结构方程论坛SEM-Structural·Equation·Modeling's Archiver

abroad 发表于 2011-7-3 10:17

请大家帮忙看看LISREL结果的拟合指数是否合理?

Degrees of Freedom = 477<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;Minimum Fit Function Chi-Square = 1290.85 (P = 0.0)<br /> &nbsp; &nbsp;&nbsp; &nbsp; Normal Theory Weighted Least Squares Chi-Square = 1151.96 (P = 0.0)<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp; Estimated Non-centrality Parameter (NCP) = 674.96<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;90 Percent Confidence Interval for NCP = (579.34 ; 778.26)<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;Minimum Fit Function Value = http://bbs.pinggu.org/5.16<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp; Population Discrepancy Function Value (F0) = 2.70<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp;&nbsp;90 Percent Confidence Interval for F0 = (2.32 ; 3.11)<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp; Root Mean Square Error of Approximation (RMSEA) = 0.075<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;90 Percent Confidence Interval for RMSEA = (0.070 ; 0.081)<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;P-Value for Test of Close Fit (RMSEA &lt; 0.05) = 0.00<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;Expected Cross-Validation Index (ECVI) = 5.28<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp; 90 Percent Confidence Interval for ECVI = (4.90 ; 5.69)<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp; ECVI for Saturated Model = 4.49<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp;&nbsp;ECVI for Independence Model = 31.48<br /> &nbsp; &nbsp;&nbsp;&nbsp;Chi-Square for Independence Model with 528 Degrees of Freedom = 7804.40<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp; Independence AIC = 7870.40<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp; Model AIC = 1319.96<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp;&nbsp;Saturated AIC = 1122.00<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;Independence CAIC = 8019.74<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp; Model CAIC = 1700.10<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp;&nbsp;Saturated CAIC = 3660.78<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp;&nbsp;Normed Fit Index (NFI) = 0.83<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;Non-Normed Fit Index (NNFI) = 0.88<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;Parsimony Normed Fit Index (PNFI) = 0.75<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;Comparative Fit Index (CFI) = 0.89<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;Incremental Fit Index (IFI) = 0.89<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp; Relative Fit Index (RFI) = 0.82<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp;&nbsp;Critical N (CN) = 107.86<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp; Root Mean Square Residual (RMR) = 0.37<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp;&nbsp;Standardized RMR = 0.087<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;Goodness of Fit Index (GFI) = 0.78<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp; Adjusted Goodness of Fit Index (AGFI) = 0.74<br /> &nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;&nbsp; &nbsp;Parsimony Goodness of Fit Index (PGFI) = 0.66

sopheia 发表于 2011-7-3 10:22

阁下问的是指这些指数是否反映可接受的拟合程度,还是??

database 发表于 2011-7-3 10:27

<strong> 2# <i>stephen1981</i> </strong>是的。就是想请教该结果是否是在可接受的拟合程度之内。请多多指教。

sopheia 发表于 2011-7-3 10:32

<strong> 3# <i>aku2008</i> </strong> <br /> <br /> 整体来看,这个模型拟合结果不是太好。

潜变量分析 发表于 2011-7-3 10:37

<strong> 4# <i>辛勤工作</i> </strong><br /> <br /> 那请问,这些拟合指数中哪些最需要修改呢?如何要调整的话,您看哪些需要做哪些工作呢?

统计分析 发表于 2011-7-3 10:42

不好!!!!!!!!!!

variable 发表于 2011-7-3 10:47

<strong> 5# <i>aku2008</i> </strong>其实你很多指数都是接近达标的临界值,都是差一点,建议你可以考虑以下几个方面:<br /> A. 检查MODIFICATION INDEX, 看看有没有一些变量之间变量对应的误差存在较显著的相关,尤其是那些MI值超过34的,这个好像是邱皓政和林碧芳老师书里有提到。<br /> B.若有,可考虑添加一条相关路径,但必须慎重考虑,这是否有理论意义,即是否有文献支持该路径的添加。<br /> C 你还可以观察哪些观测变量之间的相关非常之大,如大于.90,那你可能可以考虑删除其中一题,减少题目数还有助于提高GFI值。<br /> D.你还可以看看在某一潜在变量下,哪些观察变量的standardized regression weight小于.50,则可以考虑删去,但记得最好能保证相应的潜在变量含至少三个观测变量。<br /> E.再次检查你的AMOS报表中的“regression weight”和&quot;variances&quot;的数据, 若发现有不显著的路径(对于CFA的measurement model而言),可考虑将相应的观测变量删除。那些standardized regression weight有超过1的可考虑删除。<br /> F.这一步其实应该是要在前面几步之前要先考虑的,就是在你做CFA之前应该要检查一元异常值和多元异常值(univariate and multivariate outliers,如检验D值及其相伴概率是否小于.001,以及标准残差值等见Tabachnick &amp; Fidell (1996,2001,2007)),以及样本数据的是否满足正态分布的假设(如峰态值与偏态值等)。<br /> G.以上问题如果你都考虑了而且还是得到一样的结果的话,你可能还要考虑样本数和题目数的比例关系,一般最好样本数是题目数是的10倍以上为佳,5倍左右基本可以接受吧。譬如你还可以考虑收集多些样本来另外检验你的模型,或者如果你原来的样本数超多,那可以试着随机分成两组,分别做CFA检验,看看两组出来的拟合度指数如何。另外,你可能需要回到文献,考虑性别,年龄等因素可能对你模型的影响,因为有些模型可能对男生的数据模拟程度好过对女生的数据,等等。<br /> 以上意见仅供参考!祝你成功!!                   <ul class="s_clear">

wmhhotel 发表于 2011-7-3 10:52

感觉还可以,是可以接受的,在临界值附近也是可以接受的

amos 发表于 2011-7-3 10:57

<strong> 1# <i>aku2008</i> </strong><br /> <br /> CFI, NFI, GFI 都小于.90. RMSEA上限到了.081.这些指数都是不行的。

潜变量分析 发表于 2011-7-3 11:02

感谢各位的帮助!

结构方程爱好者 发表于 2011-7-3 11:07

谢谢楼主的分享

结构方程爱好者 发表于 2011-7-3 11:12

基本都是在临界值附近,感觉还行吧,但不是很好!

页: [1]

Powered by Discuz! Archiver 7.2  © 2001-2009 Comsenz Inc.