|
|
 |
| Monte Carlo Power Analysis Help |
 
|
| Message/Author |
|
|
|
|
Dear Drs. Muthen,
We are attempting to run a univariate latent difference score analysis, measuring the latent change attributed to constance/nonstationarity and change attributed to autoregressive/proportionality – the change in our variable of interest as attributed to the other variable’s estimate at the previous time point. We have 4 time points that have observed measurements.
We need to determine if our n is sufficiently powering this analysis and therefore want to run the Monte Carlo procedure. The example I believe we want to follow is your 12.7 parts 1 and 2, wherein we specify our current data as population parameters to be used as reference for the simulated iterations. Is this an appropriate use of that example?
Additionally, when we attempt to do step 1, we receive warnings indicating that each individual variable is not correlated with the other. We want to be able to specify that there exists an auto-correlation structure for the population values, but unsure how to do so. Could you please advise?
Best Regards,
Luke |
|
|
|
|
Q1: That should work.
Q2: See UG ex6.17. |
|
|
| Back to top |
|
|
