Unexpected Singularities

[Amanda Bertram]

Hello,

I am trying to run a logistic regression and when I do I get the following
errors…



For split file Unit = Hal 5 and Hal 6, there are 419 (39.1%) cells (i.e.,
dependent variable levels by subpopulations) with zero frequencies.
For split file Unit = Hal 5 and Hal 6, unexpected singularities in the
Hessian matrix are encountered. This indicates that either some predictor
variables should be excluded or some categories should be merged.
The NOMREG procedure continues despite the above warning(s). Subsequent
results shown are based on the last iteration. Validity of the model fit is
uncertain.






Now, the “zero frequency” warning I have been ignoring because I needed to
use the zeros to calculate mean, so I am assuming they would be included in
this type of regression.

Where I am really unclear is the “unexpected singularities in the Hessian
matrix are encountered. This indicates that either some predictor variables
should be excluded or some categories should be merged” warning. I am not
sure what it means or what I do to go about fixing it.

I would appreciate any help anybody can give me.

Thanks,

=====================
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

[vlad simion]

Hi Amanda,

Resolution number: 71043 on spss support site:

"Resolution Description:
This warning will be produced when there is a category of the
dependent variable for which one of the predictors is constant. You
can diagnose the problem by examing the regression coefficients
resulting from the last iteration, which are shown in the Parameter
Estimates table. Look for a set of coefficients where the magnitude of
the intercept is very large, and one of the predictor coefficients is
also large, in the opposite direction. If there is only one logit
where this pattern occurs, the category of the dependent variable used
in the numerator of that logit is where the problem is occurring. If
all sets of coefficients have this pattern, the category used as the
reference category, which is used in the denominator in forming all
logits, is the problematic category. You should find that for the
value of the dependent variable identified, the value of the
identified predictor is a constant. This results in the suggestion
that you may want to consider combining categories of the dependent
variable."

Hth,
Vlad

On Thu, Nov 12, 2009 at 4:12 PM, Amanda Bertram <[hidden email]> wrote:

> Hello,
>
> I am trying to run a logistic regression and when I do I get the following
> errors…
>
>
>
> For split file Unit = Hal 5 and Hal 6, there are 419 (39.1%) cells (i.e.,
> dependent variable levels by subpopulations) with zero frequencies.
> For split file Unit = Hal 5 and Hal 6, unexpected singularities in the
> Hessian matrix are encountered. This indicates that either some predictor
> variables should be excluded or some categories should be merged.
> The NOMREG procedure continues despite the above warning(s). Subsequent
> results shown are based on the last iteration. Validity of the model fit is
> uncertain.
>
>
>
>
>
>
> Now, the “zero frequency” warning I have been ignoring because I needed to
> use the zeros to calculate mean, so I am assuming they would be included in
> this type of regression.
>
> Where I am really unclear is the “unexpected singularities in the Hessian
> matrix are encountered. This indicates that either some predictor variables
> should be excluded or some categories should be merged” warning. I am not
> sure what it means or what I do to go about fixing it.
>
> I would appreciate any help anybody can give me.
>
> Thanks,
>
> =====================
> 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
>

=====================
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