Message/Author |
|
|
I am needing to perform a power analysis for a grant proposal. the data will be clustered (individuals in intervention groups), and so i ran a power analysis for a twolevel, cross-setional regression analysis, using the design effect i anticipate. however, i then realized that i should be able to perform the analysis using type=complex, since i mainly am interested in getting correct parameter and standard error estimates. is there much known about whether i should continue to take the design effect into account when calculating power for a type=complex model (in the same way as i would for a two-level model)? |
|
|
If you are going to analyze the data using TYPE=COMPLEX; then the power study should use this. Note that the data are generated using TYPE=TWOLEVEL but analyzed using an external Monte Carlo using TYPE=COMPLEX; This is done in Example 11.6. |
|
Tim Walker posted on Wednesday, March 29, 2017 - 11:48 am
|
|
|
I am in a similar situation where I would like to determine how many cases (and organizations) should be collected for an EFA and CFA for developing and testing a measure. We are collecting cases from different organizations and therefore would plan on using type=complex or two-way. My question is, what are the power and sample size implications between using type=complex versus two-way? Is approximately the same number of cases and organizations required regardless of what method is used or would you expect there to be a difference - especially in the number of organizations required? |
|
|
You should post a general question like this on a general discussion forum like SEMNET or MultilevelNet. |
|
Back to top |