multiple membership models - error 0142

[rebvas86]

Trying to run some multiple membership models and am getting this error message:
error while obeying batch file...at line number 77:
MCMC 0 5000 1 5.8 10 2 2 2 1 1 2
MCMC error 0142: The 773th MM unit for obs 2 is absent but has positive weight.
I checked the mm ids and weights for the 773th individual (I hope I understood the error message and am checking the right information) and i have id33 (mm1) and id33 (mm2) and weights 1 and 0 respectively.
Not sure what is the problem here.
Would appreciate any feedback.
Thanks,
Rebecca

[rebvas86]

just run this line of code to get a list of the cases where the same mm (interviewer code) is listed across both classifications (survey waves) and it seems the 773th individual is the first case when this happens.
am i coding the data incorrectly?

list clid clid2 w1v w2v if w1v==1

+--------------------------+
| clid clid2 w1v w2v |
|--------------------------|
773. | 33 33 1 0 |
992. | 42 42 1 0 |
1110. | 47 47 1 0 |
1325. | 56 56 1 0 |
1952. | 82 82 1 0 |
|--------------------------|
1953. | 82 82 1 0 |
2579. | 108 108 1 0 |
2819. | 118 118 1 0 |
2967. | 124 124 1 0 |
3181. | 133 133 1 0 |
|--------------------------|
3252. | 136 136 1 0 |
3253. | 136 136 1 0 |
3401. | 142 142 1 0 |
4220. | 176 176 1 0 |
4386. | 183 183 1 0 |
|--------------------------|
4630. | 193 193 1 0 |
4774. | 199 199 1 0 |
5327. | 222 222 1 0 |
5328. | 222 222 1 0 |
5422. | 226 226 1 0 |
|--------------------------|
5495. | 229 229 1 0 |
+--------------------------+

[GeorgeLeckie]

Hi Rebecca,

When an interviewee belongs to only one interviewer then the first MM ID should be the ID of that interviewer, the second MM ID should take the value zero. The weight associated with the first MM ID should then be one, while the weight associated with the second MM ID should be zero.

So where you have

773. | 33 33 1 0 |

You should have

773. | 33 0 1 0 |

and so on for all other interviewees who belong to only one interviewer

Best wishes

George

[GeorgeLeckie]

Hi Rebecca,

As you can probably tell, the error message you received is a bit jumbled. I believe we implemented a more intelligible message in the latest version of MLwiN (2.26). It would be great if you could upgrade to 2.26 and confirm that the message is more meaningful

http://www.bristol.ac.uk/cmm/software/m ... rades.html

You should at that point also update to the latest version of runmlwin if you have not already done so

Code: Select all

. adoupdate runmlwin

Best wishes

George

[rebvas86]

Hi George,

Thank you for your message. It seems to be working now.

Will download the new version of MLwiN and update my runmlwin version.

Kind regards,
Rebecca