|
|
Monte Carlo Power Estimation for ESEM |
|
Message/Author |
|
|
Hi, I'm attempting to conduct a Monte Carlo power analysis for an ESEM model (3 correlated factors, 27 items, 9 items per factor), and I'm noticing some peculiar results. The column for the percentage of confidence intervals that capture the true population estimate is decreasing as I increase the sample size. For example, for a particular loading the CI% for n=300 is .831, .649 for n=750, and .252 for n=2000. This seems contrary to what should be occurring because shouldn't the estimation become increasingly more accurate as sample size increases? Does anyone have any insight into why this may be happening? Does the Monte Carlo Method for power estimation not generalize to ESEM? I've noticed this with both Geomin and Target rotations and using different sets of population values. |
|
|
ESEM simulations can be tricky to carry out. See our FAQ: Lambda is not compatible with the notion of simplicity of the rotation criterion |
|
Back to top |
|
|