An update to my previous posts (see below). David M (SPSS Support) was able to replicate the problem and forwarded the issue to SPSS Development. He has contacted me with the following information and verified the work-around of manually pooling the results from the individual imputations:

================= SPSS Development reports that this problem has been fixed for Statistics 18.0.1, which should be available by the end of this year. The error occurred because the pooling of estimates was halted when any of the coefficients had an empty standard error cell. The parameter labeled "(Scale)" had no standard error in the affected models, but this condition is legitimate for many of the models available in GENLIN. The pooling was successful with the LOGISTIC REGRESSION procedure results for the same model because that procedure does not print a "(Scale)" Parameter, so all of the nonredundant parameters had nonempty Standard Error cells.

In Statistics 18.0.1 and later versions, the pooling will not be halted when the Scale parameter has no standard error. When the model is only available through GENLIN, applying Rubin's rules to the set of parameter estimates is a work-around until Statistics 18.0.1 is available. =================

Thanks again to David for picking up my tech support case and following through with this.

Cheers, Kylie.

Quoting Kylie Lange <kylie.lange@adelaide.edu.au>:

> David Matheson from SPSS has been kind enough to take a look at my data and > syntax and has been able to replicate the problem in version 18. He thinks > it may be a bug with the pooling in GENLIN, but once he or I know more we'll > summarise back to the list. > > Thanks, > Kylie. > > > -----Original Message----- > From: Reutter, Alex [mailto:areutter@spss.com] > Sent: Tuesday, 15 September 2009 11:53 PM > To: kylie.lange@adelaide.edu.au; SPSSX-L@LISTSERV.UGA.EDU > Subject: RE: multiple imputation/GENLIN error > > That is weird. Can you post the GENLIN syntax that doesn't work and the > LOGISTIC and REGRESSION syntax that does work? > > Alex > > -----Original Message----- > From: SPSSX(r) Discussion [mailto:SPSSX-L@LISTSERV.UGA.EDU] On Behalf Of > Kylie Lange > Sent: Monday, September 14, 2009 10:16 PM > To: SPSSX-L@LISTSERV.UGA.EDU > Subject: multiple imputation/GENLIN error > > Hi everyone, > > Has anyone encountered the following message when using GENLIN on a multiply > imputed dataset? Any ideas of what could be causing it? > > 'At least one of the vectors or matrices in the model being added is > different in size from the corresponding vector or matrix in previously > added models. This model cannot be added.' > > I have successfully analysed this MI dataset using other procedures like > LOGISTIC REGRESSION and REGRESSION, however GENLIN is giving this error. The > Pooled section of the Parameter Estimates table is empty and each of the > empty B cells contains a superscript note to the error given above. The > model runs without any other kind of error or warning and the results from > each of the individual imputations look complete and stable. > > I have opened a Tech Support case with SPSS asking for more information on > the error and what could cause it, however have been stuck for a week with > two support reps who don't appear to have any experience with the multiple > imputation procedures and they haven't been able to give me any information > on the error at all. > > Thanks, > Kylie. > > ===================== > To manage your subscription to SPSSX-L, send a message to > LISTSERV@LISTSERV.UGA.EDU (not to SPSSX-L), with no body text except the > command. To leave the list, send the command > SIGNOFF SPSSX-L > For a list of commands to manage subscriptions, send the command > INFO REFCARD

===================== To manage your subscription to SPSSX-L, send a message to LISTSERV@LISTSERV.UGA.EDU (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD