Linear mixed model: warning message

[Charlotte-9]

Dear all,

I have been using the linear mixed model facility in SPSS to fit a model
with both random and fixed effects.  However, I keep getting the following
warning message and wondered if anyone could explain the implications of
this to me?  I'm wondering if it means that my random term is basically
redundant?

"The final Hessian matrix is not positive definite although convergence
critera are satisfied.  The MIXED procedure continues despite this
warning.  Validity of subsequent results cannot be ascertained".

Thanks for any help,

Charlotte

[David Futrell]

Hi Everyone:

Is there any way to get SPSS to read and write files to a shared space accessible over a company intranet? Specifically, I need to read (import) data from an Excel file which is maintained by a different group, do some manipulation to the values, then write a different dataset back to the same space.

The files are accessed via URL, so what I need is the equivalent of a "Get File" or "Get Translate" to access a file like this:

http://companynet.global.company.com/sites/DeptName/TestData.xls


Then a way to export it to the same place, like this:

http://companynet.global.company.com/sites/DeptName/ReportFile.xls

Is this even possible?

Thanks,
David Futrell
Senior Workforce Research Consultant
Eli Lilly and Company


---------------------------------
Moody friends. Drama queens. Your life? Nope! - their life, your story.
 Play Sims Stories at Yahoo! Games.

[peter link]

Hello -

Generally this warning message has to do with encountering boundary
constraints.  Check to see if any of your variance components are close to
the boundary (ex. variance close to zero for a random slope or intercept).
Typical recommendation is to simplify the stochastic portion of the model.
One way to do this is the fix the guilty parameter, if known (ex. for
variance component that is close to zero, fix that parameter to zero).  If
there isn't one parameter that is obvious, another way to go about this is
to compare a set of models, removing a random effect in some systematic way
and find the best model, that does converge.

Something else to check would be to increase the number of iterations.  In
SPSS this is done by clicking on the 'estimation' button in the MIXED
procedure dialog.  Try increasing the number of iterations and/or the max.
number of step-halvings.

Good luck,

Peter Link
VA San Diego Healthcare System

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]]On Behalf Of
Charlotte
Sent: Thursday, September 27, 2007 6:01 AM
To: [hidden email]
Subject: Linear mixed model: warning message


Dear all,

I have been using the linear mixed model facility in SPSS to fit a model
with both random and fixed effects.  However, I keep getting the following
warning message and wondered if anyone could explain the implications of
this to me?  I'm wondering if it means that my random term is basically
redundant?

"The final Hessian matrix is not positive definite although convergence
critera are satisfied.  The MIXED procedure continues despite this
warning.  Validity of subsequent results cannot be ascertained".

Thanks for any help,

Charlotte

[Peck, Jon]

If you can use programmability, the key piece of what you need already exists.  The spssaux module, downloadable from SPSS Developer Central (www.spss.com/devcentral) has a function called
OpenDataFileFromUrl.

It retrieves the file to a local copy and opens it with GET FILE.

Currently it works only for SAV files, but I can generalize that easily enough.

Writing a file back would be a little simpler, but it wouldn't be hard to come up with a Put method.

HTH.
Jon Peck

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of David Futrell
Sent: Thursday, September 27, 2007 10:20 AM
To: [hidden email]
Subject: [SPSSX-L] Reading/Writing to Intranet Files

Hi Everyone:

Is there any way to get SPSS to read and write files to a shared space accessible over a company intranet? Specifically, I need to read (import) data from an Excel file which is maintained by a different group, do some manipulation to the values, then write a different dataset back to the same space.

The files are accessed via URL, so what I need is the equivalent of a "Get File" or "Get Translate" to access a file like this:

http://companynet.global.company.com/sites/DeptName/TestData.xls


Then a way to export it to the same place, like this:

http://companynet.global.company.com/sites/DeptName/ReportFile.xls

Is this even possible?

Thanks,
David Futrell
Senior Workforce Research Consultant
Eli Lilly and Company


---------------------------------
Moody friends. Drama queens. Your life? Nope! - their life, your story.
 Play Sims Stories at Yahoo! Games.

[Marks, Jim]

Yes, I believe you can. Try using the full path within ' ' in the FILE
subcommand.

I have used these forms:

SPSS files:
GET FILE = '\\networkname\path1\path2\filename.sav'.

SAVE FILE = '\\networkname\path1\path2\filename.sav'  /COMPRESSED.

EXCEL files:
GET DATA /TYPE=XLS
 /FILE= '\\networkname\path1\path2\filename.xls'
   /SHEET=name 'Sheet1'
   /CELLRANGE=full
   /READNAMES=on
   /ASSUMEDSTRWIDTH=321.

SAVE TRANSLATE OUTFILE =
  '\\NETWORKNAME\PAATH1\PATH2\FILENAME.XLS'
  /TYPE=XLS /VERSION=8 /MAP /REPLACE /FIELDNAMES
  /CELLS=VALUES .

Syntax files:

INSERT FILE =  '\\network\path\name.sps'.

I havent tried the http:// form since my intranet can be reached using
\\networkserver...

--jim

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
David Futrell
Sent: Thursday, September 27, 2007 10:20 AM
To: [hidden email]
Subject: Reading/Writing to Intranet Files

Hi Everyone:

Is there any way to get SPSS to read and write files to a shared space
accessible over a company intranet? Specifically, I need to read
(import) data from an Excel file which is maintained by a different
group, do some manipulation to the values, then write a different
dataset back to the same space.

The files are accessed via URL, so what I need is the equivalent of a
"Get File" or "Get Translate" to access a file like this:

http://companynet.global.company.com/sites/DeptName/TestData.xls


Then a way to export it to the same place, like this:

http://companynet.global.company.com/sites/DeptName/ReportFile.xls

Is this even possible?

Thanks,
David Futrell
Senior Workforce Research Consultant
Eli Lilly and Company


---------------------------------
Moody friends. Drama queens. Your life? Nope! - their life, your story.
 Play Sims Stories at Yahoo! Games.

[Peck, Jon]

You can use a network path as you suggest below, but you cannot use an http url in standard syntax.  That's why we provide the openDataFileFromUrl function.

-Jon Peck

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Marks, Jim
Sent: Thursday, September 27, 2007 11:31 AM
To: [hidden email]
Subject: Re: [SPSSX-L] Reading/Writing to Intranet Files

Yes, I believe you can. Try using the full path within ' ' in the FILE
subcommand.

I have used these forms:

SPSS files:
GET FILE = '\\networkname\path1\path2\filename.sav'.

SAVE FILE = '\\networkname\path1\path2\filename.sav'  /COMPRESSED.

EXCEL files:
GET DATA /TYPE=XLS
 /FILE= '\\networkname\path1\path2\filename.xls'
   /SHEET=name 'Sheet1'
   /CELLRANGE=full
   /READNAMES=on
   /ASSUMEDSTRWIDTH=321.

SAVE TRANSLATE OUTFILE =
  '\\NETWORKNAME\PAATH1\PATH2\FILENAME.XLS'
  /TYPE=XLS /VERSION=8 /MAP /REPLACE /FIELDNAMES
  /CELLS=VALUES .

Syntax files:

INSERT FILE =  '\\network\path\name.sps'.

I havent tried the http:// form since my intranet can be reached using
\\networkserver...

--jim

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
David Futrell
Sent: Thursday, September 27, 2007 10:20 AM
To: [hidden email]
Subject: Reading/Writing to Intranet Files

Hi Everyone:

Is there any way to get SPSS to read and write files to a shared space
accessible over a company intranet? Specifically, I need to read
(import) data from an Excel file which is maintained by a different
group, do some manipulation to the values, then write a different
dataset back to the same space.

The files are accessed via URL, so what I need is the equivalent of a
"Get File" or "Get Translate" to access a file like this:

http://companynet.global.company.com/sites/DeptName/TestData.xls


Then a way to export it to the same place, like this:

http://companynet.global.company.com/sites/DeptName/ReportFile.xls

Is this even possible?

Thanks,
David Futrell
Senior Workforce Research Consultant
Eli Lilly and Company


---------------------------------
Moody friends. Drama queens. Your life? Nope! - their life, your story.
 Play Sims Stories at Yahoo! Games.

[Swank, Paul R]

I would agree that the most likely problem here is that your model (the
random portion) is over parameterized and you need to simplify it.
Another possibility however is that you have reached a saddle point in
the maximizing process, that is a local minimum/maximum rather than the
actual one. You can check this be using different starting values and
see if you arrive at the same place.

Paul R. Swank, Ph.D. Professor
Director of Reseach
Children's Learning Institute
University of Texas Health Science Center-Houston

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Charlotte
Sent: Thursday, September 27, 2007 8:01 AM
To: [hidden email]
Subject: Linear mixed model: warning message

Dear all,

I have been using the linear mixed model facility in SPSS to fit a model
with both random and fixed effects.  However, I keep getting the
following warning message and wondered if anyone could explain the
implications of this to me?  I'm wondering if it means that my random
term is basically redundant?

"The final Hessian matrix is not positive definite although convergence
critera are satisfied.  The MIXED procedure continues despite this
warning.  Validity of subsequent results cannot be ascertained".

Thanks for any help,

Charlotte

[Ornelas, Fermin-2]

When conducting non linear estimation it usually helps to check the
values of the gradient coefficients to ensure that the estimation is
converging properly. If the estimation has problems converging one can
re scale the variable affecting a specific gradient coefficient.

Fermin Ornelas, Ph.D.
Management Analyst III, AZ DES
1789 W. Jefferson Street
Phoenix, AZ 85032
Tel: [phone]
E-mail: [hidden email]


-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
peter link
Sent: Thursday, September 27, 2007 9:09 AM
To: [hidden email]
Subject: Re: Linear mixed model: warning message

Hello -

Generally this warning message has to do with encountering boundary
constraints.  Check to see if any of your variance components are close
to
the boundary (ex. variance close to zero for a random slope or
intercept).
Typical recommendation is to simplify the stochastic portion of the
model.
One way to do this is the fix the guilty parameter, if known (ex. for
variance component that is close to zero, fix that parameter to zero).
If
there isn't one parameter that is obvious, another way to go about this
is
to compare a set of models, removing a random effect in some systematic
way
and find the best model, that does converge.

Something else to check would be to increase the number of iterations.
In
SPSS this is done by clicking on the 'estimation' button in the MIXED
procedure dialog.  Try increasing the number of iterations and/or the
max.
number of step-halvings.

Good luck,

Peter Link
VA San Diego Healthcare System

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]]On Behalf Of
Charlotte
Sent: Thursday, September 27, 2007 6:01 AM
To: [hidden email]
Subject: Linear mixed model: warning message


Dear all,

I have been using the linear mixed model facility in SPSS to fit a model
with both random and fixed effects.  However, I keep getting the
following
warning message and wondered if anyone could explain the implications of
this to me?  I'm wondering if it means that my random term is basically
redundant?

"The final Hessian matrix is not positive definite although convergence
critera are satisfied.  The MIXED procedure continues despite this
warning.  Validity of subsequent results cannot be ascertained".

Thanks for any help,

Charlotte

NOTICE: This e-mail (and any attachments) may contain PRIVILEGED OR
CONFIDENTIAL information and is intended only for the use of the
specific individual(s) to whom it is addressed.  It may contain
information that is privileged and confidential under state and federal
law.  This information may be used or disclosed only in accordance with
law, and you may be subject to penalties under law for improper use or
further disclosure of the information in this e-mail and its
attachments. If you have received this e-mail in error, please
immediately notify the person named above by reply e-mail, and then
delete the original e-mail.  Thank you.

[David Futrell]

This worked perfectly on the first try.

Thanks!!




"Marks, Jim" <[hidden email]> wrote: Yes, I believe you can. Try using the full path within ' ' in the FILE
subcommand.

I have used these forms:

SPSS files:
GET FILE = '\\networkname\path1\path2\filename.sav'.

SAVE FILE = '\\networkname\path1\path2\filename.sav'  /COMPRESSED.

EXCEL files:
GET DATA /TYPE=XLS
 /FILE= '\\networkname\path1\path2\filename.xls'
   /SHEET=name 'Sheet1'
   /CELLRANGE=full
   /READNAMES=on
   /ASSUMEDSTRWIDTH=321.

SAVE TRANSLATE OUTFILE =
  '\\NETWORKNAME\PAATH1\PATH2\FILENAME.XLS'
  /TYPE=XLS /VERSION=8 /MAP /REPLACE /FIELDNAMES
  /CELLS=VALUES .

Syntax files:

INSERT FILE =  '\\network\path\name.sps'.

I havent tried the http:// form since my intranet can be reached using
\\networkserver...

--jim

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
David Futrell
Sent: Thursday, September 27, 2007 10:20 AM
To: [hidden email]
Subject: Reading/Writing to Intranet Files

Hi Everyone:

Is there any way to get SPSS to read and write files to a shared space
accessible over a company intranet? Specifically, I need to read
(import) data from an Excel file which is maintained by a different
group, do some manipulation to the values, then write a different
dataset back to the same space.

The files are accessed via URL, so what I need is the equivalent of a
"Get File" or "Get Translate" to access a file like this:

http://companynet.global.company.com/sites/DeptName/TestData.xls


Then a way to export it to the same place, like this:

http://companynet.global.company.com/sites/DeptName/ReportFile.xls

Is this even possible?

Thanks,
David Futrell
Senior Workforce Research Consultant
Eli Lilly and Company


---------------------------------
Moody friends. Drama queens. Your life? Nope! - their life, your story.
 Play Sims Stories at Yahoo! Games.



---------------------------------
Moody friends. Drama queens. Your life? Nope! - their life, your story.
 Play Sims Stories at Yahoo! Games.

[Charlotte-9]

Dear all,

Thanks for the advice.  I will have a play around with the model, taking
into account your suggestions, and see how I get on.

All the best,

Charlotte

[Peck, Jon]

The latest version of the spssaux module on SPSS Developer Central (www.spss.com/devcentral) now supports reading data files from SPSS, Excel, SAS, and Stata locally or from a URL.
The openDataFile function reads a file from a file system, and
openDataFileFromUrl
does the same for internet addresses.

These functions accept arguments for specifying the various parameters associated with the file types but use the same defaults as the corresponding SPSS syntax.

The spssaux module and the spssdata module updates also have some additional support for the Unicode mode available in SPSS 16.

Regards,
Jon Peck

-----Original Message-----
From: Peck, Jon
Sent: Thursday, September 27, 2007 11:18 AM
To: 'David Futrell'; [hidden email]
Subject: RE: [SPSSX-L] Reading/Writing to Intranet Files

If you can use programmability, the key piece of what you need already exists.  The spssaux module, downloadable from SPSS Developer Central (www.spss.com/devcentral) has a function called
OpenDataFileFromUrl.

It retrieves the file to a local copy and opens it with GET FILE.

Currently it works only for SAV files, but I can generalize that easily enough.

Writing a file back would be a little simpler, but it wouldn't be hard to come up with a Put method.

HTH.
Jon Peck

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of David Futrell
Sent: Thursday, September 27, 2007 10:20 AM
To: [hidden email]
Subject: [SPSSX-L] Reading/Writing to Intranet Files

Hi Everyone:

Is there any way to get SPSS to read and write files to a shared space accessible over a company intranet? Specifically, I need to read (import) data from an Excel file which is maintained by a different group, do some manipulation to the values, then write a different dataset back to the same space.

The files are accessed via URL, so what I need is the equivalent of a "Get File" or "Get Translate" to access a file like this:

http://companynet.global.company.com/sites/DeptName/TestData.xls


Then a way to export it to the same place, like this:

http://companynet.global.company.com/sites/DeptName/ReportFile.xls

Is this even possible?

Thanks,
David Futrell
Senior Workforce Research Consultant
Eli Lilly and Company


---------------------------------
Moody friends. Drama queens. Your life? Nope! - their life, your story.
 Play Sims Stories at Yahoo! Games.