Negative binomial regression

[Eric Janssen]

Dear all

I wish to perform a negative binomial regression.
Dependent variable is a rate, with variance strongly superior to the mean.
Independent variables are both categorical and quantitative (continuous).
Any suggestion to get a syntax and/or macro?
Many thanks in advance
Eric

[Weeks, Kyle]

Eric, in SPSS 15 negative binomial regression can be conducted in the new Generalized Linear Models procedure for non-negative integer outcomes.  To see what else is in SPSS 15, see http://www.spss.com/spss/whats_new_modules.htm and http://www.spss.com/spss/whats_new_modules.htm.

Regards.

Kyle Weeks, Ph.D.
Director of Product Management, SPSS Product Line
Product Management
SPSS Inc.
[hidden email]
www.spss.com
SPSS Inc. helps organizations turn data into insight through predictive analytics.


-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Eric Janssen
Sent: Monday, October 09, 2006 12:38 PM
To: [hidden email]
Subject: Negative binomial regression

Dear all

I wish to perform a negative binomial regression.
Dependent variable is a rate, with variance strongly superior to the mean.
Independent variables are both categorical and quantitative (continuous).
Any suggestion to get a syntax and/or macro?
Many thanks in advance
Eric

[Eric Janssen]

Dear Kyle
I heard about the new stuff coming with the SPSS 15th version, unfortunately I
was given quite recently the 14th and cannot manage to upgrade once again.
Any other option?
E.

> Eric, in SPSS 15 negative binomial regression can be conducted in the new
> Generalized Linear Models procedure for non-negative integer outcomes.  To
> see what else is in SPSS 15, see
> http://www.spss.com/spss/whats_new_modules.htm and
> http://www.spss.com/spss/whats_new_modules.htm.
>
> Regards.
>
> Kyle Weeks, Ph.D.
> Director of Product Management, SPSS Product Line
> Product Management
> SPSS Inc.
> [hidden email]
> www.spss.com
> SPSS Inc. helps organizations turn data into insight through predictive
> analytics.
>
>
> -----Original Message-----
> From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Eric
> Janssen
> Sent: Monday, October 09, 2006 12:38 PM
> To: [hidden email]
> Subject: Negative binomial regression
>
> Dear all
>
> I wish to perform a negative binomial regression.
> Dependent variable is a rate, with variance strongly superior to the mean.
> Independent variables are both categorical and quantitative (continuous).
> Any suggestion to get a syntax and/or macro?
> Many thanks in advance
> Eric
>

[Weeks, Kyle]

Eric, you should check with you local office about upgrade pricing for and availability of SPSS 15.  Also, you could get a maintenance contract for SPSS which would entitle you to get new versions of SPSS that are released during the duration of your maintenance period.

In releases prior to 15, the following approach may be of use. (From the Tech Support website)

The CNLR procedure fits nonlinear regression models, including ones with user defined loss functions. You may be able to use the following set of command syntax, after editing, to fit negative binomial regression models:

* Change y to the actual dependent variable.
* Add as many parameters b0, b1, ... to the model program as needed.
* Change v1, v2, v3... into the names of the independent variables.
* Modify "compute bx" to be the sum of parameters times independent variables.

* MLE FOR NEGATIVE BINOMIAL (x = threshold p = prob) .
Model program x = 1.5 b0 = 0.0 b1 = 1 b2 = 1 b3 = 1 .
compute bx = b0+b1*v1+b2*v2+b3*v3 .
compute k = exp(bx) .
compute pred_ = x/k .
COMPUTE loss_ = -(lngamma(x+y)-lngamma(x)-lngamma(y+1)+x*bx-(x+y)*ln(1+k)) .
CNLR y
/PRED pred_
/LOSS loss_
/BOUNDS x >= 1 .

Note that analytic standard errors are not available when defining your own loss function in CNLR, so the BOOTSTRAP option would be required to get standard errors for the parameter estimates. Also, models with a prediction form involving an exponential function are often very difficult to fit in general nonlinear regression procedures, so it will likely require a large sample and well behaved data in order for the procedure to be able to reliably estimate the parameters. We suggest trying different sets of starting values for parameters to try to ensure that a true global solution is being found.

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Eric Janssen
Sent: Tuesday, October 10, 2006 2:19 AM
To: [hidden email]
Subject: Re: Negative binomial regression

Dear Kyle
I heard about the new stuff coming with the SPSS 15th version, unfortunately I
was given quite recently the 14th and cannot manage to upgrade once again.
Any other option?
E.

> Eric, in SPSS 15 negative binomial regression can be conducted in the new
> Generalized Linear Models procedure for non-negative integer outcomes.  To
> see what else is in SPSS 15, see
> http://www.spss.com/spss/whats_new_modules.htm and
> http://www.spss.com/spss/whats_new_modules.htm.
>
> Regards.
>
> Kyle Weeks, Ph.D.
> Director of Product Management, SPSS Product Line
> Product Management
> SPSS Inc.
> [hidden email]
> www.spss.com
> SPSS Inc. helps organizations turn data into insight through predictive
> analytics.
>
>
> -----Original Message-----
> From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Eric
> Janssen
> Sent: Monday, October 09, 2006 12:38 PM
> To: [hidden email]
> Subject: Negative binomial regression
>
> Dear all
>
> I wish to perform a negative binomial regression.
> Dependent variable is a rate, with variance strongly superior to the mean.
> Independent variables are both categorical and quantitative (continuous).
> Any suggestion to get a syntax and/or macro?
> Many thanks in advance
> Eric
>