|
|
Estimating FS for Skewnormal, T and S... |
|
Message/Author |
|
Jan Ivanouw posted on Friday, October 03, 2014 - 1:37 am
|
|
|
Hi I am very excited by the new developments of modeling with non-normal distributions. You have described estimation of factor scores under the normal condition in Appendix 11 of the Technical appendix, especially by formulas (230) and (231). Which formulas do you use for estimating factor scores for SKEWNORMAL, T and SKEWT? |
|
|
See Section 3.6 in http://statmodel.com/download/Skew.pdf for the general case of skewt (the skewnormal and t are special cases of skewt). The relationship is non-linear if BOTH the residuals for the factor and the indicators have skew. In all other cases it is linear. What we compute if E(factor | indicators and covariates), i.e., we use the posterior mean (given the observed data and estimated parameters) as the factor score. |
|
Jan Ivanouw posted on Friday, October 03, 2014 - 4:40 pm
|
|
|
Thank you very much |
|
Back to top |
|
|