Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: Convergence never achieved with MI impute chained

From   Lena Lindbjerg Sperling <>
Subject   st: Convergence never achieved with MI impute chained
Date   Wed, 20 Jun 2012 18:47:09 +0200

Dear all,

I'm working on several complex survey data sets exploring the wages and sectoral distribution of workers.
Of course these data sets all have missing observations at different degrees and for different variables. Therefore MI impute chained seemed like the optimal choice. But whatever I do I get the message convergence not achieved (after a very long 2 hours of thinking). A description of my mi set data is here:

  Style:  wide
          last mi update 20jun2012 17:02:46, 0 seconds ago

  Obs.:   complete          340
          incomplete      7,314  (M = 0 imputations)
          total           7,654

  Vars.:  imputed:  19; born(78) everattend(52) write(55) marital(106) origin(99) inmig(153) regorigin(174) atschool(157)
                    occup(126) ocusec(144) whours(4074) edulevel(659) ocu(228) migrmoved(5638) migryears(5638) migreason(5664)
                    recmig(6645) industry(204) mwage(598)

          passive:  0

          regular:  9; soc reg gender head age urb vocational lstatus unitwage

          system:   1; _mi_miss

         (there are 17 unregistered variables)

I would like to use as many variables as possible as predictors, as both wage and sector of work would probably be dependent on both education, migration etc. But even if I limit myself as much as possible to avoid perfect collinarity I get convergence not achieved.

My MI code looks like this:

mi impute chained (mlogit) ocu occup ocusec industry (ologit, noimp) edulevel (regress) mwage whours =  soc reg gender age urb if lstatus==1 [pweight=wgt], chainonly burnin(100) savetrace(impstats, replace) augment

Any help on why I can never get convergence is very much appreciated! My data is quite far from monotone, so need to use something else than that simple approach.

mi misstable nested ocu occup ocusec industry edulevel mwage whours 

     1.  occup(126)
     2.  ocusec(144)
     3.  industry(204)
     4.  ocu(228)
     5.  mwage(598)
     6.  edulevel(659)
     7.  whours(4074)

Thank you in advance!

*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index