Message/Author |
|
daniel posted on Tuesday, August 19, 2003 - 5:06 am
|
|
|
I ran the following model and obtained the following results. How would I compare specific path coefficients for significant differences between the two groups? FOr instance, I am interested in knowing whether the values for the path from rlevel to slevel are significantly different between the two groups. How would I do that given my data is ordered-categorical? variable: names are id race gender smoke9 smoke10f smoke10s smoke11 soma9 depaff9 posaff9 interp9 soma10 depaff10 posaff10 interp10 soma11 depaff11 posaff11 interp11 totdep9 totdep10 totdep11 recep9 recep10 recep11 expose9; missing are .; idvariable is id; grouping totdep9 (0=low 1=high); usevariables are smoke9-smoke11 recep9-recep11 gender race expose9; categorical are smoke9-smoke11 recep9-recep11; define: cut smoke9-smoke11 (0 2); cut totdep9 (16); Analysis: Type=meanstructure; iterations = 20000; model: slevel by smoke9-smoke11@1; strend by smoke9@0 smoke10f@1 smoke10s@1.422 smoke11@2.099; rlevel by recep9-recep11@1; rtrend by recep9@0 recep10@1 recep11@2; strend on slevel rlevel rtrend; slevel on rlevel; rtrend on slevel; strend rtrend on gender-expose9; slevel rlevel on gender-expose9; [smoke9$1 smoke10f$1 smoke10s$1 smoke11$1] (4); [smoke9$2 smoke10f$2 smoke10s$2 smoke11$2] (5); [slevel@0 strend]; {smoke9@1 smoke10f smoke10s smoke11}; [recep9$1 recep10$1 recep11$1] (1); [recep9$2 recep10$2 recep11$2] (2); [recep9$3 recep10$3 recep11$3] (3); [rlevel@0 rtrend]; {recep9@1 recep10 recep11}; model high: strend@.067; output: tech4 standardized cinterval; INPUT READING TERMINATED NORMALLY SUMMARY OF ANALYSIS Number of groups 2 Number of observations Group LOW 668 Group HIGH 290 Number of y-variables 7 Number of x-variables 3 Number of continuous latent variables 4 Observed variables in the analysis SMOKE9 SMOKE10F SMOKE10S SMOKE11 RECEP9 RECEP10 RECEP11 GENDER RACE EXPOSE9 Grouping variable TOTDEP9 ID variable ID Categorical variables SMOKE9 SMOKE10F SMOKE10S SMOKE11 RECEP9 RECEP10 RECEP11 Continuous latent variables in the analysis SLEVEL STREND RLEVEL RTREND Estimator WLSMV Maximum number of iterations 20000 Convergence criterion 0.500D-04 Parameterization DELTA Input data file(s) z:\sas\adlgm.dat Input data format FREE THE MODEL ESTIMATION TERMINATED NORMALLY TESTS OF MODEL FIT Chi-Square Test of Model Fit Value 72.855* Degrees of Freedom 22** P-Value 0.0000 * The chi-square value for MLM, MLR, MLMV, MUMLM, MUMLMV, WLSM and WLSMV cannot be used for chi-square difference tests. MLM and MLR chi-square difference testing is described on page 360 in the Mplus User's Guide. ** The degrees of freedom for MLMV and WLSMV are estimated according to formula 110 (page 358) in the Mplus User's Guide. Chi-Square Test of Model Fit for the Baseline Model Value 19692.277 Degrees of Freedom 31 P-Value 0.0000 CFI/TLI CFI 0.997 TLI 0.996 RMSEA (Root Mean Square Error Of Approximation) Estimate 0.069 WRMR (Weighted Root Mean Square Residual) Value 1.267 MODEL RESULTS Estimates S.E. Est./S.E. Std StdYX Group LOW SLEVEL BY SMOKE9 1.000 0.000 0.000 1.093 0.988 SMOKE10F 1.000 0.000 0.000 1.093 1.001 SMOKE10S 1.000 0.000 0.000 1.093 1.005 SMOKE11 1.000 0.000 0.000 1.093 1.013 STREND BY SMOKE9 0.000 0.000 0.000 0.000 0.000 SMOKE10F 1.000 0.000 0.000 0.202 0.185 SMOKE10S 1.422 0.000 0.000 0.287 0.264 SMOKE11 2.099 0.000 0.000 0.424 0.393 RLEVEL BY RECEP9 1.000 0.000 0.000 0.762 0.732 RECEP10 1.000 0.000 0.000 0.762 0.733 RECEP11 1.000 0.000 0.000 0.762 0.734 RTREND BY RECEP9 0.000 0.000 0.000 0.000 0.000 RECEP10 1.000 0.000 0.000 0.209 0.201 RECEP11 2.000 0.000 0.000 0.418 0.402 STREND ON SLEVEL -0.058 0.025 -2.362 -0.315 -0.315 RLEVEL 0.039 0.040 0.975 0.149 0.149 RTREND 0.380 0.251 1.514 0.393 0.393 SLEVEL ON RLEVEL 0.463 0.094 4.920 0.323 0.323 RTREND ON SLEVEL -0.013 0.030 -0.429 -0.067 -0.067 STREND ON GENDER 0.037 0.040 0.929 0.183 0.091 RACE -0.060 0.045 -1.340 -0.298 -0.138 EXPOSE9 -0.004 0.043 -0.089 -0.019 -0.010 RTREND ON GENDER -0.025 0.049 -0.508 -0.118 -0.059 RACE 0.051 0.051 0.990 0.244 0.113 EXPOSE9 -0.028 0.057 -0.486 -0.132 -0.066 SLEVEL ON GENDER 0.051 0.107 0.474 0.046 0.023 RACE 0.074 0.110 0.673 0.067 0.031 EXPOSE9 0.738 0.115 6.419 0.675 0.338 RLEVEL ON GENDER -0.411 0.082 -4.997 -0.539 -0.270 RACE -0.081 0.087 -0.932 -0.106 -0.049 EXPOSE9 0.425 0.083 5.139 0.557 0.279 Intercepts SLEVEL 0.000 0.000 0.000 0.000 0.000 STREND 0.000 0.000 0.000 0.000 0.000 RLEVEL 0.000 0.000 0.000 0.000 0.000 RTREND 0.000 0.000 0.000 0.000 0.000 Thresholds SMOKE9$1 0.561 0.072 7.772 0.561 0.561 SMOKE9$2 2.061 0.092 22.427 2.061 2.061 SMOKE10F$1 0.561 0.072 7.772 0.561 0.561 SMOKE10F$2 2.061 0.092 22.427 2.061 2.061 SMOKE10S$1 0.561 0.072 7.772 0.561 0.561 SMOKE10S$2 2.061 0.092 22.427 2.061 2.061 SMOKE11$1 0.561 0.072 7.772 0.561 0.561 SMOKE11$2 2.061 0.092 22.427 2.061 2.061 RECEP9$1 -0.872 0.059 -14.677 -0.872 -0.872 RECEP9$2 0.111 0.058 1.917 0.111 0.111 RECEP9$3 0.444 0.059 7.481 0.444 0.444 RECEP10$1 -0.872 0.059 -14.677 -0.872 -0.872 RECEP10$2 0.111 0.058 1.917 0.111 0.111 RECEP10$3 0.444 0.059 7.481 0.444 0.444 RECEP11$1 -0.872 0.059 -14.677 -0.872 -0.872 RECEP11$2 0.111 0.058 1.917 0.111 0.111 RECEP11$3 0.444 0.059 7.481 0.444 0.444 Residual Variances SLEVEL 0.865 0.051 17.125 0.724 0.724 STREND 0.030 0.013 2.250 0.727 0.727 RLEVEL 0.498 0.030 16.435 0.857 0.857 RTREND 0.042 0.015 2.770 0.974 0.974 Scales SMOKE9 1.000 0.000 0.000 1.000 1.000 SMOKE10F 1.000 0.000 0.000 1.000 1.000 SMOKE10S 1.000 0.000 0.000 1.000 1.000 SMOKE11 1.000 0.000 0.000 1.000 1.000 RECEP9 1.000 0.000 0.000 1.000 1.000 RECEP10 1.000 0.000 0.000 1.000 1.000 RECEP11 1.000 0.000 0.000 1.000 1.000 Group HIGH SLEVEL BY SMOKE9 1.000 0.000 0.000 1.073 0.944 SMOKE10F 1.000 0.000 0.000 1.073 0.859 SMOKE10S 1.000 0.000 0.000 1.073 0.793 SMOKE11 1.000 0.000 0.000 1.073 0.688 STREND BY SMOKE9 0.000 0.000 0.000 0.000 0.000 SMOKE10F 1.000 0.000 0.000 0.355 0.284 SMOKE10S 1.422 0.000 0.000 0.505 0.373 SMOKE11 2.099 0.000 0.000 0.745 0.478 RLEVEL BY RECEP9 1.000 0.000 0.000 0.804 0.785 RECEP10 1.000 0.000 0.000 0.804 0.625 RECEP11 1.000 0.000 0.000 0.804 0.719 RTREND BY RECEP9 0.000 0.000 0.000 0.000 0.000 RECEP10 1.000 0.000 0.000 0.356 0.276 RECEP11 2.000 0.000 0.000 0.711 0.636 STREND ON SLEVEL 0.197 0.094 2.101 0.596 0.596 RLEVEL -0.149 0.076 -1.968 -0.336 -0.336 RTREND 0.532 0.246 2.162 0.533 0.533 SLEVEL ON RLEVEL 0.550 0.099 5.555 0.412 0.412 RTREND ON SLEVEL -0.065 0.050 -1.299 -0.197 -0.197 STREND ON GENDER 0.121 0.089 1.359 0.340 0.164 RACE -0.120 0.094 -1.280 -0.339 -0.163 EXPOSE9 -0.041 0.115 -0.354 -0.114 -0.055 RTREND ON GENDER -0.110 0.073 -1.502 -0.309 -0.149 RACE 0.109 0.078 1.401 0.305 0.147 EXPOSE9 0.199 0.089 2.223 0.559 0.271 SLEVEL ON GENDER -0.435 0.143 -3.042 -0.406 -0.196 RACE 0.305 0.162 1.888 0.285 0.137 EXPOSE9 0.769 0.188 4.098 0.717 0.348 RLEVEL ON GENDER -0.104 0.134 -0.771 -0.129 -0.062 RACE -0.211 0.131 -1.604 -0.262 -0.126 EXPOSE9 0.411 0.131 3.125 0.511 0.248 Intercepts SLEVEL 0.000 0.000 0.000 0.000 0.000 STREND -0.010 0.117 -0.087 -0.029 -0.029 RLEVEL 0.000 0.000 0.000 0.000 0.000 RTREND -0.058 0.076 -0.754 -0.162 -0.162 Thresholds SMOKE9$1 0.561 0.072 7.772 0.561 0.561 SMOKE9$2 2.061 0.092 22.427 2.061 2.061 SMOKE10F$1 0.561 0.072 7.772 0.561 0.561 SMOKE10F$2 2.061 0.092 22.427 2.061 2.061 SMOKE10S$1 0.561 0.072 7.772 0.561 0.561 SMOKE10S$2 2.061 0.092 22.427 2.061 2.061 SMOKE11$1 0.561 0.072 7.772 0.561 0.561 SMOKE11$2 2.061 0.092 22.427 2.061 2.061 RECEP9$1 -0.872 0.059 -14.677 -0.872 -0.872 RECEP9$2 0.111 0.058 1.917 0.111 0.111 RECEP9$3 0.444 0.059 7.481 0.444 0.444 RECEP10$1 -0.872 0.059 -14.677 -0.872 -0.872 RECEP10$2 0.111 0.058 1.917 0.111 0.111 RECEP10$3 0.444 0.059 7.481 0.444 0.444 RECEP11$1 -0.872 0.059 -14.677 -0.872 -0.872 RECEP11$2 0.111 0.058 1.917 0.111 0.111 RECEP11$3 0.444 0.059 7.481 0.444 0.444 Residual Variances SLEVEL 0.679 0.065 10.433 0.590 0.590 STREND 0.067 0.000 0.000 0.532 0.532 RLEVEL 0.598 0.072 8.285 0.926 0.926 RTREND 0.115 0.051 2.272 0.909 0.909 Scales SMOKE9 1.000 0.000 0.000 1.000 1.000 SMOKE10F 0.919 0.061 15.134 0.919 0.919 SMOKE10S 0.841 0.068 12.448 0.841 0.841 SMOKE11 0.717 0.071 10.133 0.717 0.717 RECEP9 1.000 0.000 0.000 1.000 1.000 RECEP10 0.796 0.080 9.923 0.796 0.796 RECEP11 0.940 0.072 13.084 0.940 0.940 R-SQUARE Group LOW Observed Residual Variable Variance R-Square SMOKE9 0.028 0.977 SMOKE10F 0.093 0.922 SMOKE10S 0.097 0.918 SMOKE11 0.074 0.936 RECEP9 0.502 0.536 RECEP10 0.465 0.569 RECEP11 0.343 0.682 Latent Variable R-Square SLEVEL 0.276 STREND 0.273 RLEVEL 0.143 RTREND 0.026 Group HIGH Observed Residual Variable Variance R-Square SMOKE9 0.140 0.892 SMOKE10F 0.023 0.985 SMOKE10S 0.053 0.971 SMOKE11 0.177 0.927 RECEP9 0.402 0.617 RECEP10 0.906 0.453 RECEP11 0.145 0.884 Latent Variable R-Square SLEVEL 0.410 STREND 0.468 RLEVEL 0.074 RTREND 0.091 CONFIDENCE INTERVALS OF MODEL RESULTS Lower 1% Lower 5% Estimates Upper 5% Upper 1% Group LOW SLEVEL BY SMOKE9 1.000 1.000 1.000 1.000 1.000 SMOKE10F 1.000 1.000 1.000 1.000 1.000 SMOKE10S 1.000 1.000 1.000 1.000 1.000 SMOKE11 1.000 1.000 1.000 1.000 1.000 STREND BY SMOKE9 0.000 0.000 0.000 0.000 0.000 SMOKE10F 1.000 1.000 1.000 1.000 1.000 SMOKE10S 1.422 1.422 1.422 1.422 1.422 SMOKE11 2.099 2.099 2.099 2.099 2.099 RLEVEL BY RECEP9 1.000 1.000 1.000 1.000 1.000 RECEP10 1.000 1.000 1.000 1.000 1.000 RECEP11 1.000 1.000 1.000 1.000 1.000 RTREND BY RECEP9 0.000 0.000 0.000 0.000 0.000 RECEP10 1.000 1.000 1.000 1.000 1.000 RECEP11 2.000 2.000 2.000 2.000 2.000 STREND ON SLEVEL -0.122 -0.107 -0.058 -0.010 0.005 RLEVEL -0.065 -0.040 0.039 0.119 0.143 RTREND -0.266 -0.112 0.380 0.871 1.026 SLEVEL ON RLEVEL 0.221 0.279 0.463 0.648 0.706 RTREND ON SLEVEL -0.090 -0.071 -0.013 0.046 0.064 STREND ON GENDER -0.065 -0.041 0.037 0.115 0.139 RACE -0.176 -0.148 -0.060 0.028 0.056 EXPOSE9 -0.116 -0.089 -0.004 0.081 0.108 RTREND ON GENDER -0.150 -0.120 -0.025 0.071 0.101 RACE -0.082 -0.050 0.051 0.152 0.184 EXPOSE9 -0.174 -0.139 -0.028 0.084 0.119 SLEVEL ON GENDER -0.225 -0.159 0.051 0.261 0.327 RACE -0.209 -0.141 0.074 0.289 0.356 EXPOSE9 0.442 0.513 0.738 0.963 1.034 RLEVEL ON GENDER -0.623 -0.572 -0.411 -0.250 -0.199 RACE -0.304 -0.251 -0.081 0.089 0.142 EXPOSE9 0.212 0.263 0.425 0.587 0.638 Intercepts SLEVEL 0.000 0.000 0.000 0.000 0.000 STREND 0.000 0.000 0.000 0.000 0.000 RLEVEL 0.000 0.000 0.000 0.000 0.000 RTREND 0.000 0.000 0.000 0.000 0.000 Thresholds SMOKE9$1 0.375 0.419 0.561 0.702 0.747 SMOKE9$2 1.824 1.881 2.061 2.241 2.297 SMOKE10F 0.375 0.419 0.561 0.702 0.747 SMOKE10F 1.824 1.881 2.061 2.241 2.297 SMOKE10S 0.375 0.419 0.561 0.702 0.747 SMOKE10S 1.824 1.881 2.061 2.241 2.297 SMOKE11$ 0.375 0.419 0.561 0.702 0.747 SMOKE11$ 1.824 1.881 2.061 2.241 2.297 RECEP9$1 -1.025 -0.988 -0.872 -0.755 -0.719 RECEP9$2 -0.038 -0.003 0.111 0.224 0.259 RECEP9$3 0.291 0.328 0.444 0.561 0.597 RECEP10$ -1.025 -0.988 -0.872 -0.755 -0.719 RECEP10$ -0.038 -0.003 0.111 0.224 0.259 RECEP10$ 0.291 0.328 0.444 0.561 0.597 RECEP11$ -1.025 -0.988 -0.872 -0.755 -0.719 RECEP11$ -0.038 -0.003 0.111 0.224 0.259 RECEP11$ 0.291 0.328 0.444 0.561 0.597 Residual Variances SLEVEL 0.735 0.766 0.865 0.964 0.995 STREND -0.004 0.004 0.030 0.055 0.064 RLEVEL 0.420 0.439 0.498 0.557 0.576 RTREND 0.003 0.012 0.042 0.072 0.082 Scales SMOKE9 1.000 1.000 1.000 1.000 1.000 SMOKE10F 1.000 1.000 1.000 1.000 1.000 SMOKE10S 1.000 1.000 1.000 1.000 1.000 SMOKE11 1.000 1.000 1.000 1.000 1.000 RECEP9 1.000 1.000 1.000 1.000 1.000 RECEP10 1.000 1.000 1.000 1.000 1.000 RECEP11 1.000 1.000 1.000 1.000 1.000 Group HIGH SLEVEL BY SMOKE9 1.000 1.000 1.000 1.000 1.000 SMOKE10F 1.000 1.000 1.000 1.000 1.000 SMOKE10S 1.000 1.000 1.000 1.000 1.000 SMOKE11 1.000 1.000 1.000 1.000 1.000 STREND BY SMOKE9 0.000 0.000 0.000 0.000 0.000 SMOKE10F 1.000 1.000 1.000 1.000 1.000 SMOKE10S 1.422 1.422 1.422 1.422 1.422 SMOKE11 2.099 2.099 2.099 2.099 2.099 RLEVEL BY RECEP9 1.000 1.000 1.000 1.000 1.000 RECEP10 1.000 1.000 1.000 1.000 1.000 RECEP11 1.000 1.000 1.000 1.000 1.000 RTREND BY RECEP9 0.000 0.000 0.000 0.000 0.000 RECEP10 1.000 1.000 1.000 1.000 1.000 RECEP11 2.000 2.000 2.000 2.000 2.000 STREND ON SLEVEL -0.045 0.013 0.197 0.381 0.439 RLEVEL -0.343 -0.297 -0.149 -0.001 0.046 RTREND -0.102 0.050 0.532 1.013 1.165 SLEVEL ON RLEVEL 0.295 0.356 0.550 0.744 0.805 RTREND ON SLEVEL -0.195 -0.164 -0.065 0.033 0.064 STREND ON GENDER -0.108 -0.053 0.121 0.295 0.350 RACE -0.362 -0.304 -0.120 0.064 0.122 EXPOSE9 -0.336 -0.265 -0.041 0.184 0.255 RTREND ON GENDER -0.298 -0.253 -0.110 0.033 0.079 RACE -0.091 -0.043 0.109 0.261 0.308 EXPOSE9 -0.032 0.024 0.199 0.374 0.429 SLEVEL ON GENDER -0.804 -0.715 -0.435 -0.155 -0.067 RACE -0.111 -0.012 0.305 0.622 0.722 EXPOSE9 0.286 0.401 0.769 1.137 1.253 RLEVEL ON GENDER -0.450 -0.367 -0.104 0.160 0.243 RACE -0.549 -0.468 -0.211 0.047 0.128 EXPOSE9 0.072 0.153 0.411 0.668 0.749 Intercepts SLEVEL 0.000 0.000 0.000 0.000 0.000 STREND -0.312 -0.240 -0.010 0.220 0.292 RLEVEL 0.000 0.000 0.000 0.000 0.000 RTREND -0.254 -0.207 -0.058 0.092 0.139 Thresholds SMOKE9$1 0.375 0.419 0.561 0.702 0.747 SMOKE9$2 1.824 1.881 2.061 2.241 2.297 SMOKE10F 0.375 0.419 0.561 0.702 0.747 SMOKE10F 1.824 1.881 2.061 2.241 2.297 SMOKE10S 0.375 0.419 0.561 0.702 0.747 SMOKE10S 1.824 1.881 2.061 2.241 2.297 SMOKE11$ 0.375 0.419 0.561 0.702 0.747 SMOKE11$ 1.824 1.881 2.061 2.241 2.297 RECEP9$1 -1.025 -0.988 -0.872 -0.755 -0.719 RECEP9$2 -0.038 -0.003 0.111 0.224 0.259 RECEP9$3 0.291 0.328 0.444 0.561 0.597 RECEP10$ -1.025 -0.988 -0.872 -0.755 -0.719 RECEP10$ -0.038 -0.003 0.111 0.224 0.259 RECEP10$ 0.291 0.328 0.444 0.561 0.597 RECEP11$ -1.025 -0.988 -0.872 -0.755 -0.719 RECEP11$ -0.038 -0.003 0.111 0.224 0.259 RECEP11$ 0.291 0.328 0.444 0.561 0.597 Residual Variances SLEVEL 0.512 0.552 0.679 0.807 0.847 STREND 0.067 0.067 0.067 0.067 0.067 RLEVEL 0.412 0.457 0.598 0.740 0.784 RTREND -0.015 0.016 0.115 0.214 0.245 Scales SMOKE9 1.000 1.000 1.000 1.000 1.000 SMOKE10F 0.762 0.800 0.919 1.038 1.075 SMOKE10S 0.667 0.708 0.841 0.973 1.015 SMOKE11 0.535 0.578 0.717 0.855 0.899 RECEP9 1.000 1.000 1.000 1.000 1.000 RECEP10 0.589 0.638 0.796 0.953 1.002 RECEP11 0.755 0.799 0.940 1.081 1.125 |
|
|
You can use the WLS estimator and do a chi-square difference test between the model with the path constrained and the mode with the path not constrained. |
|
Daniel posted on Monday, August 25, 2003 - 7:31 am
|
|
|
In the above analysis, there was a significant effect for strend on rtrend in the high but not the low depression group. However, when I constrained the paths to equality and compared the model with constrained paths to the model above with freely estimated path coefficients for each group, the WLS chi-square difference was not significant. DOes that mean that there is no interaction? How meaningful are the above results, given there are significant differences within each group? Or is this finding completely invalidated because the chi-square difference was not significant? |
|
|
If the chi-square difference was not significant, this means there is not an interaction. The fact that the path was significant in one group and not the other could be a matter of power. Was the sample size smaller in the group without significance? Also, was the constrained path significant? |
|
daniel posted on Monday, August 25, 2003 - 11:32 am
|
|
|
The sample size was smaller for the group with significance. Group LOW 668 (not significant) Group HIGH 290 (significant) The constrained path was significant. |
|
|
This can occur. I would need to see full outputs from the constrained and not constraied runs to comment further. |
|
daniel posted on Tuesday, August 26, 2003 - 11:18 am
|
|
|
This is the unconstrained model with WLS estimation INPUT INSTRUCTIONS Data: File is z:\sas\adlgm.dat; variable: names are id race gender smoke9 smoke10f smoke10s smoke11 soma9 depaff9 posaff9 interp9 soma10 depaff10 posaff10 interp10 soma11 depaff11 posaff11 interp11 totdep9 totdep10 totdep11 recep9 recep10 recep11 expose9; missing are .; idvariable is id; grouping is totdep9 (0=low 1=high); usevariables are smoke9-smoke11 recep9-recep11 gender race expose9; categorical are smoke9-smoke11 recep9-recep11; define: cut smoke9-smoke11 (0 2); define: cut totdep9 (16); Analysis: Type=meanstructure; estimator=wls; iterations = 20000; model: slevel by smoke9-smoke11@1; strend by smoke9@0 smoke10f@1 smoke10s@1.422 smoke11@2.099; rlevel by recep9-recep11@1; rtrend by recep9@0 recep10@1 recep11@2; strend on slevel; strend on rlevel; strend on rtrend; slevel on rlevel; rtrend on slevel; strend rtrend on gender-expose9; slevel on gender race; slevel on expose9; rlevel on gender race; rlevel on expose9; [smoke9$1 smoke10f$1 smoke10s$1 smoke11$1] (4); [smoke9$2 smoke10f$2 smoke10s$2 smoke11$2] (5); [slevel@0 strend]; {smoke9@1 smoke10f smoke10s smoke11}; [recep9$1 recep10$1 recep11$1] (1); [recep9$2 recep10$2 recep11$2] (2); [recep9$3 recep10$3 recep11$3] (3); [rlevel@0 rtrend]; {recep9@1 recep10 recep11}; model high: strend@.067; INPUT READING TERMINATED NORMALLY SUMMARY OF ANALYSIS Number of groups 2 Number of observations Group LOW 668 Group HIGH 290 Number of y-variables 7 Number of x-variables 3 Number of continuous latent variables 4 Observed variables in the analysis SMOKE9 SMOKE10F SMOKE10S SMOKE11 RECEP9 RECEP10 RECEP11 GENDER RACE EXPOSE9 Grouping variable TOTDEP9 ID variable ID Categorical variables SMOKE9 SMOKE10F SMOKE10S SMOKE11 RECEP9 RECEP10 RECEP11 Continuous latent variables in the analysis SLEVEL STREND RLEVEL RTREND Estimator WLS Maximum number of iterations 20000 Convergence criterion 0.500D-04 Parameterization DELTA Input data file(s) z:\sas\adlgm.dat Input data format FREE THE MODEL ESTIMATION TERMINATED NORMALLY TESTS OF MODEL FIT Chi-Square Test of Model Fit Value 211.479 Degrees of Freedom 65 P-Value 0.0000 Chi-Square Test of Model Fit for the Baseline Model Value 43883.996 Degrees of Freedom 84 P-Value 0.0000 CFI/TLI CFI 0.997 TLI 0.996 RMSEA (Root Mean Square Error Of Approximation) Estimate 0.069 MODEL RESULTS Estimates S.E. Est./S.E. Group LOW SLEVEL BY SMOKE9 1.000 0.000 0.000 SMOKE10F 1.000 0.000 0.000 SMOKE10S 1.000 0.000 0.000 SMOKE11 1.000 0.000 0.000 STREND BY SMOKE9 0.000 0.000 0.000 SMOKE10F 1.000 0.000 0.000 SMOKE10S 1.422 0.000 0.000 SMOKE11 2.099 0.000 0.000 RLEVEL BY RECEP9 1.000 0.000 0.000 RECEP10 1.000 0.000 0.000 RECEP11 1.000 0.000 0.000 RTREND BY RECEP9 0.000 0.000 0.000 RECEP10 1.000 0.000 0.000 RECEP11 2.000 0.000 0.000 STREND ON SLEVEL -0.009 0.020 -0.458 RLEVEL 0.014 0.040 0.358 RTREND 0.171 0.282 0.608 SLEVEL ON RLEVEL 0.484 0.085 5.667 RTREND ON SLEVEL -0.007 0.029 -0.230 SLEVEL ON GENDER -0.051 0.097 -0.523 RACE -0.036 0.105 -0.342 EXPOSE9 0.577 0.100 5.744 STREND ON GENDER 0.133 0.034 3.972 RACE 0.046 0.039 1.169 EXPOSE9 0.042 0.031 1.374 RTREND ON GENDER -0.030 0.040 -0.742 RACE 0.024 0.047 0.502 EXPOSE9 -0.014 0.049 -0.297 RLEVEL ON GENDER -0.403 0.075 -5.387 RACE -0.085 0.081 -1.040 EXPOSE9 0.414 0.076 5.415 Intercepts SLEVEL 0.000 0.000 0.000 STREND 0.000 0.000 0.000 RLEVEL 0.000 0.000 0.000 RTREND 0.000 0.000 0.000 Thresholds SMOKE9$1 0.570 0.066 8.583 SMOKE9$2 1.803 0.080 22.456 SMOKE10F$1 0.570 0.066 8.583 SMOKE10F$2 1.803 0.080 22.456 SMOKE10S$1 0.570 0.066 8.583 SMOKE10S$2 1.803 0.080 22.456 SMOKE11$1 0.570 0.066 8.583 SMOKE11$2 1.803 0.080 22.456 RECEP9$1 -0.839 0.057 -14.766 RECEP9$2 0.133 0.055 2.424 RECEP9$3 0.478 0.057 8.430 RECEP10$1 -0.839 0.057 -14.766 RECEP10$2 0.133 0.055 2.424 RECEP10$3 0.478 0.057 8.430 RECEP11$1 -0.839 0.057 -14.766 RECEP11$2 0.133 0.055 2.424 RECEP11$3 0.478 0.057 8.430 Residual Variances SLEVEL 0.835 0.048 17.370 STREND 0.005 0.010 0.529 RLEVEL 0.529 0.029 17.975 RTREND 0.035 0.014 2.466 Scales SMOKE9 1.000 0.000 0.000 SMOKE10F 1.000 0.000 0.000 SMOKE10S 1.000 0.000 0.000 SMOKE11 1.000 0.000 0.000 RECEP9 1.000 0.000 0.000 RECEP10 1.000 0.000 0.000 RECEP11 1.000 0.000 0.000 Group HIGH SLEVEL BY SMOKE9 1.000 0.000 0.000 SMOKE10F 1.000 0.000 0.000 SMOKE10S 1.000 0.000 0.000 SMOKE11 1.000 0.000 0.000 STREND BY SMOKE9 0.000 0.000 0.000 SMOKE10F 1.000 0.000 0.000 SMOKE10S 1.422 0.000 0.000 SMOKE11 2.099 0.000 0.000 RLEVEL BY RECEP9 1.000 0.000 0.000 RECEP10 1.000 0.000 0.000 RECEP11 1.000 0.000 0.000 RTREND BY RECEP9 0.000 0.000 0.000 RECEP10 1.000 0.000 0.000 RECEP11 2.000 0.000 0.000 STREND ON SLEVEL 0.201 0.087 2.315 RLEVEL -0.095 0.072 -1.311 RTREND 0.525 0.194 2.712 SLEVEL ON RLEVEL 0.599 0.091 6.580 RTREND ON SLEVEL -0.033 0.044 -0.763 SLEVEL ON GENDER -0.416 0.125 -3.326 RACE 0.299 0.149 2.000 EXPOSE9 0.523 0.120 4.375 STREND ON GENDER 0.112 0.081 1.378 RACE -0.125 0.082 -1.516 EXPOSE9 0.060 0.088 0.683 RTREND ON GENDER -0.109 0.070 -1.543 RACE 0.047 0.076 0.622 EXPOSE9 0.233 0.074 3.147 RLEVEL ON GENDER 0.030 0.102 0.294 RACE -0.128 0.117 -1.088 EXPOSE9 0.398 0.107 3.709 Intercepts SLEVEL 0.000 0.000 0.000 STREND -0.022 0.077 -0.284 RLEVEL 0.000 0.000 0.000 RTREND -0.124 0.066 -1.875 Thresholds SMOKE9$1 0.570 0.066 8.583 SMOKE9$2 1.803 0.080 22.456 SMOKE10F$1 0.570 0.066 8.583 SMOKE10F$2 1.803 0.080 22.456 SMOKE10S$1 0.570 0.066 8.583 SMOKE10S$2 1.803 0.080 22.456 SMOKE11$1 0.570 0.066 8.583 SMOKE11$2 1.803 0.080 22.456 RECEP9$1 -0.839 0.057 -14.766 RECEP9$2 0.133 0.055 2.424 RECEP9$3 0.478 0.057 8.430 RECEP10$1 -0.839 0.057 -14.766 RECEP10$2 0.133 0.055 2.424 RECEP10$3 0.478 0.057 8.430 RECEP11$1 -0.839 0.057 -14.766 RECEP11$2 0.133 0.055 2.424 RECEP11$3 0.478 0.057 8.430 Residual Variances SLEVEL 0.744 0.063 11.745 STREND 0.067 0.000 0.000 RLEVEL 0.586 0.058 10.079 RTREND 0.136 0.046 2.983 Scales SMOKE9 1.000 0.000 0.000 SMOKE10F 0.844 0.049 17.384 SMOKE10S 0.773 0.056 13.872 SMOKE11 0.672 0.061 11.030 RECEP9 1.000 0.000 0.000 RECEP10 0.940 0.061 15.458 RECEP11 0.870 0.060 14.507 THis is the constrained model INPUT INSTRUCTIONS Data: File is z:\sas\adlgm.dat; variable: names are id race gender smoke9 smoke10f smoke10s smoke11 soma9 depaff9 posaff9 interp9 soma10 depaff10 posaff10 interp10 soma11 depaff11 posaff11 interp11 totdep9 totdep10 totdep11 recep9 recep10 recep11 expose9; missing are .; idvariable is id; grouping is totdep9 (0=low 1=high); usevariables are smoke9-smoke11 recep9-recep11 gender race expose9; categorical are smoke9-smoke11 recep9-recep11; define: cut smoke9-smoke11 (0 2); define: cut totdep9 (16); Analysis: Type=meanstructure; estimator=wls; iterations = 20000; model: slevel by smoke9-smoke11@1; strend by smoke9@0 smoke10f@1 smoke10s@1.422 smoke11@2.099; rlevel by recep9-recep11@1; rtrend by recep9@0 recep10@1 recep11@2; strend on slevel; strend on rlevel; strend on rtrend (6); slevel on rlevel; rtrend on slevel; strend rtrend on gender-expose9; slevel on gender race; slevel on expose9; rlevel on gender race; rlevel on expose9; [smoke9$1 smoke10f$1 smoke10s$1 smoke11$1] (4); [smoke9$2 smoke10f$2 smoke10s$2 smoke11$2] (5); [slevel@0 strend]; {smoke9@1 smoke10f smoke10s smoke11}; [recep9$1 recep10$1 recep11$1] (1); [recep9$2 recep10$2 recep11$2] (2); [recep9$3 recep10$3 recep11$3] (3); [rlevel@0 rtrend]; {recep9@1 recep10 recep11}; model high: strend@.067; INPUT READING TERMINATED NORMALLY SUMMARY OF ANALYSIS Number of groups 2 Number of observations Group LOW 668 Group HIGH 290 Number of y-variables 7 Number of x-variables 3 Number of continuous latent variables 4 Observed variables in the analysis SMOKE9 SMOKE10F SMOKE10S SMOKE11 RECEP9 RECEP10 RECEP11 GENDER RACE EXPOSE9 Grouping variable TOTDEP9 ID variable ID Categorical variables SMOKE9 SMOKE10F SMOKE10S SMOKE11 RECEP9 RECEP10 RECEP11 Continuous latent variables in the analysis SLEVEL STREND RLEVEL RTREND Estimator WLS Maximum number of iterations 20000 Convergence criterion 0.500D-04 Parameterization DELTA Input data file(s) z:\sas\adlgm.dat Input data format FREE THE MODEL ESTIMATION TERMINATED NORMALLY TESTS OF MODEL FIT Chi-Square Test of Model Fit Value 212.558 Degrees of Freedom 66 P-Value 0.0000 Chi-Square Test of Model Fit for the Baseline Model Value 43883.996 Degrees of Freedom 84 P-Value 0.0000 CFI/TLI CFI 0.997 TLI 0.996 RMSEA (Root Mean Square Error Of Approximation) Estimate 0.068 MODEL RESULTS Estimates S.E. Est./S.E. Group LOW SLEVEL BY SMOKE9 1.000 0.000 0.000 SMOKE10F 1.000 0.000 0.000 SMOKE10S 1.000 0.000 0.000 SMOKE11 1.000 0.000 0.000 STREND BY SMOKE9 0.000 0.000 0.000 SMOKE10F 1.000 0.000 0.000 SMOKE10S 1.422 0.000 0.000 SMOKE11 2.099 0.000 0.000 RLEVEL BY RECEP9 1.000 0.000 0.000 RECEP10 1.000 0.000 0.000 RECEP11 1.000 0.000 0.000 RTREND BY RECEP9 0.000 0.000 0.000 RECEP10 1.000 0.000 0.000 RECEP11 2.000 0.000 0.000 STREND ON SLEVEL 0.004 0.024 0.167 RLEVEL -0.006 0.036 -0.160 RTREND 0.477 0.162 2.936 SLEVEL ON RLEVEL 0.499 0.082 6.085 RTREND ON SLEVEL -0.017 0.028 -0.620 SLEVEL ON GENDER -0.049 0.097 -0.501 RACE -0.030 0.105 -0.289 EXPOSE9 0.572 0.100 5.708 STREND ON GENDER 0.137 0.038 3.620 RACE 0.040 0.045 0.876 EXPOSE9 0.047 0.037 1.269 RTREND ON GENDER -0.032 0.040 -0.799 RACE 0.023 0.047 0.489 EXPOSE9 -0.006 0.048 -0.117 RLEVEL ON GENDER -0.402 0.075 -5.373 RACE -0.086 0.081 -1.063 EXPOSE9 0.412 0.076 5.389 Intercepts SLEVEL 0.000 0.000 0.000 STREND 0.000 0.000 0.000 RLEVEL 0.000 0.000 0.000 RTREND 0.000 0.000 0.000 Thresholds SMOKE9$1 0.572 0.066 8.609 SMOKE9$2 1.803 0.080 22.489 SMOKE10F$1 0.572 0.066 8.609 SMOKE10F$2 1.803 0.080 22.489 SMOKE10S$1 0.572 0.066 8.609 SMOKE10S$2 1.803 0.080 22.489 SMOKE11$1 0.572 0.066 8.609 SMOKE11$2 1.803 0.080 22.489 RECEP9$1 -0.841 0.057 -14.821 RECEP9$2 0.132 0.055 2.407 RECEP9$3 0.478 0.057 8.436 RECEP10$1 -0.841 0.057 -14.821 RECEP10$2 0.132 0.055 2.407 RECEP10$3 0.478 0.057 8.436 RECEP11$1 -0.841 0.057 -14.821 RECEP11$2 0.132 0.055 2.407 RECEP11$3 0.478 0.057 8.436 Residual Variances SLEVEL 0.824 0.049 16.963 STREND -0.001 0.011 -0.088 RLEVEL 0.540 0.028 18.954 RTREND 0.029 0.013 2.253 Scales SMOKE9 1.000 0.000 0.000 SMOKE10F 1.000 0.000 0.000 SMOKE10S 1.000 0.000 0.000 SMOKE11 1.000 0.000 0.000 RECEP9 1.000 0.000 0.000 RECEP10 1.000 0.000 0.000 RECEP11 1.000 0.000 0.000 Group HIGH SLEVEL BY SMOKE9 1.000 0.000 0.000 SMOKE10F 1.000 0.000 0.000 SMOKE10S 1.000 0.000 0.000 SMOKE11 1.000 0.000 0.000 STREND BY SMOKE9 0.000 0.000 0.000 SMOKE10F 1.000 0.000 0.000 SMOKE10S 1.422 0.000 0.000 SMOKE11 2.099 0.000 0.000 RLEVEL BY RECEP9 1.000 0.000 0.000 RECEP10 1.000 0.000 0.000 RECEP11 1.000 0.000 0.000 RTREND BY RECEP9 0.000 0.000 0.000 RECEP10 1.000 0.000 0.000 RECEP11 2.000 0.000 0.000 STREND ON SLEVEL 0.191 0.084 2.275 RLEVEL -0.092 0.071 -1.292 RTREND 0.477 0.162 2.936 SLEVEL ON RLEVEL 0.597 0.091 6.593 RTREND ON SLEVEL -0.032 0.044 -0.716 SLEVEL ON GENDER -0.414 0.125 -3.314 RACE 0.301 0.149 2.022 EXPOSE9 0.525 0.119 4.393 STREND ON GENDER 0.104 0.078 1.327 RACE -0.126 0.080 -1.568 EXPOSE9 0.069 0.085 0.806 RTREND ON GENDER -0.109 0.071 -1.531 RACE 0.048 0.077 0.625 EXPOSE9 0.236 0.075 3.168 RLEVEL ON GENDER 0.030 0.102 0.296 RACE -0.130 0.117 -1.104 EXPOSE9 0.400 0.108 3.717 Intercepts SLEVEL 0.000 0.000 0.000 STREND -0.021 0.075 -0.281 RLEVEL 0.000 0.000 0.000 RTREND -0.125 0.067 -1.886 Thresholds SMOKE9$1 0.572 0.066 8.609 SMOKE9$2 1.803 0.080 22.489 SMOKE10F$1 0.572 0.066 8.609 SMOKE10F$2 1.803 0.080 22.489 SMOKE10S$1 0.572 0.066 8.609 SMOKE10S$2 1.803 0.080 22.489 SMOKE11$1 0.572 0.066 8.609 SMOKE11$2 1.803 0.080 22.489 RECEP9$1 -0.841 0.057 -14.821 RECEP9$2 0.132 0.055 2.407 RECEP9$3 0.478 0.057 8.436 RECEP10$1 -0.841 0.057 -14.821 RECEP10$2 0.132 0.055 2.407 RECEP10$3 0.478 0.057 8.436 RECEP11$1 -0.841 0.057 -14.821 RECEP11$2 0.132 0.055 2.407 RECEP11$3 0.478 0.057 8.436 Residual Variances SLEVEL 0.741 0.063 11.829 STREND 0.067 0.000 0.000 RLEVEL 0.591 0.058 10.148 RTREND 0.143 0.045 3.148 Scales SMOKE9 1.000 0.000 0.000 SMOKE10F 0.852 0.047 18.141 SMOKE10S 0.782 0.054 14.456 SMOKE11 0.681 0.060 11.450 RECEP9 1.000 0.000 0.000 RECEP10 0.930 0.059 15.764 RECEP11 0.861 0.059 14.669 |
|
|
You may have been suprised that the weighted average of the two unconstrained estimates is not the estimated value of the constrained estimate. This can come about when you have a model with many degrees of freedom and a certan amount of misfit. For your sample size, the model does not fit very well. If you modify your model to fit better, these estimates are likely to change. When you want to send full outputs, please send them to support@statmodel.com not to the discussion board. |
|
daniel posted on Wednesday, August 27, 2003 - 4:14 am
|
|
|
Thanks |
|
anonymous posted on Tuesday, August 05, 2008 - 6:58 pm
|
|
|
Hello, I'm running an LGM with two groups. In the first group, there is an intercept, linear slope, and quadratic slope. In the second group, there is only an intercept and quadratic slope. In addition, I tested differences between the intercept of the first and second group by constraining: 1) the intercept mean; and 2) the intercept variance. Chi-square difference tests indicated that both parameters differed across groups, so I have left them free. Given this difference in intercepts across groups, is it appropriate to test equivalence of path coefficients from a covariate to the intercepts? Or should these also remain free? |
|
|
A regression coefficient can be the same across classes even if the mean and variance of the dependent variable differ, but it is more likely that it also is different. |
|
gibbon lab posted on Tuesday, October 11, 2011 - 7:20 pm
|
|
|
I am running a two group analysis on LGM with binary endogenous variables. The proportions of the higher category increase in both groups. My code is: categorical are t3smklast3bin t4smk3cat t5smk3cat t6smk3bin; grouping=drd4bin (0=nonrisk 1=risk); analysis:parameterization is theta; MODEL: smk_i by t3smklast3bin@1 t4smk3cat@1 t5smk3cat@1 t6smk3bin@1; smk_s by t3smklast3bin@0 t4smk3cat@3 t5smk3cat@6 t6smk3bin@8; [t3smklast3bin$1 t4smk3cat$1 t5smk3cat$1 t6smk3bin$1] (1); The output says: ... WARNING: THE RESIDUAL COVARIANCE MATRIX (THETA) IN GROUP RISK IS NOT POSITIVE DEFINITE. PROBLEM INVOLVING VARIABLE T6SMK3BI. THE STANDARD ERRORS OF THE MODEL PARAMETER ESTIMATES COULD NOT BE COMPUTED. THE MODEL MAY NOT BE IDENTIFIED. CHECK YOUR MODEL. PROBLEM INVOLVING PARAMETER 13. THE CONDITION NUMBER IS 0.419D-18. However, when I run the exactly same model separately in each group, it worked in both groups. What to do? Thanks a lot. |
|
|
You should use the bar (|) language to set up your growth model correctly. |
|
|
I am running a multiple group latent growth curve model. I have two groups and three time points. The model fit is poor so I examined the modification indices. One suggestion is to add [affw1] for one group. affw1 is the observed variable for the first time point of the growth model. The intercept is set to the first time point and is free to vary between groups. So my question is -- what is happening in the model if I request this group mean? I thought it was already estimated since I am estimating different intercepts for each group. Thank you! |
|
|
The intercepts for the outcomes must be fixed at zero. This is part of the growth model parametrization. I would suggest fitting the growth model in each group separately as a first step. If the same growth model does not fit in each group, you should not proceed to a multiple group analysis. |
|
Back to top |