multilevel poisson with genlinfixed?

[Maguin, Eugene]

Is it possible to estimate a two level poisson model using genlinmixed? I had expected that it would be given an reply to a question about multilevel mixed model with a non-normal level 1 distribution, I think, it was a logistic distribution. My syntax statement is

 

GENLINMIXED /DATA_STRUCTURE SUBJECTS=REC*MON/FIELDS TARGET=MONDV/

   TARGET_OPTIONS DISTRIBUTION=POISSON LINK=LOG/

   RANDOM USE_INTERCEPT=TRUE SUBJECTS=MON.

 

I get a warning

 

Warnings

The target must have a continuous measurement level if the Poisson probability distribution is specified.

Execution of this command stops.

 

Which suggests my model can not be estimated. But, have I omitted something?

I noticed that an example of a multilevel normal distribution model but nothing with logistic regression, for example.

 

Gene Maguin

[Ryan]

GENLINMIXED can certainly fit a multilevel Poisson. I don't have time to investigate your code, but to start I suggest you change the variable measure to scale. 

Ryan

On Fri, May 2, 2014 at 10:49 AM, Maguin, Eugene <[hidden email]> wrote:

Is it possible to estimate a two level poisson model using genlinmixed? I had expected that it would be given an reply to a question about multilevel mixed model with a non-normal level 1 distribution, I think, it was a logistic distribution. My syntax statement is

 

GENLINMIXED /DATA_STRUCTURE SUBJECTS=REC*MON/FIELDS TARGET=MONDV/

   TARGET_OPTIONS DISTRIBUTION=POISSON LINK=LOG/

   RANDOM USE_INTERCEPT=TRUE SUBJECTS=MON.

 

I get a warning

 

Warnings

The target must have a continuous measurement level if the Poisson probability distribution is specified.

Execution of this command stops.

 

Which suggests my model can not be estimated. But, have I omitted something?

I noticed that an example of a multilevel normal distribution model but nothing with logistic regression, for example.

 

Gene Maguin

[Alex Reutter]

Hi Gene,

Based on the error message, I think you need

        VARIABLE LEVEL mondv (SCALE).

prior to the GENLINMIXED command.

Alex



From:        "Maguin, Eugene" <[hidden email]>
To:        [hidden email],
Date:        05/02/2014 10:53 AM
Subject:        multilevel poisson with genlinfixed?
Sent by:        "SPSSX(r) Discussion" <[hidden email]>

---

Is it possible to estimate a two level poisson model using genlinmixed? I had expected that it would be given an reply to a question about multilevel mixed model with a non-normal level 1 distribution, I think, it was a logistic distribution. My syntax statement is
 
GENLINMIXED /DATA_STRUCTURE SUBJECTS=REC*MON/FIELDS TARGET=MONDV/
   TARGET_OPTIONS DISTRIBUTION=POISSON LINK=LOG/
   RANDOM USE_INTERCEPT=TRUE SUBJECTS=MON.
 
I get a warning
 

Warnings

The target must have a continuous measurement level if the Poisson probability distribution is specified.

Execution of this command stops.

 
Which suggests my model can not be estimated. But, have I omitted something?
I noticed that an example of a multilevel normal distribution model but nothing with logistic regression, for example.
 
Gene Maguin

[Maguin, Eugene]

That was the problem. It wouldn’t have occurred to me. Thank you.

 

[Morrell, Ronald]

Eugene,

 

Which fixed the problem, the line of syntax suggested by Alex or switching the variable measure to scale suggested by Ryan?

 

Thanks,

Ron Morrell

 

 

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Maguin, Eugene
Sent: Friday, May 02, 2014 1:41 PM
To: [hidden email]
Subject: Re: multilevel poisson with genlinfixed?

 

That was the problem. It wouldn’t have occurred to me. Thank you.

 

[Maguin, Eugene]

They actually said the same thing.

Alex:

Based on the error message, I think you need VARIABLE LEVEL mondv (SCALE). prior to the GENLINMIXED command.

Ryan:

… change the variable measure to scale

Gene Maguin

 

 

From: Morrell, Ronald [mailto:[hidden email]]
Sent: Friday, May 02, 2014 3:57 PM
To: Maguin, Eugene; [hidden email]
Subject: RE: multilevel poisson with genlinfixed?

 

Eugene,

 

Which fixed the problem, the line of syntax suggested by Alex or switching the variable measure to scale suggested by Ryan?

 

Thanks,

Ron Morrell

 

 

From: SPSSX(r) Discussion [[hidden email]] On Behalf Of Maguin, Eugene
Sent: Friday, May 02, 2014 1:41 PM
To: [hidden email]
Subject: Re: multilevel poisson with genlinfixed?

 

That was the problem. It wouldn’t have occurred to me. Thank you.