Hi, I posted this in the psychologiy statistics forum as well, but haven't gotten a response. I thought maybe it would fit better here.

I have this output from Stata (an explanation of the variables follows). As far as I understand, the policyfoundation coefficient is statistically significant. But, I'm not sure what to make of the random effects parameters, how do I interpret that? Also, am I correct that given the very (negatively) large log likelihood, this is a pretty lousy model (I have lots of other variables, I just wanted to see if alone, policyfoundation, influenced local.


. xtmelogit local policyfoundation, || grantmakernum:, covariance(independent)
Note: single-variable random-effects specification in grantmakernum equation; covariance structure set
to identity

Refining starting values:

Iteration 0: log likelihood = -7923.7706
Iteration 1: log likelihood = -7914.381
Iteration 2: log likelihood = -7914.1983

Performing gradient-based optimization:

Iteration 0: log likelihood = -7914.1983
Iteration 1: log likelihood = -7914.1979

Mixed-effects logistic regression Number of obs = 18590
Group variable: grantmakernum Number of groups = 1363

Obs per group: min = 1
avg = 13.6
max = 347

Integration points = 7 Wald chi2(1) = 38.34
Log likelihood = -7914.1979 Prob > chi2 = 0.0000

----------------------------------------------------------------------------------
local | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-----------------+----------------------------------------------------------------
policyfoundation | -.6397503 .1033141 -6.19 0.000 -.8422421 -.4372585
_cons | 2.286715 .0783974 29.17 0.000 2.133059 2.440371
----------------------------------------------------------------------------------

------------------------------------------------------------------------------
Random-effects Parameters | Estimate Std. Err. [95% Conf. Interval]
-----------------------------+------------------------------------------------
grantmaker~m: Identity |
sd(_cons) | 1.09229 .0613483 .9784313 1.219398
------------------------------------------------------------------------------

my data set contains 18,590 observations (grants) grouped into 1363 foundations (level 2 groups).

local (dependent variable) is a dichotomous variable where foundation grants are either targeted at local government (=1) or other levels of government (=0). In the MLM, this the level 1.

policyfoundation (independent variable) is dichotomous variable where foundations are coded as being policy foundations (=1) or not policy foundations (=0). In the MLM, this is the level 2.

the identity variable, grantmakernum, is the level 2 variable that defines the groups within in level 2. it is an id number for each foundation in the data set.

thanks!!!