|
 |
Longitudinal Factor Analysis |
 
|
Message/Author |
|
|
This post could probably fit in multiple discussion categories, so I apologize for the redundancy. Is there such a thing as a longitudinal factor analysis (exploratory and/or confirmatory) where the number of factors is not constant across years? Does this even make sense? I have an investigator who believes that may be the case with her data. Thanks. |
|
|
I think it makes sense - age differentation seems reasonable. What degree of invariance of common factors over time is, however, a challenging topic to study. A CFA would study all time points together. |
|
|
I would like to conduct an exploratory factor analysis on longitudinal data. I have a median of 8 time points (range is 3 to 70) per participants (n = 588) and a measure with just 7 items. As such, a longitudinal factor analysis seems like a reasonable approach (I would need to purchase mplus), but I was wondering whether there is such a thing as a priori power analysis for this analysis? I simply want to check if I have sufficient sample size/etc. to warrant such an analysis. Thank you for your help, Kelly |
|
|
That seems like a big sample given the number of time points that you have. Seems like a 2-level, long-data-format, analysis is called for given the many time points for some subjects. This calls for measurement invariance over time for your items. You can do a power analysis in Mplus in line with 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. which is on our website under Papers, SEM. |
|
Back to top |
|
|