www.cmm.bristol.ac.uk/forum

Hi,
I am getting an error message: error while obeying batchfile, while running runmlwin command for random coefficient model.
. runmlwin x_out cons age2 age3 gender primary secondary higher notknowedu smoke moreindoorsmoking drink yesdiab dontknowdiab malnrsh seproom less
> crowd morecrowd mostcrowd POORER MIDDLE RICHER RICHEST urban SC ST OBC dontknow nocaste muslim christian otherrel if gender~=. & pwt1~=., level
> 2(group: cons RICHER, weightvar(pwt2)) level1(ind:, weightvar(pwt1)) discrete(distribution(binomial) link(logit) denominator(cons) pql2) maxite
> rations(100) or

The error is showing in MLwiN window.

Any help on this would be most helpful.

Thanks,
Soumava

Is there any further error text? If it isn't visible in the error message window then you may be able to see it by opening the "command interface" window and clicking the "output" button.

Hi Chris,
The full error message is long and I am trying to copy it below:

Error while obeying batchfile
C:\Users\Soumava\AppData\Local\Temp\ST_01000007.tmp at line number 1216:
calc 'f~~(h)'='f~(h)'*(1-expo('h'))/1+expo('h'))
284862 numeric errors in calculate command, first 20 affected entries listed.
Affected entries replaced with system missing.
See output window for details.

The output window showed:
->OBEY "c:\Users\Soumava\AppData\Local\Temp\ST_01000007.tmp"
->WSET
->WSET

Thanks,

This appears to be an estimation error, which may be resolved by providing better starting values. I would suggest building up the model from simpler versions, using the previous one as starting values for the next. For example you could start without the weights, with mql1 instead of pql2 and with fewer variables. This might also help you determine which aspect is causing the problem.