Dear Statalist users,
I have trouble replicating results I obtained with xttobit in previous versions of Stata. In particular, xttobit gives substantially different results in versions 8, 9, and 10 despite using the same set of variables and the same dataset.
Below you can see the output from each version. Does anyone know why this is happening or encountered a similar problem?
VERSION 8:
(unfortunately, I no longer have the information about the iteration for this version)
. xttobit mycont three asym_level strong, i(group) ul(20) ll(0)
Random-effects tobit regression Number of obs = 1320
Group variable (i): group Number of groups = 33
Random effects u_i ~ Gaussian Obs per group: min = 40
avg = 40.0
max = 40
Wald chi2(3) = 38.02
Log likelihood = -3000.3546 Prob > chi2 = 0.0000
------------------------------------------------------------------------------
mycont | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
three | 3.144437 .5602632 5.61 0.000 2.046342 4.242533
asym_level | .1939476 .2007664 0.97 0.334 -.1995474 .5874425
strong | -1.369307 .5868873 -2.33 0.020 -2.519585 -.2190286
_cons | 11.52007 .6803106 16.93 0.000 10.18668 12.85345
-------------+----------------------------------------------------------------
/sigma_u | 7.943933 1.000435 7.94 0.000 5.983116 9.904751
/sigma_e | 6.75084 .1805469 37.39 0.000 6.396975 7.104705
-------------+----------------------------------------------------------------
rho | .58066 .0621167 .4570053 .6967743
------------------------------------------------------------------------------
Observation summary: 58 left-censored observations
774 uncensored observations
488 right-censored observations
VERSION 9.2:
. xttobit mycont asym_level strong three, ll(0) ul(20) i(group)
Obtaining starting values for full model:
Iteration 0: log likelihood = -3918.8279
Iteration 1: log likelihood = -3918.8114
Iteration 2: log likelihood = -3918.8114
Fitting full model:
Iteration 0: log likelihood = -3113.7583
Iteration 1: log likelihood = -2977.791
Iteration 2: log likelihood = -2974.585
Iteration 3: log likelihood = -2974.5706
Iteration 4: log likelihood = -2974.5706
Random-effects tobit regression Number of obs = 1320
Group variable: group Number of groups = 33
Random effects u_i ~ Gaussian Obs per group: min = 40
avg = 40.0
max = 40
Wald chi2(3) = 7.82
Log likelihood = -2974.5706 Prob > chi2 = 0.0498
------------------------------------------------------------------------------
mycont | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
asym_level | .082819 .7468687 0.11 0.912 -1.381017 1.546655
strong | -1.312225 .5864039 -2.24 0.025 -2.461555 -.1628941
three | 3.021097 1.854002 1.63 0.103 -.6126807 6.654875
_cons | 13.92143 2.425767 5.74 0.000 9.167013 18.67585
-------------+----------------------------------------------------------------
/sigma_u | 7.531323 .9907186 7.60 0.000 5.58955 9.473096
/sigma_e | 6.668207 .1836982 36.30 0.000 6.308165 7.028249
-------------+----------------------------------------------------------------
rho | .5605611 .0652327 .4317845 .6831824
------------------------------------------------------------------------------
Observation summary: 58 left-censored observations
774 uncensored observations
488 right-censored observations
VERSION 10:
. xttobit mycont asym_level strong three, ll(0) ul(20) i(group)
Obtaining starting values for full model:
Iteration 0: log likelihood = -3918.8279
Iteration 1: log likelihood = -3918.8114
Iteration 2: log likelihood = -3918.8114
Fitting full model:
Iteration 0: log likelihood = -3112.6087
Iteration 1: log likelihood = -2976.4144
Iteration 2: log likelihood = -2974.0119
Iteration 3: log likelihood = -2973.9993
Iteration 4: log likelihood = -2973.9993
Random-effects tobit regression Number of obs = 1320
Group variable: group Number of groups = 33
Random effects u_i ~ Gaussian Obs per group: min = 40
avg = 40.0
max = 40
Wald chi2(3) = 8.47
Log likelihood = -2973.9993 Prob > chi2 = 0.0372
------------------------------------------------------------------------------
mycont | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
asym_level | -.080391 .9446354 -0.09 0.932 -1.931842 1.77106
strong | -1.291137 .5891479 -2.19 0.028 -2.445846 -.1364281
three | 4.746394 2.791425 1.70 0.089 -.7246978 10.21749
_cons | 14.47964 3.339526 4.34 0.000 7.934287 21.02499
-------------+----------------------------------------------------------------
/sigma_u | 7.178246 .9502204 7.55 0.000 5.315848 9.040644
/sigma_e | 6.683428 .1857913 35.97 0.000 6.319284 7.047573
-------------+----------------------------------------------------------------
rho | .5356514 .066032 .4066286 .6610002
------------------------------------------------------------------------------
Observation summary: 58 left-censored observations
774 uncensored observations
488 right-censored observations
Thank you in advance for your responses.
Nikos Nikiforakis
Nikos Nikiforakis
