Message/Author 

jeon small posted on Friday, March 01, 2013  8:05 pm



How do you perform a posthoc power analysis. I have a sample of 438 girls and boys. We used 11 indicator variable to form four variables. The fit of the structural model was satisfactory with CFI=.954, RMSEA=.05, DF=40 CHI SQUARED=95.85 PROBABILITY=.0000 The standardized regression weights were significant and ranged in size form .65.03. 


See Example 12.7 where the values from a real data analysis are saved and used as population parameter values for a Monte Carlo simulation study that can examine power. This method has its critics. You may want to explore this issue on SEMNET. 

jeon small posted on Monday, March 04, 2013  1:38 pm



I ran this model in Amos using maximumlikelihood estimation (for the missing data)however, I can not get the model to converge in Mplus. What am I doing wrong? My output is below. STRUCTURAL MODEL OF ADOLESCENT ALCOHOL USE AND PERCEPTION OF IPV SUMMARY OF ANALYSIS Number of groups 1 Number of observations 553 Number of dependent variables 10 Number of independent variables 0 Number of continuous latent variables 4 Estimator ML Information matrix OBSERVED Maximum number of iterations 1000 Convergence criterion 0.500D04 Maximum number of steepest descent iterations 20 Maximum number of iterations for H1 2000 Convergence criterion for H1 0.100D03 Input data file(s) h:/M_russell/mplus10_08_12v1.csv Input data format FREE 


Please send the outputs and your license number to support@statmodel.com. 


Dear Professors Muthen, I would like to estimate the power to detect that a path coefficient is different from zero and am trying to use https://www.statmodel.com/power.shtml as a procedure. Alas it hasn't worked so far. A simplification of my model looks like the following, with three IVs and three DVs. Two of those are latent, four manifest: IV1 by Indicator1 Indicator2 Indicator3; DV1 by Indicator4 Indicator5 Indicator6; DV1 on IV1 IV2 IV3; DV2 on IV1 IV2 IV3; When I follow step 1 of the procedure on https://www.statmodel.com/power.shtml, fixing values for the factor loadings and the path coefficients, the resulting covariance matrix of the residuals contains zeros, which is not accepted by Mplus in step 2. Do I have to fix more in step 1 than the factor loadings and the path coefficients? (In step 1, I have used the most simple matrix possible, in the form of: 0 0 0 0 0 0 0 0 1 0 1 0 0 1 0 0 0 1 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 ) 


Please send the output from Step 1 to Support along with your license number. 

Back to top 