I'm having some problem about irt prediction model.
When i use irt hybrid model, it cannot predict the latent value even it has successfully converge during the irt model processed.
It contain binary variable and ordinal variable.
global pvars1 = "varA"
global pvars2 = "varB"
set seed 2453
* IRT model *
irt hybrid (2pl $pvars1) (grm $pvars2)
estat report $pvars1 $pvars2, byparm sort(b)
predict latent_hybird, latent se(latent_se_hybird)
Hybrid IRT model Number of obs = 4,589
Log likelihood = -1421.6966
------------------------------------------------------------------------------
| Coefficient Std. err. z P>|z| [95% conf. interval]
------------------------------------------------------------------------------
2pl
------------------------------------------------------------------------------
varA~y |
Discrim | 250.6541 36.29445 6.91 0.000 179.5183 321.7899
Diff | .7950825 .0022131 359.26 0.000 .7907448 .7994202
------------------------------------------------------------------------------
grm
------------------------------------------------------------------------------
varB~y |
Discrim | 301.6219 53.6402 5.62 0.000 196.4891 406.7548
Diff |
>=2 | .7976474 .0021011 .7935294 .8017655
>=3 | .8030163 .0022279 .7986498 .8073829
>=4 | .8036267 .0022378 .7992407 .8080127
=5 | .8057767 .0022524 .8013621 .8101913
------------------------------------------------------------------------------
. estat report $pvars1 $pvars2, byparm sort(b)
Hybrid IRT model Number of obs = 4,589
Log likelihood = -1421.6966
-----------------------------------------------------------------------------------------
| Coefficient Std. err. z P>|z| [95% conf. interval]
------------------------+----------------------------------------------------------------
Discrim |
VarA | 250.6541 36.29445 6.91 0.000 179.5183 321.7899
VarB | 301.6219 53.6402 5.62 0.000 196.4891 406.7548
Diff |
VarA=1 | .7950825 .0022131 359.26 0.000 .7907448 .7994202
VarB>=2 | .7976474 .0021011 .7935294 .8017655
VarB>=3 | .8030163 .0022279 .7986498 .8073829
VarB>=4 | .8036267 .0022378 .7992407 .8080127
VarB=5 | .8057767 .0022524 .8013621 .8101913
-----------------------------------------------------------------------------------------
predict latent_hybird, latent se(latent_se_hybird)
(option ebmeans assumed)
(using 7 quadrature points)
could not compute empirical Bayes means;
missing values were returned by the evaluator
any clues about this problem?
thank you so much
When i use irt hybrid model, it cannot predict the latent value even it has successfully converge during the irt model processed.
It contain binary variable and ordinal variable.
global pvars1 = "varA"
global pvars2 = "varB"
set seed 2453
* IRT model *
irt hybrid (2pl $pvars1) (grm $pvars2)
estat report $pvars1 $pvars2, byparm sort(b)
predict latent_hybird, latent se(latent_se_hybird)
Hybrid IRT model Number of obs = 4,589
Log likelihood = -1421.6966
------------------------------------------------------------------------------
| Coefficient Std. err. z P>|z| [95% conf. interval]
------------------------------------------------------------------------------
2pl
------------------------------------------------------------------------------
varA~y |
Discrim | 250.6541 36.29445 6.91 0.000 179.5183 321.7899
Diff | .7950825 .0022131 359.26 0.000 .7907448 .7994202
------------------------------------------------------------------------------
grm
------------------------------------------------------------------------------
varB~y |
Discrim | 301.6219 53.6402 5.62 0.000 196.4891 406.7548
Diff |
>=2 | .7976474 .0021011 .7935294 .8017655
>=3 | .8030163 .0022279 .7986498 .8073829
>=4 | .8036267 .0022378 .7992407 .8080127
=5 | .8057767 .0022524 .8013621 .8101913
------------------------------------------------------------------------------
. estat report $pvars1 $pvars2, byparm sort(b)
Hybrid IRT model Number of obs = 4,589
Log likelihood = -1421.6966
-----------------------------------------------------------------------------------------
| Coefficient Std. err. z P>|z| [95% conf. interval]
------------------------+----------------------------------------------------------------
Discrim |
VarA | 250.6541 36.29445 6.91 0.000 179.5183 321.7899
VarB | 301.6219 53.6402 5.62 0.000 196.4891 406.7548
Diff |
VarA=1 | .7950825 .0022131 359.26 0.000 .7907448 .7994202
VarB>=2 | .7976474 .0021011 .7935294 .8017655
VarB>=3 | .8030163 .0022279 .7986498 .8073829
VarB>=4 | .8036267 .0022378 .7992407 .8080127
VarB=5 | .8057767 .0022524 .8013621 .8101913
-----------------------------------------------------------------------------------------
predict latent_hybird, latent se(latent_se_hybird)
(option ebmeans assumed)
(using 7 quadrature points)
could not compute empirical Bayes means;
missing values were returned by the evaluator
any clues about this problem?
thank you so much