I am estimating a three-level model: Persons in PSUs in Countries with melogit. My code is below. The output shows that my clusters are unique at PSU level and are unique from the nation. I have 39103 PSU clusters and 63 nation clusters. However, my ICC is EXACTLY the same. Is there an explanation for this?
melogit termpreg [pw=normwgt] || country: || newidPSU:
estat icc
Mixed-effects logistic regression Number of obs = 1034809
-------------------------------------------------------------
| No. of Observations per Group
Group Variable | Groups Minimum Average Maximum
----------------+--------------------------------------------
countryname | 63 2,615 16,425.5 122,351
newidPSU | 39,103 1 26.5 122,351
-------------------------------------------------------------
Integration method: mvaghermite Integration pts. = 7
Wald chi2(0) = .
Log pseudolikelihood = -8628.1485 Prob > chi2 = .
(Std. Err. adjusted for 63 clusters in countryname)
--------------------------------------------------------------------------------------
| Robust
termpreg | Coef. Std. Err. z P>|z| [95% Conf. Interval]
---------------------+----------------------------------------------------------------
_cons | -1.696166 .0661216 -25.65 0.000 -1.825762 -1.56657
---------------------+----------------------------------------------------------------
countryname |
var(_cons)| .1978985 .04542 .1262065 .3103152
---------------------+----------------------------------------------------------------
countryname>newidPSU |
var(_cons)| 2.26e-35 3.13e-35 1.50e-36 3.41e-34
--------------------------------------------------------------------------------------
. est store uncond
. estat icc
Intraclass correlation
------------------------------------------------------------------------------
Level | ICC Std. Err. [95% Conf. Interval]
-----------------------------+------------------------------------------------
countryname | .0567407 .0122837 .0369449 .0861943
newidPSU|countryname | .0567407 .0122837 .0369449 .0861943
------------------------------------------------------------------------------
Related Posts with Exact Same ICC for Two Levels
number of records teffects matching estimators use to estimate effectsHow can I identify the number of records at each treatment level the teffects matching estimators us…
matching FrontierI wonder if anyone has attempted a Stata version of the matching Frontier program of King et al. The…
Sorting string variable fromDear All, I'm having hassle sorting a string variable. Even after using the Code: sort code, the en…
Using coefplot to plot the effect of categorical variable with interactionDear Statalist users, I wondered if you could help me with this issue in particular. I am running …
sample used by teffects matching estimatorsDoes the teffects suite of commands include a command showing how many records at each treatment lev…
Subscribe to:
Post Comments (Atom)
0 Response to Exact Same ICC for Two Levels
Post a Comment