Hello all,
I am using the two Tobit models in the structural equation model (SEM) framework as below,
the code is
I got the result as
I have two questions here, 1) the first is how to get the marginal effect of these two models?
I got the result like below,
however, I expect to get the result like the below picture, Could you please give me some advice about it?
Please ignore the black area.
Besides, it is really strange that I cannot god of fit by following codes
Do you know how to get the god of fit of these kinds of models (such as RMSEA, CFL TLI)?
Many thanks in advance.
I am using the two Tobit models in the structural equation model (SEM) framework as below,
the code is
Code:
gsem (Country -> LN_PAT_1, family(gaussian, lcensored(0)) link(identity)) (Country -> RDTA, family( > gaussian, lcensored(0)) link(identity)) (Industry -> LN_PAT_1, family(gaussian, lcensored(0)) link( > identity)) (Industry -> RDTA, family(gaussian, lcensored(0)) link(identity)) (year -> LN_PAT_1, fam > ily(gaussian, lcensored(0)) link(identity)) (year -> RDTA, family(gaussian, lcensored(0)) link(iden > tity)) (LIQUIDITY -> LN_PAT_1, family(gaussian, lcensored(0)) link(identity)) (LIQUIDITY -> RDTA, f > amily(gaussian, lcensored(0)) link(identity)) (LN_TA -> LN_PAT_1, family(gaussian, lcensored(0)) li > nk(identity)) (LN_TA -> RDTA, family(gaussian, lcensored(0)) link(identity)) (PPETA -> LN_PAT_1, fa > mily(gaussian, lcensored(0)) link(identity)) (PPETA -> RDTA, family(gaussian, lcensored(0)) link(id > entity)) (LEV -> LN_PAT_1, family(gaussian, lcensored(0)) link(identity)) (LEV -> RDTA, family(gaus > sian, lcensored(0)) link(identity)), nocapslatent
Code:
Refining starting values:
Grid node 0: log likelihood = -56980.295
Fitting full model:
Iteration 0: log likelihood = -56980.295
Iteration 1: log likelihood = -50103.258
Iteration 2: log likelihood = -49285.87
Iteration 3: log likelihood = -49258.821
Iteration 4: log likelihood = -49258.782
Iteration 5: log likelihood = -49258.782
Generalized structural equation model Number of obs = 57,490
Response : LN_PAT_1 Number of obs = 57,490
Lower limit : 0 Uncensored = 34,977
Family : Gaussian Left-censored = 22,513
Link : identity Right-censored = 0
Response : RDTA Number of obs = 43,266
Lower limit : 0 Uncensored = 41,788
Family : Gaussian Left-censored = 1,478
Link : identity Right-censored = 0
Log likelihood = -49258.782
---------------------------------------------------------------------------------
| Coef. Std. Err. z P>|z| [95% Conf. Interval]
----------------+----------------------------------------------------------------
LN_PAT_1 |
Country | .0089105 .0008246 10.81 0.000 .0072943 .0105267
Industry | -.0017527 .0007134 -2.46 0.014 -.0031509 -.0003545
year | -.031676 .0016251 -19.49 0.000 -.0348612 -.0284908
LIQUIDITY | .0965281 .003857 25.03 0.000 .0889685 .1040877
LN_TA | .4295258 .0058899 72.93 0.000 .4179818 .4410698
PPETA | -.273905 .0259423 -10.56 0.000 -.324751 -.223059
LEV | -1.111073 .0517833 -21.46 0.000 -1.212566 -1.00958
_cons | 57.96435 3.258487 17.79 0.000 51.57783 64.35086
----------------+----------------------------------------------------------------
RDTA |
Country | .0006957 .0000469 14.84 0.000 .0006038 .0007876
Industry | .0005416 .0000365 14.82 0.000 .00047 .0006133
year | .000565 .0000829 6.82 0.000 .0004026 .0007274
LIQUIDITY | .0022623 .0002236 10.12 0.000 .001824 .0027006
LN_TA | -.0172278 .0003179 -54.18 0.000 -.0178509 -.0166046
PPETA | -.0318991 .0013397 -23.81 0.000 -.0345248 -.0292733
LEV | -.0402615 .0026459 -15.22 0.000 -.0454474 -.0350757
_cons | -.8705099 .1662382 -5.24 0.000 -1.196331 -.544689
----------------+----------------------------------------------------------------
var(e.LN_PAT_1)| 3.577723 .0293979 3.520566 3.635809
var(e.RDTA)| .0079849 .0000555 .0078768 .0080945
---------------------------------------------------------------------------------
I got the result like below,
Code:
. mfx compute
Marginal effects after gsem
y = Predicted mean (LN_PAT_1) (predict)
= .75266612
------------------------------------------------------------------------------
variable | dy/dx Std. Err. z P>|z| [ 95% C.I. ] X
---------+--------------------------------------------------------------------
Country | .0050871 .00085 5.97 0.000 .003417 .006758 25.6201
Industry | -.0048731 .00073 -6.65 0.000 -.00631 -.003436 27.8678
year | -.0288715 .00178 -16.23 0.000 -.032357 -.025386 2001.6
LIQUID~Y | .0672255 .00417 16.11 0.000 .059048 .075403 8.11123
LN_TA | .4726395 .0063 74.98 0.000 .460284 .484995 13.0144
PPETA | -.3848692 .02974 -12.94 0.000 -.443168 -.326571 .565122
LEV | -1.00272 .05403 -18.56 0.000 -1.10862 -.896819 .206329
CAPEXTA | 2.735847 .21486 12.73 0.000 2.31472 3.15697 .05374
Q | .1318931 .00548 24.06 0.000 .121147 .142639 2.00173
------------------------------------------------------------------------------
.
however, I expect to get the result like the below picture, Could you please give me some advice about it?
Please ignore the black area.
Besides, it is really strange that I cannot god of fit by following codes
Code:
. estat gof, stats(all) estat gof not valid
Many thanks in advance.
Comment