我们的网站为什么显示成这样?

可能因为您的浏览器不支持样式,您可以更新您的浏览器到最新版本,以获取对此功能的支持,访问下面的网站,获取关于浏览器的信息:

|本期目录/Table of Contents|

 验证性因素分析中模型拟合的判断(PDF)

《心理学探新》[ISSN:1003-5184/CN:36-1228/B]

期数:
 2007年27卷04期
页码:
 83
栏目:
 心理统计与测量
出版日期:
 2007-12-20

文章信息/Info

Title:
 Assessing the Mode lFitin Confirmatory Factor Analysis
作者:
 郭庆科 王炜丽 陈雪霞 韩 丹
 辽宁师范大学心理学系,大连116029
Author(s):
 Guo Qi ngke Wang Weili Chen Xuexi a
 Liaoning Normal University,Department of Psychology,Dalian 116029
关键词:
 结构方程模型模型拟合拟合指数α与β错误
Keywords:
 Structural Equation Modeling Model fit Fit Indices αand βtype error 78
分类号:
 B841·2
DOI:
 -
文献标识码:
 A
摘要:
 采用与Hu &Bentler 不同的设计进行模拟研究,发现在他们推荐的8个指数中,NNFI、CFI、IFI 在所有样本量条件下几乎都能得到最低的α与β错误率,而且最佳界值都在0·95左右,属于优良的指数。但Mc 与Gamma Hat 的表现却不好,因此不建议使用。他们建议的2指标策略被证明能降低α与β错误率,在样本量较小时尤其值得尝试。
Abstract:
 Inthis si mulation study,we designed 6loading size conditions(0·03~0·08)and 6sample size conditions(150,250,500,1000,2000,5000)to test the capacity ofthe 8fit indices and 2-indexstrategies recommended by Hu &Bentlerto avoid αand βtype error·The truemodel is also a CFAmodel with3factors and 15items ,but eachitemonlyloads on one factor ,andinthe wrong model there is a mis -specifiedfactor loading·The results showthat amongthe 8fit indices NNFI ,CFI andIFI always produce the least sumof αand βtype error ,their best cut-off values consistentlyfall around 0·95across all sample size conditions·2-indexstrategies can produce less sumof αand βtype error com—pared to single indexstrategies ,they are especially useful when sample size is small·

参考文献/References

 1 Jreskog K G,Srbom D.LISREL8:User’s reference guide .Chicago ,IL:Scientific Software international ,1996·
2 Hu LT,Bentler P.Evaluating model fit .In:R .H.Hoyle ,Ed·Structural Equation Modeling·Concepts ,Issues ,and Appli—cations·London:Sage ,1995·76-99·
3 Bentler P M·EQSstructural equations program manual·Encino ,CA:Multivariate Software ,1995·
4 Fan X,Thompson B,Wang L·Effects of sample size ,esti mationmethods ,and model specification on structural equation modelingfit indexes·Structural Equation Modeling ,1999,(6):56-83·
5 Muth n LK,Muth n BO·Mplus User’s Guide·Los Angeles ,CA,2001·
6 Bollen K A·Overall fit in covariance structure models :Two types68心理学探新2007年of sample size effects·Psychological Bulletin,1990,107:256-259·
7 Marsh H W,Balla JR ,McDonald R P·Goodness -of -fit in—dexes in confirmatory factor analysis :The effect of samplesize·Psychological Bulletin,1988,103(3):391-410·
8 McDonald R P,Marsh H W·Choosing a multivariate model :Non-centrality and goodness -of -fit·Psychological Bulletin,1990,107:247-255·
9 Marsh H W,Hau K-T·Assessing goodness of fit :Is parsi monyalways desirable ?The Journal of Experi mental Education,64,1996:364-390·
10 Hu L,Bentler P M·Fit indices in covariance structure analysis :Sensitivity to underparameterized model misspecifica—tion·Psychological Methods ,1998(3):424-453·
11 Hu L,Bentler P M·Cutoff criteriafor fit indexes in covariancestructure analysis :Conventional criteria versus new alterna—tives·Structural Equation Modeling ,1999,6(1):1-55·
12 Marsh H W,Hau K -T,Wen Z·In:search of golden rules :Comment on hypothesis testing approaches to setting cut offvalues for fit indexes and dangers in overgeneralising Hu &Bentler‘s(1999)findings·Structural Equation Modeling ,2004,11(3):320-341·
13 温忠麟,侯杰泰,马什赫伯特·结构方程模型检验:拟合指数与卡方准则·心理学报,2005,36(2):186-194·
14 Yuan K-H·Fit Indices Versus Test Statistics·Multivariate Be—havioral Research,2005,40(1):115-148·
15 Sivo S A,Fan X,Willse JT·The search for“Opti mal”CutoffProperties :Fit Index Criteria in Structural Equation Model—ing·The Journal of Experi mental Education,2006,74(3):267-288·
16 Schmukle S C,Hardt J·Acautionary note on Incremental FitIndices reported by LISREL·Methodology,2005,1(2):81-85·

备注/Memo

备注/Memo:
 基金项目:辽宁师范大学省心理学重点实验室开放基金资助。
更新日期/Last Update:  2009-01-16