We have an Accelerated Longitudinal Design study where participants are grouped into age bins. We want to provide factor scores of various measures, and have noticed that one could using the “grouping is…” command to set factor loadings and intercepts/thresholds to equality across the groups, with covariances and variances of latent variables varying, and the means/scales of the first grouping category being set to 0/1. While we are more familiar with using this command in a baseline model for subsequent tests of measurement invariance by overriding the factor loadings being set to equality, it seems this might not be solely the use of the this command (see http://www.ats.ucla.edu/stat/mplus/faq/two_group_measurement_model.htm where the command is used on its own without such an override).
So, are we right in thinking that this output creates factor scores which are comparable across age bins (what we are grouping), and therefore might be an appropriate strategy to deal with accelerated cohort design data should we desire such comparability in the full sample? We were also wondering if you could please explain why this grouping method has higher degrees of freedom than we would expect?
Thank you. Further to this discussion we were wondering if there is any precedence to grouping by age bin to deal with cohort issues in an ALD study when generating factor scores (CFA in this particular case). We have not found literature which addresses this issue so would very much appreciate your advice.
Accelerated Longitudinal Designs are typically analyzed as in our UG ex6.18.
See also the paper on our website under Papers, Growth Modeling:
Muthén, B. & Muthén, L. (2000). The development of heavy drinking and alcohol-related problems from ages 18 to 37 in a U.S. national sample. Journal of Studies on Alcohol, 61, 290-300. download paper contact first author show abstract