Hongbo Wang posted on Wednesday, June 11, 2008 - 10:13 pm
Dear Dr. Muthen: Can Mplus fit an endogenous-loading factor model that is adopted by Heckman, Stixrud, and Urzua's 2006 Journal of Labor Economics paper (v.30(3)). It accounts for the effects of schooling choice and background variables on the measurements of two factors. The authors used Fortran-based MCMC code to do the job. My question is: Do you think it is fully doable using Mplus, which no doubt is much more user friendly?
Thanks for you help. Hongbo
Hongbo Wang posted on Wednesday, June 11, 2008 - 10:24 pm
Follow-up on my previous post:
The model assumes two latent factors, one for cognitive ability and the other for non-cognitive ability. Conditional on family background and other individual characteristics, the two factors are stipulated to generate cognitive and non-cognitive test scores (e.g. AFQT and Rosenberg Self-esteem scale), and also cause dependence among a wide range of labor market and behavioral outcomes.