Rank: 3 Rank: 3

1^# 跳转到 » 倒序看帖

打印

字体大小: tT

angel发表于 2011-7-5 07:24 | 只看该作者

[求助]初学者，lisrel错误，怎么解决？

CFA分析

总是出现错误，路径图无法显示。

it值调整也没用。

哪里出了问题？怎么解决？

请高人帮忙看下！十分感谢

measurement model for test
observed variables: x1-x69

correlation matrix
1.000
.580 1.000
.221 -.018 1.000
.311 -.133 .035 1.000
.127 -.070 .057 .280 1.000
.160 .019 .361 -.001 .367 1.000
.237 .103 .307 .105 .221 .021 1.000
.274 .306 .198 .158 .226 -.105 -.002 1.000
.559 .399 .184 -.033 .179 .271 .000 .041 1.000
-.145 .160 .082 -.354 .159 .442 -.212 -.096 .258 1.000
-.115 -.299 .248 -.038 -.200 .162 .004 .007 -.048 .045 1.000
.039 .146 .315 -.118 .160 .131 .165 .220 -.161 -.173 .107 1.000
-.011 .016 .206 .019 .300 .363 .409 .034 -.113 .180 -.005 .421 1.000
.146 .048 .208 .133 -.012 -.114 -.036 .333 .055 -.218 .257 .456 -.136 1.000
.047 .149 -.085 -.276 .098 .308 .027 .018 .128 .378 .175 .148 .077 .100 1.000
.255 .086 .183 -.129 .116 .261 -.135 .240 .323 .208 .214 .301 -.169 .444 .553 1.000
-.042 -.268 .125 .038 -.104 .093 -.069 -.338 .153 .080 .262 -.125 -.231 -.306 .099 .086 1.000
.029 .053 .066 .019 -.131 -.291 .092 -.014 .015 -.269 -.012 .164 -.083 .198 -.130 -.242 .212 1.000
.285 .118 .170 -.052 -.054 .106 .016 -.234 .594 .396 .120 -.247 -.263 -.114 .161 .291 .640 -.044 1.000
.451 .490 .141 .061 .066 .104 .054 -.087 .389 .323 -.328 -.109 -.025 -.163 .020 .022 .236 .255 .477 1.000
.207 .541 -.084 -.130 .235 -.043 -.158 .324 .155 .357 -.389 .058 -.128 -.018 .104 .045 -.216 .163 .028 .562 1.000
.069 .128 -.056 .271 .047 -.122 -.082 .338 -.283 -.237 -.122 .131 .041 .072 .015 -.186 -.110 .510 -.514 .280 .415 1.000
.062 .268 -.188 -.050 .066 -.206 -.154 .555 .082 .001 -.150 .010 -.241 .045 .293 .155 -.134 .060 .019 .017 .410 .163 1.000
.127 .165 .079 -.129 -.057 -.093 .045 .421 .000 -.304 -.119 .312 .031 .059 .329 .214 .128 .198 -.021 -.044 -.027 .174 .567 1.000
-.062 -.102 .415 .130 -.098 -.060 .040 -.125 .029 .347 .019 .016 .225 .145 .113 .064 .119 .078 .292 .301 .014 -.013 -.199 -.042 1.000
.095 .227 .032 -.029 -.433 -.084 -.094 -.239 .105 -.080 .155 .194 -.006 -.104 -.041 -.041 .384 .496 .237 .215 -.021 .102 -.070 .216 .165 1.000
.083 .067 .095 .065 -.029 -.002 -.106 .022 .101 .242 -.014 .086 .188 .007 .087 .029 .017 .350 .179 .552 .381 .340 .182 .116 .405 .277 1.000
-.049 .186 .109 -.071 -.189 -.149 -.026 .031 .025 .302 -.113 .178 .211 .118 .018 -.034 .086 .262 .292 .575 .295 .178 .149 .122 .570 .158 .698 1.000
-.332 -.264 .134 -.301 -.044 -.014 .004 -.307 -.088 .272 .014 .108 .242 -.287 -.004 -.097 .483 .202 .329 .191 -.107 -.133 -.197 .220 .497 .394 .308 .466 1.000
.129 .232 .303 -.037 -.126 -.188 .313 -.083 -.079 -.011 -.007 .220 .181 -.040 -.261 -.157 .201 .196 .326 .366 .063 -.134 -.285 -.024 .369 .380 .109 .355 .408 1.000
.
sample size: 500

Latent Variables: com wis rug sop tre lik sin exc tra ben
Relations
x1-x9=com
x10-x16=wis
x17-x25=rug
x26-x33=sop
x34-x40=tre
x41-x47=lik
x48-x50=sin
x51-x56=exc
x57-x59=tra
x60-x69=ben

Options: nd=3 it=100 ad=off
Lisrel Output

ath Diagram
End of problem

W_A_R_N_I_N_G: Matrix to be analyzed is not positive definite,
ridge option taken with ridge constant = 0.010

measurement model for test

W_A_R_N_I_N_G: The solution has not converged after 100 iterations.
                The following solution is preliminary and is provided only
                for the purpose of tracing the source of the problem.
                Setting IT>100 may solve the problem.

Goodness of Fit Statistics

                            Degrees of Freedom = 2232
              Minimum Fit Function Chi-Square = 79691.980 (P = 0.0)
      Normal Theory Weighted Least Squares Chi-Square = 45236.891 (P = 0.0)
               Estimated Non-centrality Parameter (NCP) = 43004.891
         90 Percent Confidence Interval for NCP = (42315.658 ; 43699.492)

                       Minimum Fit Function Value = http://bbs.pinggu.org/159.703
               Population Discrepancy Function Value (F0) = 86.182
            90 Percent Confidence Interval for F0 = (84.801 ; 87.574)
             Root Mean Square Error of Approximation (RMSEA) = 0.196
            90 Percent Confidence Interval for RMSEA = (0.195 ; 0.198)
               P-Value for Test of Close Fit (RMSEA < 0.05) = 0.000

                 Expected Cross-Validation Index (ECVI) = 91.389
           90 Percent Confidence Interval for ECVI = (90.007 ; 92.781)
                         ECVI for Saturated Model = 9.679
                      ECVI for Independence Model = 124.135

    Chi-Square for Independence Model with 2346 Degrees of Freedom = 61805.444
                           Independence AIC = 61943.444
                              Model AIC = 45602.891
                             Saturated AIC = 4830.000
                          Independence CAIC = 62303.252
                              Model CAIC = 46557.164
                            Saturated CAIC = 17423.279

                         Normed Fit Index (NFI) = -0.289
                       Non-Normed Fit Index (NNFI) = -0.369
                    Parsimony Normed Fit Index (PNFI) = -0.275
                        Comparative Fit Index (CFI) = 0.0
                       Incremental Fit Index (IFI) = -0.300
                        Relative Fit Index (RFI) = -0.355

                             Critical N (CN) = 15.968

                     Root Mean Square Residual (RMR) = 0.192
                             Standardized RMR = 0.189
                       Goodness of Fit Index (GFI) = 0.276
                  Adjusted Goodness of Fit Index (AGFI) = 0.217
                  Parsimony Goodness of Fit Index (PGFI) = 0.255

Modification Indices cannot be Computed Because Iterations have not Converged

Time used: 2.309 Seconds

楼主热帖

注册会员

Rank: 2

2^#

database发表于 2011-7-5 07:29 | 只看该作者

可以尝试两种方法：

1、如果你的原始数据有缺乏值，在处理缺失值时用EM算法，不要用pairwise法

2、将迭代次数调大一点，比如 IT=20000

TOP

超级版主

Rank: 8 Rank: 8

3^#

偏最小二乘发表于 2011-7-5 07:34 | 只看该作者

根本原因是输入矩阵不正定，具体解决方法要按具体情况分析：

1、可能数据输入有误，出现极端数据。解决方法是检查数据。

2、问卷设计不合理，变量关系混乱。解决方法是重新设计问卷，合并问题。

3、问卷数据质量太差，被调查人回答问题太随意，极端值多。解决方法，检查问卷，删除不合格问卷。

4、数据变量间高度线性相关。解决方法是检查数据，删除高度相关问卷。

5、程序估计初始值不合理。解决方法是自行输入初始值。

TOP

注册会员

Rank: 2

4^#

心理学辅导发表于 2011-7-5 07:39 | 只看该作者

问题解决了，太感谢了。

TOP

注册会员

Rank: 2

5^#

dataanlysis发表于 2011-7-5 07:44 | 只看该作者

TI
DA NI=8 NO=160 MA=CM
LA
ld sd zcfz ldzczz zzczz zcbc xsml xsjl
CM FI='C:\test.cov' SY
MO NX=8 NK=3 TD=SY
LK
yl yy cz
FR LX(1,3) LX(2,3) LX(3,3) LX(4,2) LX(5,2) LX(6,1) LX(7,1) LX(8,1)
PD
OU IT=50000 AD=OFF

TI

                        Number of Input Variables  8
                        Number of Y - Variables 0
                        Number of X - Variables 8
                        Number of ETA - Variables  0
                        Number of KSI - Variables  3
                        Number of Observations 160

W_A_R_N_I_N_G: Matrix to be analyzed is not positive definite,
               ridge option taken with ridge constant = 0.001

TI

      Covariance Matrix

               ld       sd    zcfz    ldzczz    zzczz    zcbc
         -------- -------- -------- -------- -------- --------
   ld    1.00
   sd    1.00    1.00
   zcfz    -0.73    -0.73    1.00
ldzczz    -0.58    -0.58    0.97    1.00
zzczz    -0.29    -0.29    0.42    0.36    1.00
   zcbc    0.86    0.86    -0.55    -0.43    0.17    1.00
   xsml    0.33    0.33    -0.60    -0.57    -0.58    0.01
   xsjl    0.35    0.35    -0.53    -0.47    -0.84    -0.09

      Covariance Matrix

            xsml    xsjl
         -------- --------
   xsml    1.00
   xsjl    0.77    1.00

TI

Parameter Specifications

      LAMBDA-X

               yl       yy       cz
         -------- -------- --------
   ld       0       0       1
   sd       0       0       2
   zcfz       0       0       3
ldzczz       0       4       0
zzczz       0       5       0
   zcbc       6       0       0
   xsml       7       0       0
   xsjl       8       0       0

HI

               yl       yy       cz
         -------- -------- --------
   yl       0
   yy       9       0
   cz       10       11       0

      THETA-DELTA

               ld       sd    zcfz    ldzczz    zzczz    zcbc
         -------- -------- -------- -------- -------- --------
               12       13       14       15       16       17

      THETA-DELTA

            xsml    xsjl
         -------- --------
               18       19

TI

W_A_R_N_I_N_G: The solution has not converged after**** iterations.
            The following solution is preliminary and is provided only
            for the purpose of tracing the source of the problem.
            Setting IT>*** may solve the problem.

LISREL Estimates(Intermediate Solution)

      LAMBDA-X

               yl       yy       cz
         -------- -------- --------
   ld       - -       - -    1.01
   sd       - -       - -    1.01
   zcfz       - -       - -    -0.74
ldzczz       - -    0.28       - -
zzczz       - -    1.27       - -
   zcbc    0.21       - -       - -
   xsml    0.01       - -       - -
   xsjl    0.04       - -       - -

HI

               yl       yy       cz
         -------- -------- --------
   yl    1.00
   yy    0.29    1.00
   cz    4.45    -0.16    1.00

      THETA-DELTA

               ld       sd    zcfz    ldzczz    zzczz    zcbc
         -------- -------- -------- -------- -------- --------
            0.00    0.00    0.46    0.92    -0.61    0.96

      THETA-DELTA

            xsml    xsjl
         -------- --------
            1.00    1.00

LX was written to file C:\DUMP

PH was written to file C:\DUMP

TD was written to file C:\DUMP

                        Goodness of Fit Statistics

                           Degrees of Freedom = 17
            Minimum Fit Function Chi-Square = 1064.47 (P = 0.0)
      Normal Theory Weighted Least Squares Chi-Square = 516.77 (P = 0.0)
            Estimated Non-centrality Parameter (NCP) = 499.77
         90 Percent Confidence Interval for NCP = (429.27 ; 577.69)

                     Minimum Fit Function Value = http://bbs.pinggu.org/6.69
            Population Discrepancy Function Value (F0) = 3.14
            90 Percent Confidence Interval for F0 = (2.70 ; 3.63)
            Root Mean Square Error of Approximation (RMSEA) = 0.43
         90 Percent Confidence Interval for RMSEA = (0.40 ; 0.46)

-Value for Test of Close Fit (RMSEA < 0.05) = 0.00

               Expected Cross-Validation Index (ECVI) = 3.49
         90 Percent Confidence Interval for ECVI = (3.05 ; 3.98)
                     ECVI for Saturated Model = 0.45
                     ECVI for Independence Model = 9.53

   Chi-Square for Independence Model with 28 Degrees of Freedom = 1498.67
                        Independence AIC = 1514.67
                              Model AIC = 554.77
                           Saturated AIC = 72.00
                        Independence CAIC = 1547.27
                           Model CAIC = 632.20
                           Saturated CAIC = 218.71

                        Normed Fit Index (NFI) = 0.29
                     Non-Normed Fit Index (NNFI) = -0.17

arsimony Normed Fit Index (PNFI) = 0.18
                     Comparative Fit Index (CFI) = 0.29
                     Incremental Fit Index (IFI) = 0.29
                     Relative Fit Index (RFI) = -0.17

                           Critical N (CN) = 5.99

                  Root Mean Square Residual (RMR) = 0.36
                           Standardized RMR = 0.36
                     Goodness of Fit Index (GFI) = 0.55
               Adjusted Goodness of Fit Index (AGFI) = 0.050

arsimony Goodness of Fit Index (PGFI) = 0.26

Modification Indices cannot be Computed Because Iterations have not Converged

Time used: 41.688 Seconds

TOP