"Unable to compute due to numerical problems" Error with Generalized Linear Models

[Justin Meyer-3]

Hi SPSSlisters,

 

I am using generalized linear models to look at the impact of a reading
program on the attitude of first-grade students toward reading (last
week I posted a question about using ANCOVA to analyze the same data).
The DV is a continuous measure of student attitude towards reading.
There are three IVs: (1) treatment/control (some teachers are treatment,
some control); (2) school; and (3) teacher, all entered as fixed
factors. If I model teacher as nested within school, the results for
treatment/control in the Tests of Main Effects table are blank, with
this note underneath the table: "Unable to compute due to numerical
problems." If I include teacher, school, and treatment/control as fixed
factors with no nesting, I get no results and "Unable to compute..." for
both school and treatment/control.

 

Can anyone tell me why this isn't working? I'm sure I'm missing
something obvious here, but I'm not sure what's going on. As you might
have guessed, I'm new to this type of analysis.

 

Thanks,

____________________________________

Justin Meyer

Researcher

Rowland Reading Foundation

____________________________________

====================To manage your subscription to SPSSX-L, send a message to
[hidden email] (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

[Marta Garcia-Granero]

Justin Meyer wote:

> I am using generalized linear models to look at the impact of a reading
> program on the attitude of first-grade students toward reading (last
> week I posted a question about using ANCOVA to analyze the same data).
> The DV is a continuous measure of student attitude towards reading.
> There are three IVs: (1) treatment/control (some teachers are treatment,
> some control); (2) school; and (3) teacher, all entered as fixed
> factors. If I model teacher as nested within school, the results for
> treatment/control in the Tests of Main Effects table are blank, with
> this note underneath the table: "Unable to compute due to numerical
> problems." If I include teacher, school, and treatment/control as fixed
> factors with no nesting, I get no results and "Unable to compute..." for
> both school and treatment/control.
>

I don't have an exact idea of the layout of your data, but, could it be
that schools are nested withing treatment/control (*), and teachers are
nested within schools? I'd also set schools&teacher as random factors.

(*) For schools to be nested within treatment/control, you should have
that some schools belong to treatment, while others belong to control.
If it is not  the case (same school is present in both groups), then
don't nest it, and try nesting teachers within school*treatment/control.

Describe the situation in more detail, please.

Marta GG

--
For miscellaneous statistical stuff, visit:
http://gjyp.nl/marta/

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (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

[Reutter, Alex]

Can you post the syntax that generates this error to the list?

Alex


-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Justin Meyer
Sent: Monday, November 03, 2008 3:47 PM
To: [hidden email]
Subject: "Unable to compute due to numerical problems" Error with Generalized Linear Models

Hi SPSSlisters,



I am using generalized linear models to look at the impact of a reading
program on the attitude of first-grade students toward reading (last
week I posted a question about using ANCOVA to analyze the same data).
The DV is a continuous measure of student attitude towards reading.
There are three IVs: (1) treatment/control (some teachers are treatment,
some control); (2) school; and (3) teacher, all entered as fixed
factors. If I model teacher as nested within school, the results for
treatment/control in the Tests of Main Effects table are blank, with
this note underneath the table: "Unable to compute due to numerical
problems." If I include teacher, school, and treatment/control as fixed
factors with no nesting, I get no results and "Unable to compute..." for
both school and treatment/control.



Can anyone tell me why this isn't working? I'm sure I'm missing
something obvious here, but I'm not sure what's going on. As you might
have guessed, I'm new to this type of analysis.



Thanks,
____________________________________

Justin Meyer
Researcher
Rowland Reading Foundation

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (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