Tobit type - II GSEM - depvar censored from below

[Chiara Tasselli]

Good morning,

I would like to perform the model reported in the figure below.

The model regards preferences in university choice at the end of high school. I have 3 different outcomes that are continuous values (from 0 to 1) regarding the choices of stem or economics or humanistic path. The fact is that I would like to correct by students that strongly declare that they wanna work at the end of high school ( I have another dummy variable for that).

My idea was to substitute by -1 the three outcome variables when the work dummy is equal to one and then censor from below, like :

gsem (SUBJ -> Intrisict, ) (SUBJ -> Extrinsic, ) (SUBJ -> Cost, ) (SUBJ -> S_belonging) (SUBJ -> abilitysc, ) (SUBJ -> abiltygap, ) ///
(SUBJ -> ecomath_minmaxT, family(gaussian, lcensored(-1)) link(identity) ) (SUBJ -> stem_minmaxT, family(gaussian, lcensored(-1)) link(identity)) (SUBJ -> hum_minmaxT, family(gaussian, lcensored(-1)) link(identity)) (abiltygap -> ecomath_minmaxT, family(gaussian, lcensored(-1)) link(identity)) (abiltygap -> stem_minmaxT, family(gaussian, lcensored(-1)) link(identity)) (abiltygap -> hum_minmaxT, family(gaussian, lcensored(-1)) link(identity)) ///
(M1[school] -> ecomath_minmaxT, family(gaussian, lcensored(-1)) link(identity)) (M1[school] -> stem_minmaxT, family(gaussian, lcensored(-1)) link(identity)) (M1[school] -> hum_minmaxT, family(gaussian, lcensored(-1)) link(identity) ) (female dscore 1.female#c.dscore -> SUBJ, ) if time == 0, ///
covstruct(_lexogenous, diagonal) latent(SUBJ M1 ) nocapslatent

Someone could kindly comment on it or gives me other suggestions about how to perform it? Or suggestions of sites and literature on Tobit type II in Gsem? Many thanks in advance for your time.

PS: Does gsem allow replacement of missing values? does it has an option like <method(mlmv)>