I use Stata 14.1 on Windows 10
Using this code:
I get these results:
My question has to do with the random-effects parameters, specifically, can one say that the variance components (.113 and/or .326) are statistically significant? In some MLM studies, I see these variance components listed as significant, in others, I see no mention of it. I thought that the LR test was a test comparing two models using their log likelihoods - but it is not clear to me that this says anything about whether or not the variance components can be listed as statistically significant or not.
Help is appreciated.
Thanks,
Marie
Using this code:
Code:
mixed tolind c.age##c.age Inspired Fables female Liberal Moderate white ceduc marry kids south urban rural suburban crelcom mainline blackprot catholic nofaith tolindsd [fweight=wtssall]|| _all:R.cohort5 || year:
HTML Code:
-------------------------------------------------------------
| No. of Observations per Group
Group Variable | Groups Minimum Average Maximum
----------------+--------------------------------------------
_all | 1 14,604 14,604.0 14,604
year | 18 348 810.8 1,733
-------------------------------------------------------------
Wald chi2(21) = 8042.40
Log likelihood = -39253.37 Prob > chi2 = 0.0000
------------------------------------------------------------------------------
tolind | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
age | -.0024984 .0133355 -0.19 0.851 -.0286355 .0236387
|
c.age#c.age | -.0003943 .0001274 -3.10 0.002 -.0006439 -.0001446
|
Inspired | 1.680803 .072354 23.23 0.000 1.538992 1.822614
Fables | 1.882974 .1039909 18.11 0.000 1.679155 2.086792
female | .1422362 .0607907 2.34 0.019 .0230886 .2613838
Liberal | .2627643 .0785938 3.34 0.001 .1087233 .4168053
Moderate | .1032633 .0708411 1.46 0.145 -.0355828 .2421094
white | .6327955 .0920592 6.87 0.000 .4523628 .8132283
ceduc | .1892372 .0084649 22.36 0.000 .1726463 .2058281
marry | -.0779974 .0683676 -1.14 0.254 -.2119955 .0560006
kids | -.2239905 .0805381 -2.78 0.005 -.3818422 -.0661387
south | -.5834785 .0644605 -9.05 0.000 -.7098187 -.4571383
urban | .0869833 .0832515 1.04 0.296 -.0761867 .2501532
rural | -.7469693 .1015166 -7.36 0.000 -.9459382 -.5480003
suburban | .3433786 .0730651 4.70 0.000 .2001736 .4865836
crelcom | -.0815977 .0096197 -8.48 0.000 -.100452 -.0627435
mainline | .4905731 .0900192 5.45 0.000 .3141387 .6670074
blackprot | .0190671 .1368865 0.14 0.889 -.2492256 .2873598
catholic | .2306323 .0786859 2.93 0.003 .0764108 .3848538
nofaith | .4159865 .1095781 3.80 0.000 .2012174 .6307556
tolindsd | -8.500621 .1499065 -56.71 0.000 -8.794432 -8.20681
_cons | 11.76895 .3746199 31.42 0.000 11.03471 12.50319
------------------------------------------------------------------------------
------------------------------------------------------------------------------
Random-effects Parameters | Estimate Std. Err. [95% Conf. Interval]
-----------------------------+------------------------------------------------
_all: Identity |
var(R.cohort5) | .1133522 .0581664 .0414609 .3099001
-----------------------------+------------------------------------------------
year: Identity |
var(_cons) | .3264224 .1205952 .1582368 .6733679
-----------------------------+------------------------------------------------
var(Residual) | 12.57375 .1473513 12.28824 12.86589
------------------------------------------------------------------------------
LR test vs. linear model: chi2(2) = 0.00 Prob > chi2 = 1.0000
Note: LR test is conservative and provided only for reference.
Help is appreciated.
Thanks,
Marie