Message/Author |
|
|
Hi, I was wondering if the modification indices are sensitive to large sample sizes? For instance if I have a larger sample size, will more modification indices come up as significant? Thanks for your help, Lucy |
|
|
Yes. |
|
|
Thankyou. Because of the large data set I have to work with, I am planning on taking a few random samples and doing the Mulitgroup Factor Analysis on these. I understand that large sample sizes also affect the ch-squared difference test. What sample sizes would you suggest would work best for the chi-square difference testing and MIs? Thanks again, Lucy |
|
|
Tough question since it depends on so many factors such as outcome distribution and model parameter values. Often with continuous-normal outcomes n=500 seems to be a middle of the road sample size. |
|
|
Thanks. To be more specific, I have ten items, each with 5 levels of response. At the moment I'm taking sample sizes of about 1000. |
|
|
There are many factors that influence the sample size needed. See the following paper which is available on the website: Muthén, L.K. & Muthén, B.O. (2002). How to use a Monte Carlo study to decide on sample size and determine power. Structural Equation Modeling, 4, 599-620. |
|
Back to top |