悬赏20币！非线性分层模型（HGLM）的零模型怎么求出ICC？

llf2000

4396

收藏 2015-11-14

我要考察地区效应对家庭收入的影响，因变量家庭人均收入划分两个阶段，大于1689赋值为1，小于等于1689为0，所以因变量是二分类变量，用非线性分层模型考察（HGLM），首先是零模型分析，我的命令是
. xtmelogit pcinc_ad|| commid:

结果是
Refining starting values:

Iteration 0: log likelihood = -10209.148
Iteration 1: log likelihood = -10203.565
Iteration 2: log likelihood = -10203.562

Performing gradient-based optimization:

Iteration 0: log likelihood = -10203.562
Iteration 1: log likelihood = -10203.562

Mixed-effects logistic regression             Number of obs    =    18358
Group variable: commid                         Number of groups =    233

                                             Obs per group: min =       16
                                                            avg =    78.8
                                                            max =    114

Integration points = 7                      Wald chi2(0)    =       .
Log likelihood = -10203.562                   Prob > chi2       =       .

------------------------------------------------------------------------------
pcinc_ad |    Coef. Std. Err.    z P>|z|    [95% Conf. Interval]
-------------+----------------------------------------------------------------
   _cons |  -.6302385 .0796407 -7.91 0.000 -.7863313 -.4741456
------------------------------------------------------------------------------

------------------------------------------------------------------------------
  Random-effects Parameters  | Estimate Std. Err.    [95% Conf. Interval]
-----------------------------+------------------------------------------------
commid: Identity          |
               sd(_cons) | 1.178406    .06029    1.065972    1.3027
------------------------------------------------------------------------------
LR test vs. logistic regression: chibar2(01) =  3420.61 Prob>=chibar2 = 0.0000

请问如何求出零模型的ICC,就是组间关联度，这个结果只有一个sd(_CONS)，没有sd（Residual）啊，请问如何看ICC
.