SPSS專用統(tǒng)計術語

1、SPSS用統(tǒng)計術語1、X/dfx2(卡方檢驗)檢驗方法和分解融合(DF)方法x2值(Chi-Square)為22.4601,自由度(chi-squaredf)為4,P值(Pr>Chi-Square)為0.00022、RMRLISREL、AMOS等結構方程模型分析軟件，為什么我的Amos輸出中找不到RMR和GFI這一項內(nèi)容。(1)間距參數(shù)RSm與混合參數(shù)Rmr(c)稱為附加參數(shù)(2)表示殘差均方根，殘差是指樣本導出的方差協(xié)方差矩陣與模型隱含的方差協(xié)方差矩陣中對應元素的差值，RMR的值越小，模型擬合越好。rTtAu«s十口3 . anu+ AjiaL*yis:Nu：“ fui G工

2、口國+ 7ai iable ?LumimLy:?diajiBttl finiundif4 Nr:"foi咂面£ t EsiiJiates' HiniJiiaatian+ PairviEC Faram?t?Model Fit' 三尺 cut Mr； TimeMuAcIFilCAHN口更里展卷.圖里ilSlSummavyNTAKDFTCMIN/DF109S03 2B5485,0001 6565<M,0000S32076 25S561,0003 701 ELII 口/,哲胡笑涅臼曲*SAmarJJn摘妾慨琵Group r.inr.tor 1HasplinrC

3、fiHf雨KFIIFITLIFIDeftalrholJJela2ri02,613,552300,757.7303alui抵di。Jell.r)OD1.0001.000Lndependeiiccmodel,000,000000,000.000ParauHAiylirasPRA7I0PCHDetajlt.Q.530.6333ahuat&dnode：,DDO,000.DOOInJcpcndciiccnwdcl1ODO,000.003NCP匕心dslNCPLOSOHI90Ljdavitmo4由ng.2St244238400.If5SarirftpdnoiR'000,000CigLidc

4、pcndcDccmodel25152561579.5721658,470FMNFkLLN從你的圖片看，你的模型擬合度不好。AmosdoesnotreportGFI,PGFI,AGFIandRMRwhenyouestimatemeansandintercepts.Thisisbecauseitisnotclearhowtoincorporatemeansandinterceptsintotheconventionalformulasforthesestatistics.Onepossibilityistocalculatethesestatisticsbasedontheobservedvari

5、ances/covariancesandthefitted(implied)variances/covariances,leavingthemeansandinterceptsoutofthefitmeasure.Thatwayofcomputingthestatisticscouldbeusefulandiscertainlydefensible.Itwouldmakeagreatdealofsenseinthosemodelswheremeansandinterceptsareestimatedbutnotconstrained.Ontheotherhand,formodelsthatco

6、nstrainmeansandinterceptsitisimportanttopayattentiontohowwellthemodelreproducesthemeansandintercepts,aswellasthevariances/covariances.Anotherapproachwouldbetogeneralizethedefinitionsofthosestatisticstosomehowincorporatefailuretofitmeansandintercepts.Sofar,aunique,obviouslycorrect,wayofdoingthishasno

7、tbeenproposed.Forthesereasons,thedesigndecisionwasmadenottoreportGFI,PGFI,AGFIandRMRwhenmeansandinterceptsareestimated.回頭去看你的"View"里的"AnalysisProperties”,aEstimatesmeansandintercepets”是被選中了的，把那個復選框去掉后，報表里就有RMR和GFI等指數(shù)啦。3、NFI（1）神經(jīng)纖維指數(shù)；（2）NormedFitIndex,正規(guī)指數(shù)，越接近于1,說明擬合越好。本特勒一波內(nèi)特規(guī)范合指數(shù)（NFI

8、）;（3）NormedFitIndex,正規(guī)擬合指數(shù)4、rmsea近彳以誤差均方根，rootmeansquareerrorofapproximation，衡量模型協(xié)方差與數(shù)據(jù)協(xié)方差陣的差異大小，RMSFA越小越好。5、GFI擬合優(yōu)度指數(shù)goodnessoffitindex6、RFI相對擬合指數(shù)，（1）renalfunctionindex腎功能指數(shù)（2）RelativeFitIndex7、NNFI非本特內(nèi)-范內(nèi)特規(guī)范擬合指數(shù)，8、IFIIncrementalFitIndex,增量擬合指數(shù)9、AGFIadjustedGFI,調(diào)整擬合優(yōu)度指數(shù)10、TLITucker-LewisIndex指數(shù)，TLI

9、可能大于1或者小于1,但一月在01之間，越接近1模型越理想。11、CFIComparativeFitIndex,比較擬合指數(shù)TheconfirmatoryfatoranalysisshowedthatthefitindexesforX2/df=3.622,GFI=0.92,AGFI=0.91,NFI=0.89,NNFI=0.90,RMSEA=0.069,CFI=0.90,IFI=0.90.StrivingforhighgoalshadnosignificantcorrelationswithSCL-90、SDS、SASandSWLS,butwithSES（r=0.241,P<0.01）.

10、W：夜有驗證性因素分析顯示擬合指數(shù)X2/df=3.622,GFI=0.92,AGFI=0.91,NFI=0.87,NNFI=0.90,RMSEA=0.069,CFI=0.90,IFI=0.90。追求高標準與SCL-90、SDS、SAS和SWLS等效標不相關，與SES的相關為0.241（P<0.01）;信息標準指數(shù)ISI完全標準化解（MI）相對擬合指數(shù)（RFI）大于0.90,近似均方根殘差（RMSEA）小于0.08,則模型與數(shù)據(jù)的擬合程度很好。非范擬合指數(shù)（NNFI）為0.93,近似均方根誤差（RMSEA）為0.079。12個分測驗的完全標準化解（MI）分別為與自由度df之比、近似誤差均方

11、根（RMSEA）、擬合優(yōu)度指數(shù)（goodnessoffitindex,GFI）本特勒一波內(nèi)特規(guī)范擬合指數(shù)（NFI）、近似誤差平方根（RMSEA）和信息標準指數(shù)等?？筛鶕?jù)用于驗證的數(shù)據(jù)特征1.2009年7月7日.7?goodnessoffitindexNNFICFI?df=,p（p+1）/2.（CFI）=0.95IncrementalFitIndex（IFI）=0.95RelativeFitIndex（RFI）=0.88.運行l(wèi)isrel會產(chǎn)生一個GF文件，就是模型擬合指標的文件，內(nèi)容如下（文件中是居中設置的）：DegreesofFreedom=35MinimumFitFunctionChi-S

12、quare=47.95(P=0.071)NormalTheoryWeightedLeastSquaresChi-Square=47.07(P=0.083)EstimatedNon-centralityParameter(NCP)=12.0790PercentConfidenceIntervalforNCP=(0.0;34.10)MinimumFitFunctionValue=0.048PopulationDiscrepancyFunctionValue(F0)=0.01290PercentConfidenceIntervalforF0=(0.0;0.034)RootMeanSquareErro

13、rofApproximation(RMSEA)=0.01990PercentConfidenceIntervalforRMSEA=(0.0;0.031)P-ValueforTestofCloseFit(RMSEA<0.05)=1.00ExpectedCross-ValidationIndex(ECVI)=0.08790PercentConfidenceIntervalforECVI=(0.075;0.11)ECVIforSaturatedModel=0.11ECVIforIndependenceModel=1.87Chi-SquareforIndependenceModelwith45D

14、egreesofFreedom=1843.25IndependenceAIC=1863.25ModelAIC=87.07SaturatedAIC=110.00IndependenceCAIC=1922.33ModelCAIC=205.23SaturatedCAIC=434.93NormedFitIndex(NFI)=0.97Non-NormedFitIndex(NNFI)=0.99ParsimonyNormedFitIndex(PNFI)=0.76ComparativeFitIndex(CFI)=0.99IncrementalFitIndex(IFI)=0.99RelativeFitIndex

15、(RFI)=0.97CriticalN(CN)=1195.65RootMeanSquareResidual(RMR)=0.049StandardizedRMR=0.021GoodnessofFitIndex(GFI)=0.99AdjustedGoodnessofFitIndex(AGFI)=0.99ParsimonyGoodnessofFitIndex(PGFI)=0.63txt 文件，并于首行指明擬合as.numeric() 轉(zhuǎn)成數(shù)值。細我現(xiàn)在想要做的是從多個這樣的文件中抽取特定的擬合指標的值另存為一個指標的名稱？在R在能否實現(xiàn)呢？該如何操作？能。因為這個文件的規(guī)律很明確。用readLine

16、s()讀進來，用split()拆分'='，節(jié)我就不說了，里面有幾行和其它行的規(guī)律不完全一樣。用兩行文本舉個例子吧，比如我現(xiàn)在讀了兩行文本放在x中：>x1 "P-ValueforTestofCloseFit(RMSEA<0.05)=1.00"2 "ExpectedCross-ValidationIndex(ECVI)=0.087"用'='為拆分符號把x拆掉：>strsplit(x,'=')11 "P-ValueforTestofCloseFit(RMSEA<0.05)"2 "1.00