghai201203,ChinaObjective:TointroducethemethodofanalyzingrepeateddatameasuredbywatermazewithSPSS11.0,andofferareferencestatisticalmethodtoclinicalandbasicmedicineresearcherswhotakethedesignofrepeatedmeasures.Methods:Usingrepeatedmeasuresandmultivariateanalysisofvariance(ANOVA)processofthegenerallinearmodelinSPSSandgivingcomparisonamongdifferentgroupsanddifferentmeasuretimepairwise.Results:Firstly,Mauchly'stestofsphericityshouldbeusedtojudgewhethertherewererelationsamongtherepeatedlymeasureddata.Ifany(P0.05),multivariateANOVAshouldbetakennext,orGreenhouseGeissercorrectedresultsshouldbetaken.Treatedeffectcouldbeevaluatedbyestimatingbetweensubjectvariance.Repeatedmeasurementeffectoritsinteractiveeffectwithtreatedgroupcouldbeevaluatedbyestimatingwithinsubjectvariance.ThemethodofBonferronishouldbeusedtodopairwisecomparisonsoftherepeatedlymeasureddataindifferentmeasurementtimeofeachtreatedgroup.WithmultivariateANOVA,dataindifferenttreatedgroupofeachmeasurementtimecouldbecomparedpairwise.Conclusion:Therepeatedmeasuresprocessofthegenerallinearmodelissuitableforvarianceanalysisofrepeatedlymeasureddata.SPSSstatisticalpackageisavailabletofulfilthisprocess.Keywords:analysisofvariance;statistics;SPSS;linearmodels1012007151JournalofChineseIntegrativeMedicine,January2007,Vol.5,No.1QiuH,JinGQ,JinRF,ZhaoWK.JChinIntegrMed/ZhongXiYiJieHeXueBao.2007;5(1):101105.ReceivedMarch1,2006;publishedonlineJanuary15,2007.Freefulltext(PDF)isavailableat(Alzheimerdementia,AD)[1]SASSAS,,SPSSSPSS11.0,11.1资料来源(No.39830450),AD1.2动物及分组SD,(20010)g,5:+()+Aricept(Aricept)1.3空间记忆能力测试Morris,1.5cm,19~20,2,,70s,,,(latency)1.4统计学方法HuynhFeldt(Mauchlystestofsphericity),P0.05,,HuynhFeldt,;,P0.05,HuynhFeldt,,,(),(generallinearmodel,GLM),:group(1~5ADAricept),day11day12day21day22day31day32day41day42day51day525dday1~day522.1通过球形检验的结果判断重复测量数据之间是否存在相关性P0.01,5,3(Epsilon)GreenhouseGeisser11Table1MauchlystestofsphericityMeasure:MEASURE_1WithinsubjecteffectMauchly'sWApprox.ChiSquaredfSig.EpsilonGreenhouseGeisserHuynhFeldtLowerboundDay0.43567.69490.0000.7690.8400.2501022007151JournalofChineseIntegrativeMedicine,January2007,Vol.5,No.1:AnalyzeGeneralLinearModelRepeatedMeasuresWithinsubjectfactorname:day(day)Numberoflevels:5:add(5d)DefineWithinsubjectvariables:day1day5(day1day55d)Betweensubjectfactor:groupModelCustom()WithinsubjectModel:day(5)BetweensubjectModel:group()ContinueOK2.2分析时间分组因素的作用以及时间和分组之间的交互作用,,;,(withinsubject),(day)(P0.05),();(daygroup)(P0.05),2,groupP0.05,,32Table2TestsofwithinsubjecteffectsMeasure:MEASURE_1SourceTypesumofsquaresdfMeansquareFSig.DaySphericityassumed18904.95244726.23846.3870.000GreenhouseGeisser18904.9523.0756147.17946.3870.000HuynhFeldt18904.9523.3615624.64546.3870.000Lowerbound18904.9521.00018904.95246.3870.000DaygroupSphericityassumed1143.8101671.4880.7020.792GreenhouseGeisser1143.81012.30292.9810.7020.753HuynhFeldt1143.81013.44485.0770.7020.766Lowerbound1143.8104.000285.9520.7020.593Error(Day)Sphericityassumed33826.378332101.887GreenhouseGeisser33826.378255.257132.519HuynhFeldt33826.378278.971121.254Lowerbound33826.37883.000407.5473Table3TestsofbetweensubjecteffectsMeasure:MEASURE_1.Transformedvariable:averageSourceTypesumofsquaresdfMeansquareFSig.Intercept159748.89211159748.892250.0970.000Group17446.07044361.5176.8280.000Error53016.07583638.7481032007151JournalofChineseIntegrativeMedicine,January2007,Vol.5,No.1Plots5d,,51,5,15dFigure1Thetrendforaveragelatencyofeachgroupduring5days2.3每个时间点上5个分组之间的两两比较,,;AD(P0.05),AD,AriceptAD(P0.05)4:DataSplitFilesAnalyzeallcases,donotcreategroups()AnalyzeGenerallinearmodelMultivariate()Dependentvariables:day1day5Fixedfactors:groupContinueModelCustomModel:groupContinuePostTocPostTocTestsfor:groupLSD(LSD,)ContinueOptionsDisplaymeansfor:groupDisplay:Descriptivestatistics()ContinueOK4ADTable4EffectsofTiaoxinRecipeonspatialmemoryofdeleteriousnetworkofoxidativedamagedADrats(x