-------------------------------------------------------------------------------
      name:  <unnamed>
       log:  Q:\C-modelling\runmlwin\website\logfiles\2020-03-27\16\12.2.smcl
  log type:  smcl
 opened on:  27 Mar 2020, 18:22:36

. ****************************************************************************
. * Module 12: Cross-Classified Models - Stata Practical
. *
. *     P12.2: A Cross-Classified Model of Attainment
. *
. *            George Leckie
. *            Centre for Multilevel Modelling, 2011
. ****************************************************************************
. *     Stata do-file to replicate all analyses using runmlwin
. *
. *     George Leckie
. *     Centre for Multilevel Modelling, 2013
. *     http://www.bristol.ac.uk/cmm/software/runmlwin/
. ****************************************************************************
. 
. * P12.2.1 Specifying and fitting the cross-classified model
. 
. use "http://www.bristol.ac.uk/cmm/media/runmlwin/12.2.dta", clear

. 
. xtmixed attain || _all: R.schid || neighid:, mle variance

Performing EM optimization: 

Performing gradient-based optimization: 

Iteration 0:   log likelihood = -3178.3835  
Iteration 1:   log likelihood = -3178.3557  
Iteration 2:   log likelihood = -3178.3557  

Computing standard errors:

Mixed-effects ML regression                     Number of obs     =      2,310

-------------------------------------------------------------
                |     No. of       Observations per Group
 Group Variable |     Groups    Minimum    Average    Maximum
----------------+--------------------------------------------
           _all |          1      2,310    2,310.0      2,310
        neighid |        524          1        4.4         16
-------------------------------------------------------------

                                                Wald chi2(0)      =          .
Log likelihood = -3178.3557                     Prob > chi2       =          .

------------------------------------------------------------------------------
      attain |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
       _cons |   .0753532   .0722216     1.04   0.297    -.0661987     .216905
------------------------------------------------------------------------------

------------------------------------------------------------------------------
  Random-effects Parameters  |   Estimate   Std. Err.     [95% Conf. Interval]
-----------------------------+------------------------------------------------
_all: Identity               |
                var(R.schid) |    .075445   .0316491      .0331553    .1716755
-----------------------------+------------------------------------------------
neighid: Identity            |
                  var(_cons) |   .1412201   .0218651       .104257     .191288
-----------------------------+------------------------------------------------
               var(Residual) |   .7990182   .0263652      .7489788    .8524007
------------------------------------------------------------------------------
LR test vs. linear model: chi2(2) = 207.44                Prob > chi2 = 0.0000

Note: LR test is conservative and provided only for reference.

. 
. estimates store model1

. 
. 
. 
. * P12.2.2 Interpretation of the cross-classified model
. 
. * P12.2.3 Calculating variance partition coefficients (VPC) and intra-class
. *         correlations (ICC)
. 
. 
. 
. * P12.2.4 Predicting and examining school and neighbourhood effects
. 
. predict v0, reffects level(_all)

. 
. predict u0, reffects level(neighid)

. 
. predict v0se, reses level(_all)

. 
. predict u0se, reses level(neighid)

. 
. egen v0rank = rank(v0) if pickone_school==1
(2293 missing values generated)

. 
. egen u0rank = rank(u0) if pickone_neigh==1
(1786 missing values generated)

. 
. sort v0rank

. 
. list schid v0 v0se v0rank if pickone_school==1, noobs

  +---------------------------------------+
  | schid          v0       v0se   v0rank |
  |---------------------------------------|
  |     1   -.4203102    .171911        1 |
  |    10    -.365078   .0963185        2 |
  |    20   -.3377112   .0829578        3 |
  |     9    -.249002   .0929058        4 |
  |    19      -.1888   .1061417        5 |
  |---------------------------------------|
  |     8   -.1046162   .0866322        6 |
  |     2   -.0737279   .0905647        7 |
  |    17    -.019355   .0857978        8 |
  |     7    .0397741   .0718181        9 |
  |     0    .0567248   .0947632       10 |
  |---------------------------------------|
  |     5     .101812   .0891645       11 |
  |     3    .1378104   .0890324       12 |
  |    16    .1448746   .1007428       13 |
  |    15    .1958423   .0816363       14 |
  |     6    .2301028   .1029594       15 |
  |---------------------------------------|
  |    18    .2421351   .0965405       16 |
  |    13    .6095243   .1101784       17 |
  +---------------------------------------+

. 
. summarize u0 if pickone_neigh==1

    Variable |        Obs        Mean    Std. Dev.       Min        Max
-------------+---------------------------------------------------------
          u0 |        524   -3.25e-10    .2348217  -.5765654    .695379

. 
. summarize u0 if pickone_neigh==1, detail

                BLUP r.e. for neighid: _cons
-------------------------------------------------------------
      Percentiles      Smallest
 1%    -.4657722      -.5765654
 5%    -.3664438      -.5677193
10%    -.2978136      -.5482477       Obs                 524
25%    -.1672465      -.4739119       Sum of Wgt.         524

50%    -.0087194                      Mean          -3.25e-10
                        Largest       Std. Dev.      .2348217
75%     .1416191       .6051867
90%     .2975899       .6275532       Variance       .0551412
95%     .4239585       .6655953       Skewness       .2630606
99%     .5628019        .695379       Kurtosis       2.855424

. 
. display r(p75) - r(p25)
.30886561

. 
. qnorm v0 if pickone_school==1, xlabel(-.4(.2).6) aspectratio(1)

. 
. qnorm u0 if pickone_neigh==1, aspectratio(1)

. 
. serrbar v0 v0se v0rank if pickone_school==1, ///
>         scale(1.96) yline(0) mvopts(mlabel(schid))

. 
. generate labheight = v0 + 1.96*v0se + 0.05

. 
. serrbar v0 v0se v0rank if pickone_school==1, scale(1.96) yline(0) ///
>         addplot(scatter labheight v0rank, ///
>         msymbol(none) mlabel(schid) ///
>         mlabposition(1) mlabangle(vertical) mlabcolor(navy)) ///
>         ytitle("Predicted school effect") xtitle("Rank") ///
>         legend(off)

. 
. 
. serrbar u0 u0se u0rank if pickone_neigh==1, scale(1.96) yline(0)

. 
. count if ((u0 + 1.96*u0se)<0 | (u0 - 1.96*u0se)>0) & pickone_neigh==1
  19

. 
end of do-file